PhDthesis Nayak PDF

Soft X-ray Multilayers as Polarizing Elements:
Fabrication, and Studies of Surfaces and

Interfaces

A Thesis
Submitted for the Degree of
Doctor of Philosophy
In the Faculty of Science

By
Maheswar Nayak

Material Research Centre
Indian Institute of Science
Bangalore-560 012, India

August 2007

Declaration

I hereby declare that the work reported in this thesis is entirely original and has been carried
out by me in Synchrotron Utilization & Materials Research Division, Raja Ramanna Centre
For Advanced Technology, Indore-452013, India and Material Research Centre, Indian
Institute of Science, Bangalore-560012, India under the joint supervision of Dr. R. V.
Nandedkar, Dr. G. S. Lodha and Prof. S. A. Shivashankar. I further declare that to the best of
my knowledge this work has not formed the basis for award of any degree in any University
or Institution.

August, 2007 M. Nayak

To my beloved late parents

Acknowledgements
i

Acknowledgements

It is a great honor and privilege to express my deep gratitude to my supervisors, Dr.
R. V. Nandedkar and Prof. S. A. Shivashankar, for their constant guidance and words of
inspirations during my project work, which helped me in reaching the goal. They infused
enthusiasm in me at every step during my project work. Dr. R. V. Nandedkar, the as Head of
our Division, he has gone out of his way to help me and solve my problems. He has always
been a source of inspiration for me. His patience and cool temperament in handling tough
situations eased the tension in the work. I have enjoyed the academic freedom he has
afforded to me. I thank to Prof. S. A. Shivashankar for his continuous support and constant
encouragement.

I take this opportunity to express my deep sense of gratitude to Dr. G. S. Lodha, Co-
supervisor of my project work, for giving me every kind of supports and freedom I needed to
carry out my research work. He helped me develop an experimental approach towards
science and to interpret experimental results in an analytical manner. Beginning my research
career under him has been most beneficial to me because he has the great ability to sort out
experimental difficulties. He has high professional temperament, and a transparent yet a
caring attitude, which certainly benefits a person working with him. His novel and exciting
experimental ideas helped me in reaching my targets.

I express my gratefulness to the late Prof. S. M. Chaudari for his support in x-ray
photoelectron spectroscopy measurements and for fruitful scientific collaboration.

I am grateful to Dr. K. J. S. Sawhney for his valuable scientific suggestions and
constant encouragement in the early course of my research work.

I thank to Prof. Ajay Gupta sincerely for providing access to the diffractometer in
IUC, Indore, for the reflectivity measurements of test samples I needed to make during
commissioning of the e-beam system. I also thank to Amit Sarayia for his co-operation
during these measurements.
I am thankful to Mr. T. P. Tenka for his co-operation in carrying out some of experiments.
Acknowledgements
ii

I am especially thankful to Dr. M. H. Modi, Mr. S. Rai, and Dr. M. K. Tiwari for the
friendly environment and for various stimulating scientific discussions, and the cooperation
they provided in carrying out some of the experiments.

I am grateful to Dr. Ashwani Kumar for help numerical analysis and for fruitful
discussions on electrical measurements. I am thankful to Dr. A. K. Sinha for co-operation in
carrying out soft x-ray reflectivity measurements, for scientific discussions and constant
encouragement.

I am thankful to Mr. B. Gowrisankar for his support in the mechanical design of the
electron beam evaporation system and the load lock system. It was a pleasant experience. I
am thank full to Mr. S. R. Kane, Mr. C. Garg, Mr. V. Mondloi and Mr. S. P. L. Shrivastav for
sorting out many problems related to electronic instruments; without this help, it would have
been difficult to carry out some of the experiments in time.

I am thankful to members of Synchrotron Utilization and Materials Research Division
for a creating friendly atmosphere, which helped me a lot. I have joyous memories of the
time I spent with them. I thank Dr. A. K. Shrivastava, Ms. Pragya Tripathi, Mr. Adu Verma,
Mr. Vinay Raghuvanshi, Dr. Archana Jayaswal, Dr. P. Gupta, Mr. Vishal Dhamgaye, Mr.
Arijeet Das and Mr. H. Shrivastava for lending me all kinds of support. I am also thankful to
Mr. S. Naik and Mr. G. M. Vallerao for different types of help.

I am also thankful to the members of the Indus-I operation, without whom my Indus-I
measurements would not have been possible. I would like to thank Dr. K. J. S. Sokhey for his
keen interest in my work. I am thankful to Mr. P Bhatt for his co-operation in carrying out
XPS measurements.

Acknowledgements
iii
I especially thankful to the technical staff of our Division for the helping hand they
extended in the execution of different aspects of the experimental work. I cannot forget to
thank Mr. Chowhan for rendering untiring assistance in the assembly and testing of the
deposition system, and in sample preparation, even at odd hours. Thanks to Mr. R. Dhavan,
Mr. R. K. Gupta, and Mr. M. Simgh for their support during measurements on Indus-I.
Thanks to Mr. Naval Kishore, Mr. Ajit Kumar, Mr. Mahendra Babu, Mr. Arjun Kumar, Mr.
Suraj Das, Mr. P. K. Shrivastav, Mr. Raju, and Mr. Sanjay Likhar for providing different
kinds of help.

I am thankful to all of my Oriya friends in RRCAT for their joyful company and for
the many light moments. I recall the joyful time spent with Jagdish, Rama, and Khatei during
my course work at IISc, Bangalore. I am thankful to the other students of Prof. Shivashankar,
namely Ashish, Pritesh, Pranav, Girish, Sanjaya, Anirudha and Mahua for different kinds of
help during my visits to IISc.

Last but not least, I must add that it would not have been possible for me to achieve
this milestone in my academic career without the encouragement, and moral and mental
support from all of my family members. My parents, who always prayed for my success in
all walks of my life, are with me no more. I pray to God to grant peace to their soul. I cannot
imagine taking steps in achieving different milestones in my life without active support from
my wife Mili. My cute little daughter Shradha keeps me always happy, which is a great
support in my life.
Preface
iv

Preface

The exploitation of the soft x-ray/extreme ultra-violet (EUV) region of the
electromagnetic spectrum is possible mainly due to the development of multilayer (ML)
mirrors. This region of the electromagnetic spectrum offers great opportunities in both
science and technology. The shorter wavelength allows one to see smaller features in
microscopy and write finer features in lithography. High reflectivity with moderate spectral
bandwidth at normal/near-normal incidence can be achieved in soft x-ray/ EUV spectral
range using these ML mirrors, where natural crystals with the required large periodicity are
not available. These MLs are generally artificial Braggs reflectors, which consist of
alternative high and low density materials with periodicity in the nanometer range. The main
advantages of ML optics stem from the tunability of layer thickness, composition, lateral
gradient, and the gradient along the normal to the substrate; these can be tailored according to
the desired wavelength regime. They have the great advantage of being adaptable to figured
surfaces, enabling their use as reflective optics in these spectral regions, for focusing and
imaging applications. Broadband reflectivity and wavelength tunability are also possible by
using MLs with normal and lateral gradient, respectively. However, fabrication of these ML
mirrors requires the capability to deposit uniform, ultra-thin (a few angstroms-thick) films of
different materials with thickness control on the atomic scale. Thus, one requires a proper
understanding of substrate surfaces, individual layers, chemical reactivity at interfaces and,
finally, of the ML structures required for particular applications. The performance of these
MLs is limited by (the lack of) contrast in optical constants of the two materials, interfacial
roughness, the chemical reactivity of two materials and, finally, errors in the thickness of
individual layers.
Soft x-ray/extreme ultra-violet ML mirrors have found a wide range of applications in
synchrotron radiation beam lines, materials science, astronomy, x-ray microscopy, x-ray
laser, x-ray lithography, polarizers, and plasma diagnostics. The Indus1 synchrotron
radiation (SR) source is an operational 450 MeV machine, which produces radiation up to
soft x-rays. Indus-2 is a 2.5 GeV machine, which has been commissioned recently to produce
hard x-rays (E > 25 keV). The combination of Indus-1 and Indus-2 will cover a broad energy
Preface
v
spectrum from IR to hard x-rays. Therefore, there is a significant need and opportunity to
study MLs of different pairs of materials, with different parameters such as periodicity and
optimum thickness of individual layers. The goal of the present thesis is to fabricate MLs for
soft x-ray optics and to study their physics for application as polarizers in the wavelength
range from 67 to 160 on the Indus-1 synchrotron source. To accomplish this task, a
UHV electron beam evaporation system has been developed indigenously for the fabrication
of MLs. Three different ML systems viz., Mo/Si, Fe/B
4
C and Mo/Y have been fabricated,
and their surfaces and interfaces were investigated thoroughly for the polarizer application.
X-ray reflectivity (XRR) has been used extensively in the investigations of these MLs. This
is because XRR is a highly sensitive non-destructive technique for the characterization of
buried interfaces, and gives microscopic information (at atomic resolution) over a
macroscopic length scale (a few microns). Numerical analysis of XRR data has been carried
out using computer programs. Depth-graded x-ray photoelectron spectroscopy (XPS) has
been used for compositional analysis at interfaces for some of the ML structures, as a
technique complementary to XRR. The performance of some of these MLs has been tested in
the soft x-ray region, using the Indus-1 synchrotron radiation (SR) source. Prior to studying
the MLs, a detailed study of the surfaces and interfaces of thin films, bi-layers, and tri-layers
was carried out using XRR and the glancing incidence fluorescence technique. The
discontinuous-to-continuous transition and the mode of film growth, which are vital to the
optimization of layer thickness (basically for the high-atomic number or high-Z layer) in the
ML structures, were also investigated using in situ sheet resistance measurement method.
Indus-1 is a soft x-ray SR source that covers atomic absorption edges of many low-Z
materials. The present work demonstrates the possibilities of characterizing low-Z thin films
and multilayers using soft x-ray resonant reflectivity. In one case, we have shown for first
time that soft x-ray resonant reflectivity can be employed as a non-destructive technique for
the determination of interlayer composition. In a second study using the Indus-1 SR source,
we have shown, by observing the effect of the anomalous optical constant on reflectivity
pattern when photon energy is tuned across the atomic absorption edge of the constituent
low-Z element, that soft x-ray resonant reflectivity is an element-specific technique.
This thesis is organized into 7 chapters. A brief summary of individual chapters is
presented below.

Preface
vi
Chapter 1 gives a brief general introduction to x-ray ML optics. This is followed by a
discussion of the importance of the soft x-ray region of electromagnetic radiation. The optical
properties of x-rays are reviewed and optical constants are calculated for some of the
important materials used for x-ray MLs. The refractive index in the x-ray region being less
than unity (except absorption edges), the consequent limitation of conventional transmission
lenses is discussed. The limitation of glancing angle incidence optics is presented, motivating
the need for ML optics, which is discussed along with a theoretically calculated reflectivity
profile. The procedure for materials for the MLs for application in different spectral regions
is discussed, along with a survey of literature related to the present thesis. The importance of
the quality of surfaces and interfaces on the performance of ML structures has been shown
through simulations. The applications of soft x-ray MLs are discussed with emphasis on
polarization. This is followed by a review of different modes of growth of thin films. Finally,
the scope of the present work is highlighted.

Chapter 2 provides brief descriptions of the experimental techniques used in the present
investigations and of the numerical methods employed for quantitative data analysis. The
XRR technique is discussed elaborately because it has been used extensively. Detailed
calculations of x-ray reflectivity from single surfaces, thin films and bi-layers are presented,
along with simulated values. The effect of critical angle and Brewsters angle is also
discussed. Data analysis methods for computing x-ray reflectivity from multilayer structures,
based on dynamical and kinematical models, have been discussed. The effect of roughness
on XRR has been discussed based on the recursion formalism of dynamical theory.
Simulations of XRR and experimental XRR data fitting are carried out using computer
programs. The XRR experimental set up is also outlined. A theoretical background is given
for the electrical measurements on thin films. This is followed by a brief overview of x-ray
photoelectron spectroscopy (XPS) and interpretation of spectra. Finally, the glancing
incidence x-ray fluorescence (GIXRF) technique is outlined.

Chapter 3 describes in detail the ultra-high vacuum electron beam evaporation system
developed in house especially for the fabrication of thin films and x-ray multilayer optics. At
the outset, a brief overview of different deposition techniques commonly used for the
fabrication of x-ray optical elements is presented. Design, fabrication, and assembly of
Preface
vii
different accessories are discussed. The control of thickness and uniformity of the films
deposited has been checked through the experiments, whose results are provided. The results
obtained for ML test structures are presented to show the capability of system in carrying out
fabrication of high quality x-ray ML structures. Finally, the versatility of evaporation system
incorporating in situ characterization facilities such as -situ electrical measurements for
different substrate temperatures is illustrated.

Chapter 4 presents a study of the growth of ultra-thin Mo films at different substrate
temperatures using in situ sheet resistance measurements. First, a theoretical background is
given on the different stages of island growth and on factors affecting thin film growth,
followed by a discussion of the possible electrical conduction phenomena in continuous and
discontinuous metal films. The nature of thin film growth and a detailed microscopic picture
at different growth stages are derived from a modeling of sheet resistance data obtained in
situ. The various conduction mechanisms have been identified in different stages of growth.
In the island growth stage, the isotropic and anisotropic growth of Mo islands is identified
from the model. In the insulator-metal transition region, experimentally determined values of
critical exponent of conductivity agrees well with theoretically predicted values for a two-
dimensional (2D) percolating system, revealing that Mo films on float glass substrate is
predominantly a 2D structure. The minimum thickness for which Mo films becomes
continuous is obtained as 1.8 nm and 2.2 nm for Mo deposited at substrate temperatures 300
K and 100 K, respectively. An amorphous-to- crystalline transition is also observed, and
discussed.

Chapter 5 covers the detailed study of the surfaces and interfaces studies in three different
ML structures viz., Mo/Si, Fe/B
4
C and Mo/Y, meant for the polarizer application in the
wavelength range of 67 to 160 . Multilayers with varying periodicity, varying number of
layer pairs, and different ratios of high-Z layer thickness to the period, were fabricated using
the electron beam system. Initially, a brief overview of the design aspects of ML structures is
given, along with the theoretically calculated reflectivity at Brewsters angle from the best
material combinations. In Mo/Si MLs, the interlayer formed at the interfaces due to inter-
diffusion of the two elements is asymmetric in thickness, i.e., Mo-on-Si interlayer is thicker
than the Si-on-Mo interlayer. To take account of these interlayers in XRR data fitting, a four-
Preface
viii
layer model is considered. The effect of interlayers on reflectivity pattern was studied using
simulations, and differences with respect to roughness are also discussed. The mechanism of
formation of asymmetric interlayers is also discussed. The interlayer composition has
determined using depth-graded XPS. The results reveal the formation of the MoSi
2

composition at both the interfaces. The experimental results agree well with theoretical
calculations based on solid-state amorphization reaction, which is a result of large heat of
mixing. The effective heat of formation model reveals the formation of MoSi
2
as the first
phase. The soft x-ray reflectivity performance of the Mo/Si ML structure at Brewsters angle
is tested using Indus-1 synchrotron radiation (SR).

Using XRR and GIXFR, a study of the surfaces and interfaces of bilayers of B
4
C-on-Fe and
Fe-on- B
4
C, and tri-layers of Fe-B
4
C-Fe was carried out, with a systematic variation of Fe
and B
4
C layer thicknesses. A sharp interface was observed in Fe-on-B
4
C, whereas a low
density (w.r.t. Fe) interlayer is observed at the B
4
C-on-Fe interface. The interlayer properties
fluctuates w.r.t. the bottom Fe layer thickness and is independent of the top B
4
C layer
thickness. The nature of fluctuations has been discussed in detail. A study of the surfaces and
interfaces of Fe/B
4
C MLs is described. Finally, a study of the surfaces and interfaces of bi-
layers, tri-layers, and MLs of the Mo/Y system is discussed in detail.

Chapter 6 describes the application of soft x-ray resonant reflectivity for the characterization
of low-Z thin films and interfaces in multilayer structures. Initially, a discussion of the
energy dependence of atomic scattering factors and hence of optical constants is provided
with simulations, with emphasis on the atomic absorption edge. Then, a brief overview of
synchrotron radiation, with particular emphasis on the parameters of the Indus-1 synchrotron
source is given. The possibilities of determining the composition of the buried interlayer with
sub-nanometer scale sensitivity using soft x-ray resonant reflectivity are discussed. The
methodology has been applied to study the Mo/Si interface both by simulations and by
experiments on the Indus-1 SR, by tuning the photon energy to the Si L-absorption edge.
Finally, direct evidence of elemental specificity of soft x-ray resonant reflectivity through the
observation of the effect of anomalous optical constants on the reflectivity pattern is
discussed. We demonstrate the method through simulations and experiments on the B
4
C
Preface
ix
material in B
4
C thin films and Fe/ B
4
C bi-layers, using Indus-1 SR tuned to the boron K-
edge.

Chapter 7 summarizes the main findings of the present work, and provides an outlook for
further investigations in the field.

Contents
x
CONTENTS

Description Page No.

Chapter-1: Introduction 1

1.1 Introduction 2
1.2 Soft X-rays and their Importance 3
1.3 Optical Properties of X-rays 4
1.4 X-ray Optics 6
1.4.1 Glancing Incidence Optics 7
1.4.2 Multilayer Optics 8
1.4.2.1 Advantages of Multilayers 9
1.4.2.2 Materials Selection 9
1.4.2.3 Substrates 12
1.5 Factors affecting the Performance of Multilayers 12
1.6 Thin Film Growth 15
1.7 Applications of Multilayer Structures 16
1.8 Aim of the Present Work 17
1.9 References 21

Chapter-2: Characterization Techniques and Data Analysis 25

2.1 Introduction 26
2.2 X-ray Reflectivity 26
2.2.1 Scattering Geometry 26
2.2.2 Reflection and Refraction at an Interface 29
2.2.2.1 Total External Reflection 30
2.2.3 Reflection Coefficients at an Interface 31
2.2.3.1 Brewsters Angle for X-rays 34
2.2.4 Reflectivity from a Single Film 35
2.2.5 Reflectivity from a Bi-layer Structure 37
2.2.6 Reflectivity from a Multilayer Structure 38
2.2.6.1 Kinematical Theory 39
Contents
xi
2.2.6.2 Dynamical Theory 41
2.2.7 Imperfect Boundaries 43
2.2.7.1 Reduction of Reflectivity 43
2.2.7.2 Different Types of Roughness 44
2.2.7.3 Effect of Roughness on Reflectivity 45
2.2.8 Method of Data Analysis 47
2.2.8.1 Instrumental Resolution 49
2.2.9 Experimental Set-up 50
2.3 Electrical Conduction in Thin Films 50
2.3.1 General Considerations 50
2.3.2 Measurements of Thin Film Resistivity 51
2.3.2.1 Four Probe-Point Method 52
2.3.2.2 The van der Pauw Method 52
2.4 X-ray Photoelectron Spectroscopy 53
2.4.1 X-ray Photoelectron Spectral Interpretation 54
2.5 Glancing Incidence X-ray Fluorescence 56
2.6 References 57

Chapter-3: Design and Development of an UHV Electron Beam
Evaporation System 59

3.1 Introduction 60
3.2 Deposition Techniques 61
3.2.1 Sputter Deposition 61
3.2.1.1 DC Sputtering 61
3.2.1.2 RF Sputtering 62
3.2.1.3 Magnetron Sputtering 62
3.2.2 Ion Beam Sputtering 64
3.2.3 Electron Beam Evaporation 65
3.3 Developed of UHV Electron Beam Evaporation System 67
3.3.1 Design of UHV Chamber 67
3.3.2 Pumping System and Gauges 68
3.3.3 Electron Guns and Shutters 69
3.3.4 Substrate Holder and Masking Arrangements 70
Contents
xii
3.3.5 Load Lock System For Sample Transfer 70
3.3.6 Automation of Thickness Control 70
3.4 System Performance and Results 71
3.4.1 Vacuum Testing 71
3.4.2 Deposition of Thin Films of Various Materials 72
3.4.3 Deposition of Bi-layers 74
3.4.4 Deposition of Multilayers 75
3.4.5 In situ Resistivity Measurement Facility 76
3.5 Conclusions 77
3.6 References 78

Chapter-4: Study of Thin Film Growth 79

4.1 Introduction 80
4.2 Theoretical Approach 82
4.2.1 Different Stages of Film Growth 82
4.2.2 Factors Affecting the Film Growth Process 85
4.2.3 Electron Transport in Metallic Films 86
4.2.3.1 Conduction in Discontinuous Films 86
4.2.3.1.1 Transport by Thermionic Emission 86
4.2.3.1.2 Transport by Tunneling 87
4.2.3.2 Conduction in Continuous Films 87
4.2.3.2.1 Matthiessens Rule 88
4.2.3.2.2 Electron Scattering From Film Surfaces 89
4.2.3.2.3 Grain-Boundary Scattering 90
4.3 Growth of Molybdenum Thin Films 90
4.3.1 Experimental 91
4.3.2 Results and Discussion 93
4.3.2.1 Nucleation and Growth 94
4.3.2.2 Study of Infinite Clusters 97
4.3.2.3 Amorphous-to-Crystalline Transition 104
4.4 Conclusions 107
4.5 References 108
Contents
xiii

Chapter-5: Study of Soft X-ray Multilayers 111

5.1 Introduction 113
5.2 Design Criteria of ML Structures 114
5.2.1 High-Reflectivity Multilayers 117
5.3 Mo/Si Multilayer Structures 120
5.3.2 Results and Discussion of Hard XRR Measurements 124
5.3.2.1 Modeling and Simulation 125
5.3.2.2 Influence of Interlayer 126
5.3.2.3 Effect of Asymmetry 128
5.3.2.4 Application of Four-layer Model to Mo/Si MLs 129
5.3.3 Results of Depth-graded XPS Measurements 135
5.3.3.1 Discussions 139
5.3.4 ML Performance Assesment using Indus-1 SR 144
5.4 Fe/B
4
C Bi-layers, Tri-layers, and Multilayer Structures 146
5.5 Mo/Y Bi-layers, Tri-layers, and Multilayer Structures 153
5.6 Conclusions 157
5.7 References 160

Contents
xiv
Chapter-6: Soft X-ray Resonant Reflectivity for Low-Z Thin Films and
Multilayers 165

6.1 Introduction 166
6.2 Theoretical Background 167
6.2.1 Atomic Scattering Factor 169
6.3 Synchrotron Radiation 175
6.3.1 Indus-1 Synchrotron Source 176
6.4 Soft X-ray Resonant Reflectivity as a Non-destructive Technique
for Interlayer Composition Analysis 178
6.4.1 Simulations 178
6.5 Elemental Specificity of Soft X-ray Resonant Reflectivity 183
6.6 Conclusions 192
6.7 References 193

Chapter-7: Summary and Conclusions 196

7.1 Summary and Conclusions 197
7.2 Scope for Investigations in the Future 201

List of Publications 203

Curriculum Vitae 206

Chapter 1
1

CHAPTER-1

Introduction

This chapter gives a brief overview of the importance and salient features of soft x-ray
multilayer structures. The problem related to x-ray optical properties have also been
described. A special emphasis has been placed on layer growth and the effect of surfaces and
interfaces on the performance of soft x-ray multilayer optical elements. The selection of
material pairs for obtaining the best performance in x-ray multilayers is presented. A brief
over view of applications of x-ray multilayers with special emphasis on polarizers is
provided. The exiting literature on the soft x-ray multilayer structures related to the present
investigation is surveyed. Finally, the scope of present investigation is outlined.
Chapter 1
2
1.1 Introduction

X-rays were discovered by Rntgen
1
in 1895, and he stated that the refractive index
of all materials for x-rays was very close to that of vacuum or air. So, the angle of refraction
is very small and conventional refractive lenses would have impossibly long focal lengths
(f=R/, for plano-concave lens where, f is focal length, R is radius of curvature and is real
part of refractive index). In addition, due to high absorption, the transmission lens is
practically difficult for x-rays. The discovery of x-ray diffraction from naturally occurs
crystals
2
provides a method to deflect x-rays by large angle. The reconstruction of an object
from its diffraction pattern is a two-step imaging process, where the second step is a
computation. Again, problems are associated with the attempt to use natural crystals to
deflect longer wavelength x-rays at large angles, because of the non-availability of natural
crystals of large periodicity. In order to manipulate x-rays, a new version of x-ray diffraction
has been evolved during last 30 years. Well-defined, artificially made diffraction structures
are being designed, in the absence of naturally occurring crystalline diffraction structures. It
is now possible to fabricate a more stable artificial diffraction structure of required
periodicity due to advancements in thin film deposition techniques. The most important
aspect of these artificial diffraction structures is that these structures can be designed in such
a way that their diffraction pattern represents a real image of the object by performing a
Fourier transform and its inversion in one step. They are most useful in the soft x-ray region,
where computational methods have not been successful. The two main representatives of
these artificial devices are thin film multilayer structures and Fresnel zone plates
3
. This thesis
concentrates on thin film soft x-ray multilayer structures.
This Chapter is organized in the following manner: Initially, a brief overview of the
importance of the soft x-ray spectral region will be given. Next, the optical properties of x-
rays will be discussed, relating to the limitations of single surfaces in giving high reflectivity
and the consequent need for thin film multilayer structures (MLS). The importance of these
structures for soft x-ray optics will be discussed. The choice of the best combination of
materials for achieving high reflectivity over different spectral ranges will be presented. The
role of the surfaces and the interfaces of the MLS in the performance of optical elements will
then be considered, including the growth of such MLS. Thereafter, the diversity of
applications of soft x-ray multilayer optics will be presented briefly with particular emphasis
on polarimeters. Finally, the scope of the present thesis work will be presented.
Chapter 1
3

1.2 Soft X-rays and their importance

The wavelength range from ~2000 to 300 corresponds to the vacuum ultra-
violet (VUV) region, whereas the hard x-ray region extends from a wavelength of ~5 to
0.1 . The spectral range in between VUV to hard x-rays, i.e., the wavelength range from
~300 to 5 is known as the soft x-ray/extreme ultra-violet region. A large number of
atomic resonances lie in this spectral range and, hence, high degree of absorption takes place
in all materials, leading to absorption of radiation over very short distances, typically in
micrometers. While this plentitude of atomic resonances and efficient photo-absorption has
made this spectral region more difficult to access, it also provides a very sensitive tool for
chemical and elemental identification, thus creating much scientific and technological
opportunity for advances, both in science and technology. Since wavelengths are short, high-
resolution optical techniques would be possible, thus permitting direct image formation and
spatially resolved spectroscopes, with spatial resolutions measured in the tens of nanometers
will be achieved. The relative transparency of water and its natural contrast with other
elements add further to these opportunities: for instance, spectroscopy in the life and
environmental science. Exploiting these opportunities requires the development of proper
soft x-ray optical devices. These in turn lead to new scientific understanding, through surface
science, physics and chemistry, providing feedback to the technologies. In the following, we
now consider the different types of soft x-ray sources.
The general method of producing soft x-rays is the bombardment low atomic number
targets
4
, such as Al (=8.34 ), Mg (=9.8 ), C (=44.7 ) and B (=67.6 ) etc., with
high energy electrons, giving rise to intense characteristic line spectra. Soft x-rays are also
produced by laser-produced hot plasma sources, the resulting spectra being both discrete and
continuous in nature
3
. The applications of soft x-rays have increased tremendously over the
last few decades, due to the availability of synchrotron radiation (SR) sources. SR sources
produce very intense, collimated, well-polarized x-rays, continuous over a wide spectral
range
5
. A brief description of SR sources is given in Chapter 6, together with the parameters
of the Indus-1 SR source.

Chapter 1
4
1.3 Optical Properties of X-rays

In order to arrive at the optimal design of optical devices for various applications,
reliable optical data on mirror materials are essential. These optical data are also vitally
important to explaining the experimental results. The refractive index in the x-ray region can
be written as i n + = 1 , where (the dispersion term) is the real part, and (the
absorption term) is the imaginary part, of refractive index. The quantities and , generally
known as optical constants, can be written as

1
2
2
f
N r
a e
(1.1)

2
2
2
f
N r
a e
(1.2)
where is the incident photon wavelength, r
e
is the classical electron radius, N
a
atomic
density, and f
1
, f
2
the real and imaginary parts of the atomic scattering factor, respectively.
The detailed derivation of the expressions above is given in Chapter 6. The optical constants
are real, and they depend both on the material and the wavelength. In the x-ray region,
typically, ~10
-2
to 10
-5
and ~10
-2
to 10
-6
for all materials, except at atomic absorption
edges, where anomalous effects occur. The values of optical constants are smaller in the hard
x-ray region than in the soft x-ray region. Comprehensive tabulated data of these optical
constants are given by Henke et al.
6
for elements with atomic number Z=1 to 92. The optical
constants near the absorption edges are very sensitive to the configuration of an atom in its
environment. Therefore, by tuning the photon energy to be close to the absorption edges of
the constituent elements, it would be possible to distinguish its composition. However, the
fine structure produced by interaction with neighboring atoms is smoothed out in Henkes
tabulation and there are uncertainties in these tabulated values near absorption edges. There
are also materials for which optical constants are not explored.
The optical constants ( and ) of some of the materials frequently used for x-ray
multilayers are shown in Fig. 1.1, for the wavelength region, 0.5 to 400 . For short
photon wavelengths, the real part of the atomic scattering factor f
1
approaches the number of
electrons per atom (except at absorption edges), and remains constant for shorter
wavelengths (cf Chapter 6). As a result, increases as
2
(Eq.
n
(1.1).
Chapter 1
5

But f
2
is proportional to wavelength, and hence, increases as
3
, as shown in Fig. 1.1.
Therefore, as the photon wavelength tends towards lower values, x-ray absorption decreases
faster than the refractive index. Hence, materials become absorption-free dielectrics at shorter
wavelengths (higher energy). In the x-ray region, the optical properties are described by the
interaction of photons with core electrons. Near the absorption edges, the inner shell
electrons contribute to the absorption of x-rays and give rise to the series of jumps at the
binding energy of these electrons. The value of can even be negative close to an absorption
edge for some of the materials. This is known as anomalous dispersion.
In the x-ray region, the small value of means that the real part of refractive index is
very close to unity. Therefore, the angle of refraction of x-rays through a material is very
Fig. 1.1 Calculated optical constants (delta and beta) for Si, B
4
C,
Y, C, Mo, Ni, Fe and Rh in the wavelength range 0.5 to 400 .
1 10 100
10
-10
10
-8
10
-6
10
-4
10
-2
10
0
Si
B4C
Y
C
Mo
Ni
Fe
Rh
O
p
t
i
c
a
l

c
o
n
s
t
a
n
t

(
b
e
t
a
)
Wavelength ()
10
-7
10
-5
10
-3
10
-1
Si
B4C
Y
C
Mo
Ni
Fe
Rh
O
p
t
i
c
a
l

c
o
n
s
t
a
n
t

(
d
e
l
t
a
)
Chapter 1
6
small. The conventional refractive lenses would have impossibly long focal lengths. In
addition to this, the radiation is exponentially attenuated on passing through the lens
materials. If E
0
is the incident electric wave vector, then the electric vector after passing
through a material of thickness d is given by

( ) Kr t i
d
e e E E

2
0
(1.3)
where is the absorption co-efficient and is the wavelength of incident radiation.
In term of intensity, if I
0
is the incident intensity, then the intensity of radiation after passing
through a material of thickness d is given by

) exp(
0
d I I = (1.4)

where
4
= is the linear absorption coefficient. Absorption in the lens material would
also mean that hardly any of the soft x-ray photons would be actually transmitted by the lens.
Therefore, the conventional refractive lenses used for the visible region are not suitable for
soft x-rays. Now let us consider the possibility of optics for the x-ray region as follows.

1.4 X-ray Optics

At normal incidence, x-ray reflectivity of all materials is very low (typically 10
-5
or
less), due to the small values of (cf. Chapter 2). Thus, it is not possible to use conventional
reflective optics for x-rays. The single surface conventional optics can be used only in
glancing incidence reflection geometry. High reflectivity at normal incidence can be obtained
by fabricating a multilayer structure with materials of alternately high and low atomic
numbers, known as a synthetic Bragg crystal, or a multilayer mirror. For x-rays with
wavelength in the range 20 -40 (and in some cases over a wider range), gratings and
zone plates and, recently, Bragg-Fresnel optics
7
as well, are used as the optical elements, all
of which are based on the diffraction principle. Since this thesis concerns x-ray multilayer
optics, we will confine discussion in what follows to reflective x-ray optics i.e., glancing
incidence optics and multilayer optics.

Chapter 1
7
1.4.1 Glancing Incidence Optics

Glancing incidence optics is based on the phenomenon of total external reflection.
Total external reflection of x-rays from a single surface occurs for incidence angles less than
the critical angle,

e o
c
r
= 2 . A detailed derivation of this expression for
c
is
given in Chapter 2. Therefore,
c
is proportional to the incident wavelength and to the square
root of the electron density of the material. At 54 . 1 = , the critical angle for different
materials is less than 0.65
0
. For example, in case of a gold mirror, the critical angle at
54 . 1 = is 0.55
0
, and is typically a few degrees in the soft x-ray region. Above the critical
angle, the reflectivity falls as q
-4
(cf Chapter 2), where / sin 4 = q . At normal incidence,
reflectivity is extremely small. The typical value of reflectivity from a single surface lies
between 10
-3
and 10
-6
. The calculated normal angle reflectivity from a molybdenum thin film
mirror is shown in Fig. 1.2 for different wavelengths, clearly illustrating the smallness of the
reflectivity of single surfaces.
Therefore, glancing angle x-ray optics operates at angles below the critical angle. The
choice of materials for a glancing incidence mirror depends on the application of the mirror.
The thermal properties of the material are of importance in high x-ray flux applications (e.g.
Fig. 1.2 Normal incidence reflectivity from a Mo thin film mirror and a Mo
[(26)/Si (40 )]
40
ML mirror. In the x-ray region, the reflectivity from thin film a
mirror is very small whereas, in multilayer structure, the enhancement of
reflectivity at the Bragg position gives a significantly high reflectivity.
0 100 200 300 400
10
-6
10
-4
10
-2
10
0
Mo thin film
Mo/Si ML
R
e
f
l
e
c
t
i
v
i
t
y

wavelength ()
Chapter 1
8
in synchrotron application), where thermal distortion of mirrors may occur. Therefore, an
ideal material would be one with a low thermal expansion coefficient, a high thermal
conductivity, and a low susceptibility to distortion through thermal spiking. Mirror materials
should also be chemically stable. A good choice of material would be , for example, SiC, Ni,
or Au.
The main disadvantages of glancing incidence optics are as follows: (i) In extreme
glancing angle geometry, the size of the optical element required is very large, due to the
large foot print of the beam. Again the focal length of the optical elements also becomes very
large, typically a tens of meter. (ii) In glancing incidence geometries, optical aberrations such
as astigmatism, spherical aberration, and coma become large
7
. (iii) Optical elements of a
large size have to be fabricated with high precisions in terms of surfaces roughness, slope,
figured etc., demanding the application of sophisticated technology. The afore-mentioned
disadvantages can be overcome to a great extent by using artificial multilayer structures,
which are discussed in the following section.

1.4.2 Multilayer Optics

The advancements in thin film fabrication technology open up the possibility of
fabricating artificial Braggs reflectors of the desired period
8
. This artificial Bragg reflector is
commonly known as an x-ray multilayer or x-ray multilayer mirror. It consists of alternating
layers of high-Z and low-Z materials, with individual layers having thicknesses of the order
of nanometers, which can be fabricated on a suitable substrate. A schematic diagram of such
a multilayer structure is given in Chapter 2. The thicknesses are so adjusted that the path
length difference between reflections from successive layer pairs is equal to one wavelength,
so that the amplitudes of x-rays from successive layer pairs add in phase to give the
maximum reflectivity at the Bragg angle due to constructive interference, as shown in Fig.
1.2 given above. The reflectivity increases as N
2
, where N is the number of contributing layer
pairs. These multilayer structures act as multilayer interference reflectors for x-rays, soft x-
rays, and extreme ultraviolet (EUV) light. These multilayer devices allow for accessing soft
x-ray/EUV wavelengths at normal incidence geometry giving high reflectivity (Fig. 1.2),
with moderate energy bandwidth (10<E/E<100), making them valuable as optical
components for exploiting the soft x-ray/EUV region.

Chapter 1
9
1.4.2.1 Advantages of Multilayers

The main advantages of multilayers optics stem from the tunability of layer the
thickness, composition, the lateral gradient and the perpendicular gradient. These may be
tailored according to the desired wavelength region. The perpendicularly graded multilayers
are commonly known as super mirrors and act as broadband reflectors
9, 10
. The MLs graded
laterally are used as wavelength selectors
11
. The great advantage of multilayer structures is
that they can be adapted to contoured surfaces, so that focusing and imaging are achieved.
They can operate at normal angle incidence, thus overcoming limitations associated with
glancing incidence mirrors.

1.4.2.2 Selection of Materials

The four criteria for selecting pairs of materials suitable for x-ray multilayer mirrors may
be stated as follows
12
:
i. First, a material with a low absorption coefficient at the desired wavelength must be
selected. This requires a low-Z material, generally known as the spacer.
ii. A second material is then selected, which has a large reflection coefficient at the
boundary with the first material. This requires a high-Z material, generally known as
the absorber. If several materials give similar reflection coefficients, the one with the
smallest absorption coefficient should be used.
iii. These two materials must form boundaries or interfaces, which should be physically
and chemically stable.
iv. The boundaries (interfaces) must be sharp and smooth, on the atomic scale.
The best material combinations for multilayer mirrors for different spectral regions may be
identified as follows
4
.

Wavelength Region < 23
In this short wavelength region, natural crystals, which have comparable periodicity, can
be used. For example, single crystals of organic and acid phthalate, such as potassium acid
phthalate, typically have a periodicity of 10 -25 . The main disadvantage of these crystals
is their poor thermal stability. Short period artificial multilayers can be used in this spectral
range to overcome the thermal stability problem associated with single crystals. To obtain
high normal incidence reflectivity, multilayer periodicity of the order of ~15 is required
Chapter 1
10
with an interfacial roughness less than 2 . Combinations of materials with a large difference
in the atomic number Z can produce good reflectivities for these short wavelengths. The
problem is associated with fabrication of these short period multilayers is the stringent
requirement for interfacial roughness. The deposition of ultra-thin continuous layers with
atomically smooth interfaces is a difficult task
13
, even with the advances in thin film growth
technology, preventing multilayers from being used as normal incidence mirrors at shorter
wavelength. Good spacer materials in this region are C and B
4
C. The pairs of materials for
MLs good for this region are ReW/C, W/C, Pt/C and W/B
4
C
14-16
.

Wavelength Region =23 - 44
This wavelength region lies between the oxygen K- absorption edge (=23.3 ) and the
carbon K- absorption edge (=43.7 ). This region is of special interest to the x-ray
microscopy of biological specimens. In this region, water is very transparent, while carbon-
containing (organic) materials have high absorption: the region is generally known as the
water window region. Therefore, in this regime, it is possible to observe live biological
specimens in transmission in their natural environment, which is water. For this region, the
materials with low absorption are Ti (L-edge at =31.4 ), Sc (L-edge at =31.9), and V
(L-edge at =24.3 ). The best material combinations for this wavelength range are Cr/Sc,
TiO
2
/Al
2
O
3
and Ni
80
Nb
20
/MgO
17-19
.

Wavelength Region =45 -124
For wavelengths > 45 , carbon is the best choice for spacer material. In this region,
carbon has low absorption, forms stable boundaries without diffusion with most metals, and
produces very smooth films. For the wavelength range = 44 - 66 , the highest
reflectivities are obtained when C is used as the spacer material. Multilayers of Ni/C, Co/C,
Mo/C and Fe/C have been reported by many groups
20-25
.
In the wavelength range above the boron K-edge ( > 66 ), boron has lower
absorption than carbon, and multilayers with boron as the spacer material give higher
reflectivity than those with carbon. However, multilayers with boron are less stable. On the
other hand, B
4
C gives very stable multilayers as well as lower absorption than carbon.
Various groups have reported good quality multilayers mirrors with B
4
C as spacer
26, 27
.
Chapter 1
11
In the wavelength range above the yttrium M-edge ( > 89 ), yttrium has a lower
absorption than B
4
C, and multilayer mirrors with Y as spacer give higher reflectivity than
those with B
4
C. Up to now, only few multilayers with Y as the spacer have been reported
28,
29
. In the wavelength range above the beryllium L-edge ( >111 ), Be becomes the best
spacer material and Rh/Be and Mo/Be multilayer mirrors have a theoretical normal incidence
reflectivity more than 80%. But, due to the health hazard of Be, only limited effort has been
made on Be-based multilayers
30, 31
. Although Y-based MLs gives less reflectivity (10%)
than Be-based ones in this wavelength region, Y-based MLs are still the preferred choice,
due to the health hazard posed by beryllium
32
.

Wavelength Region =125 -350
Silicon has very low absorption at wavelengths above its L-edge ( >124 ), and is
suitable as spacer material. Molybdenum gives a high reflection co-efficient with silicon and
also has lower absorption in this wavelength range than other high Z-materials. Therefore,
Mo/Si multilayers offer a theoretical normal incidence reflectivity greater than 80%
33
. But in
real multilayers, the reflectivity is greatly reduced from its theoretical value due to imperfect
interfaces (such as roughness, interdifusion, and compound formation), thickness errors,
layer continuity, and oxygen contamination.
34-38
. Therefore, it is important to enhance the
performance of the multilayers with improvement of these aspects. Interdiffusion at the
interfaces can be minimized, and hence the thermal stability of the Mo/Si multilayer
improved, by depositing a thin diffusion barrier layer (~5 ) of carbon between Mo and Si,
leading to a Mo/C/Si/C structure
39
or, analogously, to a Mo/B
4
C/ Si structure
29
. At
wavelengths beyond the Al L-edge ( >170 ), Al has less absorption than Si, and
multilayers with Al as spacer would give higher reflectivity than those with Si as spacer.
However, use of Al with other high Z-metals is restricted due to stability problems, and only
a few ML systems such as Mo/Al
40
have been reported. Therefore, Mo/Si remains the ML
system of choice for wavelengths up to 350 .

Wavelength Region > 350
The magnesium L-edge is at = 251 and, therefore, for the wavelength region =252
- 600 , suitable material pairs for stable structures are Pt/Mg
2
Si
41
and W/Mg
2
Si
42
. For
>600 , the absorption of all the materials increases with increasing wavelength, and the
Chapter 1
12
maximum number of layer pairs in the multilayer structure, N
max
(given in Eq.
n
5.6), becomes
smaller than 1, even for normal incidence. Therefore, single layers of Be, Al, or Mg are good
reflectors
43
above their critical wavelengths, which are 635 , 837 , and 1198 ,
respectively. The oxidation of metal surface can be protected with an overcoat of LiF or
MaF
2
.

1.4.2.3 Substrates

The quality of substrates in terms of surface roughness, figure errors, slope and
flatness, plays an important role in the performance of single layer and ML x-ray mirrors.
The root mean square roughness of the substrate should be less than
10
1
of the multilayer
period
4
. Due to improvement in surface polishing and testing techniques
44
,

polished
substrates are commercially available with roughness values 1 over a range of spatial
frequencies from 1 mm
-1
to 1 m
-1
. Well-polished mirror substrates with roughness values <
3 are also commercially available, in the high spatial frequencies above 10 m
-1
, important
for x-rays. The substrates generally used are zerodur, silicon carbide, fused silica, (single
crystal) silicon wafer, and float glass. Float glass and polished silicon wafers with rms
roughness values 2 -5 in the high frequency range are commercially available, and have
been used extensively to test the performance of coatings.
The propagation of the substrate roughness to the multilayers can be reduced with a
buffer layer of carbon, which heals the roughness of the substrate
45
. However, a thick buffer
layer on very good substrates has opposite effect. Since, roughness of a single film on a
perfect substrate always scales with thickness, therefore, thick buffer layer of carbon may
increases roughness than the bare substrate
46
.

1.5 Factors affecting the Performance of Multilayers

There are several reasons why the measured performance of x-ray multilayer (ML)
structures is lower than theoretical values. These may be enumerated are as follows.

i. Quality of the interface boundaries
The quality of the interface plays a decisive role in the performance of the multilayer
structure. The imperfections at the boundaries arise because of roughness, interdiffusion and
compound formation. Roughness has two distinct categories: low frequency and high
Chapter 1
13
frequency. Low frequency roughness is generally called figure error, which enlarges the
half-width of the focal spot. High frequency roughness increases diffuse scattering, which
results in reduced specular reflectivity. In general, if the interface is not sharp and has a width
due to high frequency roughness or due to interdifusion, the beams reflected from points
within the boundaries are in random direction, thus reducing specular component. The
diffuse scattering generates a scattering background, which diminishes the image contrast,
for example in imaging applications
47
.
If an interlayer is formed at the interfaces due to chemical reaction between two pure
elements, the refractive index contrast at the interfaces decreases, which results in reduced
reflectivity. A typical example of the effect of roughness and interlayer on the glancing angle
reflectivity in hard x-ray region, and on near-normal incidence reflectivity in the soft x-ray
region, as calculated theoretically using Parratts formalism
48
, is shown in Fig. 1.3. In Fig.
1.3(a) is plotted reflectivity vs. q
z
(momentum transfer vector normal to the sample surface)
whereas, in Fig. 1.3(b), reflectivity is plotted vs. the angle of incidence (degree) to help
visualize near-normal incidence reflectivity directly. As interface roughness increases, the
peak reflectivity decreases. The presence of an interlayer, produced by chemical interaction
at the interface, reduces peak reflectivity by reducing the contrast in the refractive index
between two pure materials. Therefore, it is clear that, the performance of the ML structure is
strongly dependent on the quality of the interface. The effect of roughness, interlayer
Fig.1.3 The effect of roughness and interlayer thickness on reflectivity patterns
of Mo/Si ML. The roughness and interlayer thickness are represented by and
t
i
,, respectively. (a) [Mo(20 )/Si(30 )]
10
ML at Cu K
( =130 ) radiation.
(b) [Mo(25 )/Si(40 )]
50
ML. at =130 .
0.00 0.05 0.10 0.15 0.20 0.25
10
-4
10
-3
10
-2
10
-1
10
0
10
1
Ideal Mo/Si ML
ML with =0 and t
i
= 10
ML with =10 and t
i
= 0
ML with =10 and t
i
= 10
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
(a)
0 20 40 60 80 100
0.0
0.2
0.4
0.6
0.8
1.0
Ideal Mo/Si ML
ML With =10
ML With =10 and t
i
= 10
R
e
f
l
e
c
t
i
v
i
t
y
Angle of incidence (degree)
(b)
Chapter 1
14
thickness, and its composition on hard x-ray reflectivity pattern is discussed in detail in
Chapters 5 and 6. Thus, it is important to understand the physics at the interfaces in x-ray
multilayer structures. Extensive studies are being made, by probing the interfaces with
different techniques
49-54
, to understand the interfacial behavior in MLs with different material
combinations, and fabricated using different deposition techniques.

ii. Thickness Errors
For a periodic structure without any errors in layer thicknesses, the reflected amplitudes
from adjacent periods add in phase, to give
maximum reflectivity at the Bragg angles. If a
ML structure (MLS) consists of N layer pairs,
there are N-2 regular Kiessig oscillations
between successive Bragg peaks, which
correspond to the total thickness of the MLS.
The thickness errors result in phase errors
between the reflected amplitudes from
successive periods, leading to reduced Bragg
peak reflectivity. In an error-free MLS, the
amplitude maxima are greater than the
surrounding minima
4
by a factor of
2 2
sin N
higher, where is the ratio of the thickness of
the high-Z layer to the period. The reduced ratio
of intensity of the maxima that of the surrounding peaks is a measure of thickness errors. The
effect of thickness errors on the reflectivity of Mo/Si ML is shown in Fig. 1.4. In a coating
with random thickness errors, the spacing in the Kiessig oscillations between Bragg peaks
becomes irregular, and the width of the Bragg peaks increases. The thickness errors also can
be derived from the width of the Bragg peaks
55
. Thickness errors are sensitive to the
deposition system adopted. Thus, while fabricating multilayer structures, one has to control
deposition parameters carefully, so that thickness errors are minimized.

iii. Layer Contamination and Continuity
When a pure layer gets contaminated with impurities, its optical response changes. The
Bragg peak reflectivity also gets reduced if the layers (especially the spacer layer) get
Fig. 1.4 The effect of thickness
error on reflectivity pattern
calculated at =1.54 for
[Mo(20 )/Si(30 )]
10
MLs
0.00 0.05 0.10 0.15 0.20 0.25
10
-4
10
-3
10
-2
10
-1
10
0
10
1
Ideal [Mo (20)/Si(30)]
10
ML
With 0.5% thickness error
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
Chapter 1
15
contaminated with impurities. Therefore, the optical properties of the layer change, thus
affecting the reflectivity at the wavelength for which the ML is designed
29
.
For a real MLS to exhibit its theoretical reflectivity, it is essential that its layers should be
continuous
56
with sharply defined boundaries, which results in large contrast in refractive
index at the boundaries, yielding maximum reflected amplitude. This requires the film
growth mechanism to be predominantly two-dimensional.

1.6 Thin Film Growth

The thin film formation involves one of three basic growth modes
57
: (a) island (or
Volmer-Weber), (b) layer-by-layer (or Frank-van der Merwe), and (c) Stranski-Krastanov,
which are shown in Fig.1.5 In island growth, the smallest stable clusters nucleate on the
substrate and grow to form islands. This happens when atoms or molecules are more strongly
bound to each other than to the substrate. Thin films of many metals, deposited on substrates
of insulators, alkali halide crystals, graphite, and mica, display this growth mode.
Characteristics opposite to this mode are observed during layer-by-layer growth.
Here, the extension of the smallest stable nucleus occurs overwhelmingly over the entire
substrate, resulting in the formation of planar sheets. In this growth mode, the atoms are more
strongly bound with substrate than with each other. The first complete monolayer is then
covered with a less tightly bound second layer. Provided the decrease in bonding energy is
continuous toward the bulk crystal value, the layer growth mode is sustained. An example of
layer growth involves single crystal epitaxial growth of semiconductor films.
The layer-plus-island or the Stranski-Krastanov growth mode is an intermediate
combination of the aforementioned modes. In this case, after forming one or more
monolayers, subsequent layer growth becomes unfavorable, and islands form. This transition
Substrate
Island Structure
Substrate
Uniform Film
Substrate
Uniform Film
Island Structure
(a) Volmer-Weber type (b) Frank-van der Merwe type (c) Stranski-Krastanov type
Fig.1.5 Schematic of different type of thin film growth
Chapter 1
16
in growth mode may be caused by any factor that disturbs the monotonic decrease in binding
energy that is characteristic of layer growth. For example, due to film-substrate lattice
mismatch, strain energy accumulates in growing film. When released, the high energy at the
deposit-intermediate-layer interface may trigger island formation. Examples of this growth
mode are observed in metal-metal and metal-semiconductor systems.

1.7 Applications of Multilayer Structures

A great advantage of the multilayer interference coatings is that they can be applied to
curved substrates for use in imaging applications, such as microscopes and telescopes.
Furthermore, they provide high normal incidence reflectivity with a modest, well-defined
spectral band pass, which finds many additional applications. Applications cover many areas
in science and technology, such as in x-ray lithography
58
, x-ray microscopy
59
, astronomy
60
,
spectroscopy
61
, and polarization-sensitive materials science
62
etc. Some of these applications
are outlined below:
i. Lithography: Fabrication of high-density integrated circuits with line width down to
a few 100 .
ii. Astronomy: Stellar objects e.g., neutron stars, emit copious amount of soft x-rays.
Their spectral properties provide information on the production mechanism of
radiation, magnetic field strength, on the elements present, etc. Multilayer-coated
normal incidence optics is used to obtain high-resolution astronomical images for
these purposes.
iii. Polarizers: A device, which converts incident electromagnetic waves to a particular
orientation of the electric vector, is known as a polarizer. EUV/Soft x-ray polarizers
became important only after the invention of synchrotron radiation. Synchrotron
source gives high intensity continuous radiation with a definite state of polarization.
A polarizer is used to measure the degree of polarization from a synchrotron source.
Radiation from bending magnets has polarization states ranging from linear-on-the-
axis to elliptical-off-axis (depending on the observation angle with respect to the
plane of the electron orbit). Complete characterization of the polarization of the
beams produced by synchrotron sources in general requires optical elements such as
linear polarizers to measure the azimuthal linear polarization dependence, combined
with quarter wave plates to ascertain phase relationships of different components
Chapter 1
17
within the beam. Alternatively, quarter-wave plates could be used to convert the
predominantly linearly polarized part of a traditional bending-magnet into a circularly
polarized beam. X-ray multilayer optics might be used as polarization conversion
devices in the EUV/Soft x-ray spectral range. The ability of the multilayers to act as
linear polarizers in the EUV/Soft x-ray spectral range, by positioning the multilayer
Bragg peak near 45
0
(Brewsters angle) to give a scattered angle near 90
0
, has been
demonstrated
63
. Further possibilities were opened up by a transmission multilayer
prepared as a free-standing film, in order to prevent absorption by the substrate. The
transmission multilayer is used to work as polarizer or a quarter wave plate depending
on the operational angle of incidence
64
. Therefore, using multilayer polarizers and
quarter wave plates, certain types of synchrotron experiments such as linear and
circular dichroism spectroscopy, magnetic scattering, and atomic and nuclear
resonant scattering, can be performed. Furthermore, because of EUV/soft x-ray
polarizers, EUV/soft ellipsometry is expected to be a sensitive tool for the study of
materials through the strong interaction between the electromagnetic wave and core
electrons of the atoms in a material
65
.

1.8 Aim of the Present Work

The aim of the thesis is to study surfaces and interfaces of x-ray multilayer structures,
to optimize their performance as polarizing elements. The quality of the
interfaces plays a decisive role in achieving optimum performance of the real
multilayer structures. The imperfect interfaces originate from roughness,
interdiffusion, and chemical reaction at the interfaces. The interface quality
and hence the performance of the multilayer structures is also strongly
dependent on the thin film growth process. To study thin film growth, and to
fabricate x-ray multilayers for the study of their surfaces and interfaces, and
eventually to fabricate x-ray optical devices, an ultra-high vacuum electron
beam evaporation system has been developed indigenously, forming a part of
this the present thesis research. The rest of the research effort has been divided
into four parts: (a) a study of the growth of ultra-thin molybdenum films at
different substrate temperatures, (b) a detailed study of the surfaces and
interfaces of x-ray multilayers and an evaluation of the actual performance of
Chapter 1
18
some of the MLs on the Indus-1 synchrotron radiation source, (c) a study of
the possibility of using soft x-ray resonant reflectivity as a non-destructive
technique to determine interlayer composition at buried interfaces in MLs and
(d) to obtain experimental evidence of elemental specificity of soft x-ray
resonant reflectivity through a measurement of the effect of anomalous optical
constants on the reflectivity profile of the low Z-element. In situ resistivity
measurement was set up for the study of thin film growth. Soft x-ray studies
were performed using the Indus-1 synchrotron source. Hard x-ray reflectivity
studies were performed using radiation from a sealed cupper target x-ray tube.
Other characterization tools used in the present studies are x-ray photoelectron
spectroscopy and glancing angle x-ray fluorescence.

Molybdenum thin film growth studies
For a real multilayer structure to exhibit the theoretically computed reflectivity, it is
essential that layers are continuous, with sharply defined boundaries. Reflectivity would be
enhanced by minimizing absorption. For this, the thickness of the layers must be small, but
above the percolation threshold, beyond which layers become continuous
56
. Only a clear
understanding of the nucleation and growth mechanisms of ultra-thin layers in such
structures would allow one to make the optimal choice of deposition parameters. Therefore,
it is important to investigate the growth mode and the discontinuous-to-continuous transition
in ultra-thin films. A percolating system can be studied experimentally using both electron
microscopy and in situ resistance measurements. The determination, using electron
microscopy, of the insulator-metal (I-M) transition region, fractional surface coverage and
hence critical coverage, may be erroneous due to contamination of the ultra-thin film when
exposed to air (in preparation for microscopy). In situ resistance measurement is a sensitive
method for the experimental study of percolating systems, and has been applied for various
materials. Examples are platinum
66
and nickel
67
. One of the difficulties in the experimental
determination of the critical exponent of conductivity is the determination of the critical
fractional coverage
c
x and the I-M transition region. In earlier work, the I-M transition
region and critical exponent of conductivity has been determined by the method of least
squares fitting
66, 67
of in-situ resistance data.
Chapter 1
19
In the present work, we have carried out detailed growth studies of ultra-thin Mo
films because Mo, in combination with spacer materials such as Si, C, Be, and B
4
C, forms
excellent x-ray multilayer mirrors for different spectral ranges. A detailed microscopic
picture of Mo films at different growth stages was derived from resistivity measured in situ.
The thesis also describes a simple method for experimental determination of the I-M
transition region and the critical exponent of electrical conductivity in a percolating system.
We have determined structural parameters such as the relative average nucleation density,
island area, and fractional surface coverage of island Mo films. The microstructure of island
film growth, derived from in situ sheet resistance measurements is in agreement with that
predicted by percolation theory.

Study of surfaces and interfaces in x-ray multilayers
The quality of interfaces plays an important role in optimizing the performance of
multilayer structures. The kinetics and thermodynamics at the interfaces of the multilayer
structures, which cause structural modifications, are important and interesting from the basic
research point of view as well.
This present thesis describes a detailed investigation of three different multilayer
structures viz., Mo/Si, Mo/Y, and Fe/B
4
C, meant for polarization application on the Indus-1
synchrotron source. Out of these three different multilayer structures, the Mo/Si system has
been studied extensively. The various surfaces and interfaces have been characterized
primarily by hard x-ray reflectivity, while and some samples were also characterized using
depth-profiled XPS and GIXRF. The actual performance of some of the MLs was tested on
the Indus-1 synchrotron source.

Soft x-ray resonant reflectivity
X-ray reflectivity (XRR) is an important non-destructive research tool for probing
surfaces and interfaces
48
in thin film multilayer structures. The large dynamic range and high
momentum scattering vector range (q
z
-range) of conventional hard XRR allow the use of
Fourier transform methods to study the micro-structural parameters of condensed matter
films, such as thickness, surface and interface roughness, and interface diffusion profile. It is
important to recognize that electrons are responsible for the interaction of x-rays with
materials and that x-rays actually probe effective electron density, which is directly related to
the index of refraction. Normally, in XRR, the x-ray photon energy is far from the absorption
Chapter 1
20
edges of materials of interest, and the poor contrast in optical constants is not sufficient to
identify the composition of the buried interfaces on a sub-nanometer scale. Generally, high
resolution cross-sectional transmission electron microscopy
68
and depth-graded
photoelectron spectroscopy methods
69
are used to identify composition of phases formed
buried interfaces at sub-nanometer resolution. However, these are destructive methods.
In this thesis, we have reported a non-destructive method for determination of the
composition at buried interfaces, by tuning the photon energy near the absorption edge. X-
ray resonant reflectivity is a surface-sensitivity and element-specific technique. Element
specificity is achieved by tuning the energy of the x-rays to the absorption edge of the
element being investigated, and XRR is suited for the investigation of buried interfacial
layers. Although the basic principle of the resonant effect at near atomic absorption edges has
been previously recognized
70, 71
, the sensitivity of resonant x-ray reflectivity to determine the
composition of a phase formed on a sub-nanometer scale at buried interfaces has not yet been
still examined. In this thesis, we have extended the resonant principle to determine the
composition of interlayer formed at interfaces of a well-characterized Mo-Si multilayer
system, through the use of fine structure features of atomic scattering factors at the L-
absorption edge of silicon. We have demonstrated the sensitivity of soft x-ray resonant
reflectivity to interlayer composition theoretically and, using the Indus-1 synchrotron
radiation, experimentally. This should in general be applicable to any bilayer/multilayer
system near the resonant absorption edges.
In another study, we have observed the anomalous optical constant of the B
4
C
material at boron K-edge, where the real part of the forward atomic scattering factor
undergoes sign reversal. Therefore, by tuning the photon energy such that the refractive
index of the material would be very close to the refractive index of vacuum, that material
becomes virtually invisible We have observed the elemental specificity property of soft x-ray
resonant reflectivity by showing the effect of the anomalous optical constant on reflectivity
profile, through the characterization of B
4
4
C bi-layers.
Chapter 1
21
1.9 References

1.
W.C. Rntgen, Sitzungsberichte Med. Phys. Gesellschaft Wrzburg, pp. 137-141
(1895).

2.
W. Friedrich, P. Knipping, M. von Laue, itzb, Math.-Phys. Bayr. Akad. Wiss. pp.
303-322 (1912).

3.
A.G. Michette and C. J. Buckley X-ray Science and Technology (Institute of
Physics Publishing, London, 1993).

4.
E. Spiller, Soft X-ray Optics (SPIE Optical Engineering Press, Washington, 1994).

5.
Ernst-Eckhard Koch, Hand Book on Synchrotron Radiation(North-Holland
Publishing Company, 1983)

6.
B.L. Henke, E.M. Gullikson, J. C. Davis, Atomic Data Nucl. Data Tables 54, 181
(1993).

7.
A.G. Michette, Optical systems for Soft X-rays (Plenum Press, New York, 1986).

8.
T.W. Barbee, Jr., S. Mrowka, M. C. Hettrick, Appl. Opt. 24, 883 (1985).

9.
Y. Tawara, K. Yamashita, H. Kunieda, K. Tamura, A. Furuzawa, K. Haga, N.
Nakajo, T. Okajima, H. Takata, P. J. Serlemitsos, J. Tueller, R. Petre, Y. Soong, K.
Chan, G. S. Lodha, N. Namba, J. Yu., SPIE 3444, 569 (1998).

10.
H. Wang, J. Zhu, Z. Wang, Z. Zhang, S. Zhang, W. Wu, L. Chen, A.G. Michette,
A.K. Powell, S.J. Pfauntsch, F. Schafers, A. Gaupp, Thin Solid Films 515, 2523
(2006).

11.
E. Ziegler, O. Hignette, Ch. Morawe, R. Tucoulon, Nucl. Instr. and Meth. A 467-468,
954 (2001).

12.
E. Spiller, in Low Energy X-ray Diagnostics edited by D. T. Attwood and B. L.
Henke, AIP Conference Proceedings No. 75, pp.124-130, New York (1981).

13.
C.C. Walton, G. Thomas, and J. B. Kortright, Acta Mater. 46, 3767 (1998).

14.
E. Ziegler, Y. Lapetre, I.K. Schuller, E. Spiller, Appl. Phys. Lett. 48, 1354 (1986).

15.
J. B. Kortright, S. Joksch, E. Zieger, J. Appl. Phys. 69, 168 (1991).

16.
R.F.E. Christensen, Z. Shou-Hua, A. Hornstrup, H.W. Schnopper, P. Plag, J. Wood, J.
X-ray Sci. Technol. 3, 1 (1991).

17.
F. Erikssen, G. A. Johansson, H.M. Hertz, E.M. Gullikson, U. Kreissig, J. Birch, Opt.
Lett. 28, 2494 (2003).

Chapter 1
22
18.
H. Kumagai, K. Toyoda, K. Kobayashi, M. Obara, Y. Limura, Appl. Phys. Lett. 70,
2338 (1997).

19.
S. Vitta, M. Weisheit, T. Scharf, H.U. Krebs, Opt. Lett. 26, 1448 (2001).

20.
L. Golub, E. Spiller, R.J. Barlett, M.P. Hockaday, D.R. Kania, W.J. Trela, R. Tacyn,
Appl. Opt. 23, 3529 (1984).

21.
H.V. Brug, M.P. Bruijn, R.V. Pol, M.J.V. Wiel, Appl. Phys. Lett. 49, 914 (1986).

22.
J. Verhoeven, M.J. Vanderwiel, E. Puik, Vacuum 39, 711 (1989).

23.
K. Yamashita, M. Ohtani, Y. Ueno, H. Tsunemi, S. Kitamoto, I. Hatsukade, Rev. Sci.
Instrum. 60, 2006 (1989).

24.
M. Arbaoui, R. Barchewitz, C. Sella, K. B. Youn, Appl. Opt. 29, 477 (1990).

25.
B.L. Evans and Xu Shi, J. Mod. Opt. 39, 695 (1992).

26.
D.G. Stearns, R. S. Rosen, P. Vernon, Opt. Lett. 16, pp 1283 (1991).

27.
F.E. Christensen, Z.S. Hua, A. Hornstrup, H.W. Schnopper, P. Plag, J. Wood, J. X-
ray Sci. Technol.3, 1 (1991).

28.
J. Nilsen, S. Bajit, H.N. Chapman, F. Staub, J. Balmer Opt. Lett. 28 (22), 2249
(2003).

29.
B. Kjornrattanawanich and S. Bajit, Appl. Opt. 43 (32), 5955 (2004).

30.
T.W. Barbee, Jr., Rev. Sci., Instrum. 60, pp 1595 (1989).

31.
R. Philip, R. Rivoira, Y. Lepetre, G. Rasigni, Appl. Opt. 27, pp 1918 (1988).

32.
C.Montcalm, B.T.Sullivan, S.Duguay, M.Ranger, W. Steffens, H. Pepin, M. Chaker,
Opt. Lett., 20, 1450 (1995).

33.
http://www-cxro.lbl.gov/optical_constants/multi2.html.

34.
D.G. Stearns, R.S. Rosen, S.P. Vernon, J. Vac. Sci. Technol. A 9, 2662 (1991).

35.
D.L. Windt, R. Hull, W.K. Waskiewicz, J. Appl. Phys. 71, pp. 2675 (1992).

36.
E. Meltchakov, V. Vidal, H. Faik, M.J. Casanove, B. Vidal J. Phys.: Condensed
Matter 18, 3355 (2006).

37.
M. Nayak, M.H. Modi, G.S. Lodha, A.K. Shrivastava, P. Tripathi, A.K. Sinha, K. J.S.
Sawhney, R.V. Nandedkar, Nucl. Instr. and Meth. B 199, 128 (2003).

38.
T.W. Barbee, Jr., J.C. Rife, W.R. Hunter, M.P. Kowalski, R.G. Cruddace, J. F. Seely,
Appl. Opt. 32, 485 (1993).

39.
H. Takenaka, H. Ito, T. Haga, Kawamura, J. Synch. Rad. 5, 708 (1998).

Chapter 1
23
40.
Q. H. Guo, J.J. Shen, H. M. Du, E.Y. Jiang, H. L. Bai, J. Phys. D: Appl. Phys. 38,
1936 (2005).

41.
T. W. Barbee, Jr., MRS Bulletin 15(2), 37 (1990).

42.
P. Houdy, M. Kuhne, P. Muller, R. Barchewitz, J. P. Delaboudiniere, D. J. Smith, J.
X-ray Sci. Technol. 3, 118 (1992).

43.
M.L. Scott, in Short Wavelength Coherent Radiation: Generations and
Applications, edited by R. W. Falcone, J. Kirz, p. 322, Opt. Soc. America (1988).

44.
P.C. Backer, Advanced flow-polishing of exotic optical materials, Proc. SPIE 1160,
263 (1989).

45.
H.J. Stock, F. Hamelmann, U. Kleineberg, D. Menke, B. Schmiedeskamp, K.
Osterried, K. F. Heidemann, U. Heinzmann, Appl. Opt. 36, 1650 (1997).

46.
D.E. Savage, N. Schimke, Y.H. Phang, M.G. Lagally, J. Appl. Phys., 71, 3283
(1992); K.B. Nguyen, T.D. Nguyen, J. Vac. Sci. Technol. B 11, 2964 (1993).

47.
E. Spiller, D. Stearn, M. Krumery, J. Appl. Phys. 74, 107 (1993).

48.
L.G. Parratt, Phys. Rev. 95, 359 (1954).

49.
M.J.H. Kessels, F. Bijkerk, F.D. Ticheelaar, J. Verhoeven, J. Appl. Phys. 97, 093513
(2005).

50.
D.L. Windt, F.E. Christensen, W.W. Craig, C. Hailey, F.A. Harrison, M. Jimenez-
Garate, R. Kalyanaraman, P. H. Mao, J. Appl. Phys. 88, 460 (2000).

51.
A. Paul and G. S. Lodha, Phy. Rev. B. 65, 245416 (2002).

52.
M. Nayak, G.S. Lodha, P. Bhatt, S.M. Choudhari, R.V. Nandedkar, J. Electron
Spectroscopy and Related Phenomena 152, 115 (2006).

53.
A. Ulyanenkov, R. Matsuo, K. Omote, K. Inab,a, J. Harada, J. Appl. Phys. 87, 7255
(2000).

54.
G.S. Lodha, S. Pandita, A. Gupta, R.V. Nandedkar, K. Yamashita, Appl. Phys. A 62,
29 (1996).

55.
A.D. Akhsakhalyan, A.A. Fraerman, N.I. Polushkin, Y.Y. Platonov, N.N.
Salashchenko, Thin Solid Films, 203, 317 (1991).

56.
M Veldkamp, H. Zabel, Ch. Morawe, J. Appl. Phys. 83, 155 (1998).

57.
K. Wasa and S. Hayakawa, Handbook of Sputter Deposition Technology, Noyes,
Westwood, New Jersey, USA, (1992).

58.
A.M. Hawryluk and L.G. Sepala, J. Vac. Sci. Technol. B 6, 2162 (1988).

Chapter 1
24
59.
A.G. Michette, G.R. Morrison, C.J. Buckley, X-ray Microscopy III, Springer-
Verlag, Berlin (1992).

60.
A.B.C. Walker, T.W. Barbee, Jr., R.B. Hoover, J.F. Lindblom, Science 241, 1781
(1988).

61.
G.B. Stephenson, Nucl. Instrum. Meth. Phys. Res. A, 266, 447 (1988)); J. F. Seely,
C.M. Brown, Appl. Opt. 32, 6288 (1993).

62.
H.C.N. Tolentino, J.C. Cezar, Narcizo, N.M. Souza-Neto, A.Y. Ramos, J. Synch.
Rad. 12, 168 (2005).

63.
B. Kjornrattanawanich, R. Soufli, S. Bajt, D. L. Windt, J.F. Seely, Proc. SPIE 5538,
17 (2004).

64.
T. Hatano, W. Hu, M. Yamamoto, M. Watanabe, J. Electron Spectroscopy and
Related Phenomena, 92, 311 (1998).

65.
T. Kawamura, J.J. Delaunay, H. Takenaka, T. Hayashi, Y. Watanabe, J. Synch. Rad.
5, 735 (1998).

66.
I. Ostadal and R.M. Hill, Phys. Rev. B 64, 033404 (2001).

67.
L. Cheriet, H. H. Helbig, and S. Arajs, Phys. Rev. B 39, 9828 (1989).

68.
J. C. Chen, G.H. Shen, L.J. Chen, J. Appl. Phys. 83, 7653 (1998).

69.
S. Hofmann, Appl. Surf. Sci. 241, 113 (2005).

70.
C.- C. Kao, C.T. Chen, E.D. Johnson, J. B. Hastings, H.J. Lin, G.H. Ho, G. Meigs, J.-
M. Brot, S.L. Hulbert, Y.U. Idzerda, C. Vettier, Phys. Rev. B 50, 9599 (1994).

71.
D.R. Lee, S.K. Sinha D. Haskel, Y. Choi, J.C. Lang, S.A. Stepanov , G. Srajer, Phys.
Rev. B 68, 224409 (2003).

Chapter-2
25
CHAPTER-2

Characterization Techniques and Data Analysis

This Chapter describes in detail the experimental techniques and data analysis methods
used for characterization of the samples. The characterization tools used in the present
studies involve glancing angle hard x-ray reflectivity, soft x-ray reflectivity, in situ resistivity
measurements, x-ray photoelectron spectroscopy (XPS), and glancing incidence x-ray
fluorescence (GIXRF). X-ray reflectivity (XRR) is the major characterization technique used
in this thesis for the investigation of surfaces and interfaces. Because XRR has been used
extensively, this Chapter provides a detailed discussion of XRR as a non-destructive
characterization tool for the investigations of the surface and interfaces of substrates, single
layers, bilayers and multilayers. Data analysis methods for XRR from multilayer structures
based on dynamical and kinematical models has been discussed. The effect of roughness on
XRR has been discussed based on the recursion formalism of dynamical theory. Reflectivity
is simulated using a computer program based on Parratts recursion formalism and the same
is used to fit experimental data. The XRR experimental set-up also has been presented. The
different techniques of resistivity measurements are briefly discussed. The basic principles of
XPS with emphasis on spectral interpretation are presented. Finally, the experimental setup
of GIXRF is also discussed.
Chapter-2
26
2.1 Introduction

It is important to have a feedback loop for the fabrication of good quality x-ray multilayer
structures (MLS). Experimental data on the performance of the MLS made in one
deposition run should be available before the next run, and the analyzed data should yield
sufficient information to guide the way the deposition run should be modified to obtain
improved performances. This information cannot be obtained from a test of the mirror for
the intended application alone; other tools are needed to characterize the MLS more
completely. Insufficient reflectivity can result from interfacial roughness, interface mixing
and compound formation, thickness errors, discontinuous nature of layers, the crystalline
nature of the layers etc. Hence, it is desirable to employ characterization techniques that can
separate these effects. Techniques used to extract the above-mentioned important
parameters via. XRR, in situ resistivity measurements, XPS, and GIXRF, have been
discussed in this Chapter.

2.2 X-ray Reflectivity

The principal objective of x-ray reflectivity experiment is the determination of the one-
dimensional scattering potential perpendicular to the surface of the sample. When the sample
is stratified and its chemical composition is known, the XRR data can be related to its
chemical profile. Typical parameters that can be deduced from the experiment are the
thickness, mass density and interfacial roughness of the individual layers of a stratified
sample. The structural characterization with x-rays is a non-destructive way of studying the
surfaces and interfaces on the atomic scale. The range of in hard x-ray reflectivity
experiment is small, typically between 0 to 4. XRR is a coherent elastic scattering
technique. For reflectivity, a change of the scattering potential (or chemical density) may
cause interference, whereas this is not so for diffraction due to the long-range periodical
order.

2.2.1 Scattering Geometry
When an X-ray beam incident on a medium it induces, as in the case of diffraction,
secondary re-radiating dipole centers by interaction with the individual outer electrons of the
atoms that constitute the material. In the co-planar geometry for x-ray scattering, the incident
wave, the reflected wave, and the normal to the reflecting surface lie in same
Chapter-2
27
plane
1
. In the specular reflection condition, both the incident and reflected waves, defined by
the wave vectors k and k
, respectively, make equal angles with the surface boundary, i.e.,

=
, as shown in Fig. 2.1. This is called the specular condition and it means that the change
in momentum transfer is perpendicular to the physical surface of the material, as shown
in Fig. 2.1(b). The intensity and phase of the reflected waves as a function of incidence angle
depend on the vertical spatial distribution of the electron density in the material. In Fig. 2.1,
since both incident and reflected wave propagate in the same medium (vacuum), the
magnitudes of wave vectors are equal, i.e., |k|=|k
|=2. The change in momentum transfer

vector perpendicular to the surface is defined as
2
k=q=k
-k. The magnitude of momentum

transfer vector is

|k|=2ksin (2.1)

and there is no momentum change in the x- or y- direction (Snells law). In specular x-ray
reflectivity measurements, the reflected intensity is measured as a function of momentum
transfer perpendicular to the surface q. The magnitude of q is changed either by varying the
angle at fixed wavelength or by varying the wavelength at a fixed angle . The former
is the most commonly used method and hence has been adopted in this thesis work.
(a)
k
vacuum

z
o
k
k
2
(b)
K

K
n=1-+i
Incident wave
Refracted wave
Reflected wave
Fig.2.1 (a) Interface geometry for incident, reflected, and refracted waves at the
interface between vacuum and the medium with refractive index n=1-+i. The
plane of incidence is defined as containing the incident wave vector k and the
surface normal z
o
(b) Scattering of the wave vector.
Chapter-2
28
A general case of scattering geometry is shown in Fig. 2.2. The wave vectors of the
incident and scattered x-rays are k and k
, with incidence angle and exit angle
, which are
not equal. Then, there is momentum transfer along x-direction. This geometry is used for
diffuse scattering experiments to get information about the lateral surface structure. The
wave vector transfer is q=k
-k, and can be resolved into z- and x-components, taking into

account the effect of refraction, as
3

( ) ( ) | |
2 1
" 2 2
2 1
2 2
z
cos cos
2
q
+ = n n (2.2(a))
| |
cos cos
2
q
"
x
= (2.2(b))

The z- and x-components of the momentum transfer vector can vary experimentally by
independent and coupled movement of sample and detector in scans, as follows
i. Specular Scan: The sample and the detector are moved in such a way that the
incidence angle and the reflected angle are the same, i.e., =
. In this case, q
x
=0 and
q
z
0. This scan also called the -2 scan.
ii. Off-specular Scan: The sample and detector are moved in such a way that a constant
off angle is maintained between the incident and the reflected beams, i.e.,
-=.
iii. Rocking scan: In this case, the detector is fixed at a +
value, and the sample is

rocked around the central value of (+
)/2.
iv. Detector Scan: In this case, the sample is fixed at a constant incidence angle, and
detector is scanned, i.e., is fixed, and
is varied.

x
z
y
k
K
q
q
z
q
x

Fig. 2.2 Scattering geometry for general case of scattering where the incidence
angle and reflected angles are not equal. The momentum transfer along the x- and
z-direction is given by q
x
and q
z
, respectively. x-y is the sample plane, whereas x-z
is the scattering plane.
Chapter-2
29
2.2.2 Reflection and Refraction at an Interface

Let us consider an electromagnetic wave with frequency and the wave vector k incident
from vacuum on a material at an angle , as shown in Fig. 2.3 (a). Then, the incident, the
refracted, and the reflected electric field vectors may be written as,

) . (
0
r k t i
e E E

=

(incident wave) (2.3(a))

) . ( '
0
'
'
r k t i
e E E

=

(refracted wave) (2.3(b))

) . ( "
0
"
"
r k t i
e E E

=

(reflected wave) (2.4 (c ))
All waves have the same frequency, , and
c
k k

= =
"
(since both waves are in vacuum).
The refracted wave has the phase velocity
n
c
k
V = =
'
'
, thus ( )
i
c
k + = 1
'

(2.5)
where and are the real and imaginary parts of refractive index, and are discussed in
further detail in Chapter 6.
Boundary conditions:
i. The components of E and H vectors parallel to the interface must be continuous.
ii. The components of D and B vectors perpendicular to the interface must be
continuous.
Incident wave Reflected wave
Refracted wave
Vacuum
n=1
n=1-+i
k k

x
z
c
<
c
Totally reflected wave
(a)
(b)
Fig.2.3 (a) Schematics of incident, reflected, and refracted electromagnetic waves at
the interface between vacuum and a material with n=1-+i. The plane of incidence
contains the incident wave vector k and the normal to the surface (b) showing the
total reflection condition
Chapter-2
30
We consider the incident wave vector k to lie in the x-z plane. If the parallel field
components are to be continuous along the interface, then the phase and amplitude
variations of all waves must be identical along the interface. This requires that the x-
components of the wave vectors must be equal at z=0,

" " ' '
cos cos cos k k k = = (2.6)

Case1: Since k and k
propagate in vacuum, they are real and equal in magnitude

"
= (2.7)
i.e., the angle of incidence is equal to the angle of reflection.
Case2: From Eq.
n
2.6,
n
cos
cos
'
= (2.8)
This Eq.
n
is known as Snells law. This gives the refractive turning of a wave inside a
medium with complex refractive index n. Therefore, for a real incidence angle , both the
wave vector k
and turning angle
, in the medium, have real and imaginary components.

2.2.2.1 Total External Reflection

For x-rays, the real part of the refractive index is smaller than unity
4
. Therefore, Snells
law indicates that the radiation is refracted in a direction slightly away from normal to the
surface. Thus, there exists an incidence angle for which the refraction angle
can be equal
to zero, indicating that the refracted wave doesnt penetrate into the material but, rather,
propagates along the interface. To quantify the critical angle with minimal mathematical
complexity, we assumed =0. At the critical angle
c
= , 0
'
= . From Snells law,

2 =
c
(2.9 (a))
Z
c
(Z is atomic number) (2.9 (b))

Chapter-2
31
At any angle below critical angle, the incident x-ray gets totally externally reflected, as
shown in Fig. 2.3 (b). Below the critical angle, the wave vector inside the medium
(
i 2 2 sin
2
2
) is purely imaginary, and the wave cannot propagate inside the
medium for all the incidence angles
c
. All the incident radiation gets reflected, except
for small losses due to absorption. In the critical angle region, the penetration depth is
restricted to a few nanometers. Therefore, x-ray reflectivity is surface-sensitive at a glancing
angle of incidence
5
. A typical penetration depth in molybdenum and silicon in the hard x-ray
region (=1.54) is shown in Fig. 2.4, as a function of the glancing angle, normalized to the
critical angle (
c
= 0.435
0
for Mo and
c
= 0.22
0
for Si). At near the critical angle, the
penetration depth is only a few tens of angstroms. As the glancing angle increases above the
critical angle, the penetration depth increases drastically. The penetration is more in Si than
in Mo because of lower absorption in Si than in Mo.

2.2.3 Reflection Coefficients at an Interface

Let us consider a single smooth interface between vacuum (n=1) and a medium with n=1-
+i, as shown in Fig. 2.3 (a). The Fresnel reflection and transmission coefficients, r and t,
Fig. 2.4 The calculated penetration depth for molybdenum and silicon as a function
of glancing angle of incidence normalized to critical angle, at wavelength =1.54.
At this ,
5
10 88 . 2

= and
6
10 88 . 1

= for molybdenum, and
6
10 67 . 7

=
and
7
10 76 . 1

= for silicon.
0 5 10 15
10
1
10
2
10
3
10
4
Mo
Si
p
e
n
e
t
r
a
t
i
o
n

d
e
p
t
h

(
)
/
c
Chapter-2
32
for s-polarized (electric field vector perpendicular to the plane of incidence) and p-polarized
(electric field vector parallel to the plane of incidence) can be written as
6

2 2
2 2
cos sin
cos sin
+

=
n
n
r
s
(2.10 (a))

2 2 2
2 2 2
cos sin
cos sin
+

=
n n
n n
r
p
(2.10 (b))

2 2
cos sin
sin 2
+
=
n
t
s
(2.10 (c))

2 2 2
cos sin
sin 2
+
=
n n
t
p
(2.10 (d))

Since n itself is a complex quantity, these coefficients are also complex quantities. The
measured real quantity is the Fresnel reflectivity, which is the modulus square of the
reflection coefficient amplitude, i.e., R=
2
r . Note that both s- and p- polarized reflectivity
yield nearly identical results for glancing angle reflection. The simple form of the Fresnel
reflection co-efficient for s-polarized radiation can be written in terms of the momentum
transfer vector (
sin
4
n q = ) as,

medium vacuum
medium vacuum
s
q q
q q
r
+
= (2.11)
Case 1: At glancing incidence
For
c
, and where 1 2 << =
c
, the reflectivity from Eq.
n
(2.10 (a)) becomes

( )
( )
2
2 2
2
2 2
2
2

i
i
R
c
c
s
+ +
+
= ) sin ( (2.12)
Chapter-2
33
The reflectivity from a single
surface, as function of angle of
incidence normalized to the critical
angle, is shown in Fig. 2.5
(reproduced from Ref. 2). Below the
critical angle, the reflectivity is
nearly equal to 100% for negligible
absorption, . A finite and small
rounds out the sharp angular
dependency. The external reflection
region (<
c
) becomes more
sensitive as the absorption increases
or increases. For all >>
c
, the
reflectivity is inversely proportional to the fourth power of the momentum transfer vector.

Case 2: At normal incidence
For
0
90 = , 1 << and 1 << , the
reflectivity from Eq.
n
(2.10 (a)) becomes,

4
1
1
2 2
2
2
+
=
n
n
R
s

(2.13)
The normal reflectivity from an ideal
molybdenum surface (roughness zero) is
shown in Fig. 2.6. This clearly shows that the
normal reflectivity is very small. For example
at =130 ,
Mo
= 6.9310
-2
and
Mo
=5.610
-3
,
the reflectivity is only 110
-3
. The reflectivity
increases as the fourth power of wavelength.

Fig. 2.5 X-ray reflectivity of a single surface
at =1.54 vs. angle of incidence normalized
to the critical angle, for different ratios.
Fig.2.6 Calculated normal
angle (89
o
) reflectivity of ideal
molybdenum surface
0 100 200 300 400
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
Mo thin film
R
e
f
l
e
c
t
i
v
i
t
y

wavelength ()
Chapter-2
34

2.2.3.1 Brewsters Angle for X-rays

The minimum in p-polarized reflectivity occurs when the numerator in Eq.
n
2.10 (b)
becomes zero. Therefore, from geometrical interpretation,

n
B
= tan
(2.14 (a))

Since n is complex, Eq.
n
2.14 (a) does not
give a real angle for which R
p
will be zero;
rather, a minimum is achieved. Expanding
the reflectivity R
p
in and , and setting
derivative w.r.t to zero, the real Brewsters
angle is given by

2 4

B
(2.14 (b))

The physical interpretation of Brewsters angle is given in Fig. 2.7 for the parallel component
of the electric field. At Brewsters angle, k
and k
vectors are perpendicular to each other. In

this case, the atoms at the interface respond to the impressed field
'
0
E , oscillating in a
direction parallel to k
, each radiating a sin

2
pattern which is zero in the reflected direction,
and producing no reflected field component
"
0
E . But, due to a finite absorption in the
material,
"
0
E is not absolutely zero. Fig. 2.8 shows the s- and p-polarized reflectivity vs.
incidence angle for an ideal Mo surface at different photon wavelengths. The residual p-
n=1-+i
n=1
E
0
E
0

E
0
k
k

k

B
sin
2

radiation
pattern
B
Fig.2.7 Schematic of the Brewster angle
B
for which there is no reflected component,
for parallel polarization
Fig. 2.8 The s- and p-polarized reflectivity from molybdenum surfaces at
different photon wavelengths: (a) =1.54 , (b) =130 and (c) =300
0 15 30 45 60 75 90
10
-16
10
-13
10
-10
10
-7
10
-4
10
-1
S-polarized
P-polarized
R
e
f
l
e
c
t
i
v
i
t
y
angle of incident (degree)
at =1.54
(a)
0 15 30 45 60 75 90
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0 S-polarized
P-polarized
R
e
f
l
e
c
t
i
v
i
t
y
at =130
(b)
0 15 30 45 60 75 90
10
-3
10
-2
10
-1
10
0
S-polarized
P-polarized

R
e
f
l
e
c
t
i
v
i
t
y
at =300
(c)
Chapter-2
35
polarized reflectivity at the minimum is caused by absorption of radiation in the material. As
the photon wavelength increases, the absorption becomes stronger, hence producing a
shallower minimum. For longer wavelengths, the absorption increases and, hence, the ratio of
s- and p-polarized reflectivity drops. At the Brewsters angle, the phase changes by 180
0
for
p-polarization.

2.2.4 Reflectivity from a Single Film (a Single Layer)

For a thin film deposited on a substrate, there are two interfaces: the vacuum/film
interface and the film/substrate interface. The interference of the waves reflected from these
two interfaces gives rise to the final reflectivity pattern of the structure. Let r
12
and r
23
be the
Fresnel reflection coefficient from the vacuum/film and the film/substrate interfaces,
respectively. Then, the final reflectivity from the thin film structure can be written as
7

( )
( )
2
23 12
23 12
2
2 exp 1
2 exp
i r r
i r r
r R
+
+
= = (2.15)
where is the phase difference between the wave reflected from the two interfaces. The most
remarkable feature is the occurrence of
intensity oscillation in the reflectivity
pattern as function of incidence angle or
normal component of momentum transfer
vector arising from the phase difference .
This phase difference is due to the path
difference between the beams reflected
from the two interfaces. and are
related by =
2
. These intensity
oscillations are known as Kiessig
oscillations or fringes
7
. From Fig. 2.9, the
path difference between the two
reflected beams can be written as
AD n BC AB + = ) (

2 2
sin sin 2 sin 2
c t
t tn = = (2.16 (a))
Fig.2.9 Schematic representation of
reflected and refracted beams from a
single film showing the deviation of
their phase difference
t
A
B
D
C

t
Thin film
Substrate
Vacuum
Chapter-2
36
where n and t are the refractive index and the thickness of the film, respectively.
c
is the
critical angle. This equation is analogous to the Bragg equation, modified by the influence of
refraction through m = . Since, in most cases, the angle of incidence () is sufficiently
small, we may use the approximation sin, to get

2 2
2
c
t = (2.16 (b))

Maxima are observed whenever the path difference is a multiple of , i.e., m = , where m
is an integer. Indexing the oscillation maxima by m, we get

2 2 2
2
m
t
c m
|
.
|
\
|
+ =

(2.16 (c))

The slope, s, of the straight line in the
2
m
vs.
2
m plot gives thickness information, i.e.,
s
t
2
= . In modern XRR evaluation

programs, the determination of t is
performed automatically by fitting the
reflectivity curve with the thickness t as a
definable parameter. It is also clear from
Eq.
n
(2.16 (c)) that the angular distance
between two adjacent fringes for
m+1
and
m
is
t
m m
2
1

= =
+
. In terms of the
difference in wave vector transfer between
any two successive maxima or minima q,
the thickness of the film is given
as q / 2 . There are two important
aspects in the oscillating behavior of reflectivity from thin films, i.e., the period of oscillation
and the amplitude of oscillation. The former depends inversely on the thickness and directly
on the wavelength, and the latter depends directly on the contrast in electron density between
the film and the substrate. This is illustrated in Fig. 2.10, which shows the calculated
reflectivity at Cu-K
radiation (=1.54 ) for single layer thin films of Mo (density =10.2

Fig. 2.10 Calculated reflectivity for ideal
(zero roughness) Mo and Al films of
thicknesses 100 and 500 on Si-
substrate at wavelength =1.54
0 1 2 3 4 5 6
10
-7
10
-5
10
-3
10
-1
10
1
100
500
R
e
f
l
e
c
t
i
v
i
t
y
(degree)
Al
Mo
Chapter-2
37
gm/cm
3
) and Al (density =2.7 gm/cm
3
), with a thickness of 100 and 500 , deposited on
Si-substrate.

Experimental limitations:
The oscillation maxima become more closely spaced with increasing layer thickness. For
example: for =1.54 and t=1500 , the difference in the scattering angle 2 is of the order
of 0.06
0
only. Because the resolution in 2 cannot be made arbitrarily small, the measurable
thickness of thin films by XRR is limited to certain maximum value. The precise value of the
upper limit depends on the instrumental resolution (such as divergence of beam) and the
various sample parameters, such as thickness inhomogeneity, surface and interface
roughness, and other factors that contribute to the washing out of sharp fringes. As a thumb
rule, the thickness values of single layers in excess of about 1000 are hardly measurable,
when Cu-K
radiation is employed because of basically divergence of beam limit.

2.2.5 Reflectivity from a Bi-layer Structure

In the case of a bilayer film (film 1 and film 2) of different materials on the substrate,
there are three interfaces, i.e.,
vacuum/film 1, film1/ film2, and film 2/
substrate. The final reflectivity stems
from the interference of scattered
radiation from the three interfaces. The
reflectivity can be calculated in a manner
similar to that in the single layer film. In
case the thicknesses of the two films are
different, the oscillation frequency of the
thicker film is modulated over the
oscillation frequency of the thinner film.
As in the case of single layer, the
thicknesses of the thin layer and the
thick layer are calculated from the
angular distance between two successive maxima/minima of high frequency oscillation and
low frequency oscillation, respectively. This is illustrated in Fig. 2.11, which shows
calculated reflectivity from two bi-layer structures, i.e., Mo (100 )-Si (100 ) and Mo (400
Fig. 2.11 Calculated reflectivity of a
Mo-Si bi-layer on Si-substrate at
wavelength =1.54 .
(a) Mo (100 )-Si (100 ) and
(b) Mo (400 )-Si(50).
0 1 2 3 4
10
-5
10
-3
10
-1
10
1
10
3
Mo (100)-Si (100) bilayer
Mo (400)-Si (50) bilayer
R
e
f
l
e
c
t
i
v
i
t
y
(theta)
Chapter-2
38
)-Si(50), at wavelength =1.54 . In the case of the Mo (100 )-Si (100 ) bi-layer
structure, the low frequency oscillations correspond to 100 and the high frequency ones
correspond to the total film thickness 200 . However, for the Mo (400 )-Si(50 ) bi-layer
structure, the high frequency oscillation due to the 400 Mo layer gets modulated over the
low frequency oscillation due to the thinner Si layer.

2.2.6 Reflectivity from a Multilayer Structure

The Fresnel formalism for reflection from a single interface (Eq
n
2.10) can be extended
to ML structures. A schematic of a periodic ML structure built of two different materials
A and B alternately, with M layers, including vacuum, on a substrate of infinite thickness,
is shown in Fig. 2.12. Let t
A
and t
B
be the thickness of the layers of A and B, respectively.
The period of the multilayer, d= t
A
+ t
B
, with N periods (N=M/2). Let n
A
and n
B
be the
refractive index of material A and B, respectively. These materials are homogeneous in
the x-y plane, and continuous, with sharp interfaces. The density variation occurs only
x
z

E
1
E
2
E
3
E
1
R
E
2
R
E
3
R
t
A
t
B
Layer
pair
E
J
E
J+1
E
J
R
E
J+1
R
E
M
1 (Vacuum)
J
J+1
M-2
M-1
M (Substrate)
2 (Material A)
3 (Material B)
Fig.2.12 Schematic diagram of incident, reflected, and transmitted electric
fields inside the periodic multilayers of two different materials A and B. The
thickness of the layers of A and B is t
A
and t
B
, respectively.
Chapter-2
39
along the z-direction. The electric field E is incident at an angle with respect to sample
surface and undergoes reflection and transmission at the different boundaries inside
multilayer structure. There are two theories to calculate the reflectivity from such a ML
structure, namely the kinematical theory (single scattering approach), also called the first-
order Born approximation, and more exact the dynamical theory (multiple scattering
approach). These are described below.

2.2.6.1 Kinematical Theory

The kinematical theory of x-ray scattering based on the assumption that an x-ray photon,
after being scattered by an electron, cannot be scattered again by another electron. So, only
one scattering can take place of a single ray. Thus, scattering of a ray by more than one
electron is neglected. This neglect leads to the complex reflectivity being the Fourier
transform of the electron density
8
. The kinematical approach is valid in the so-called weak-
scattering regime, i.e., when the cross-section of scattering is small, and multiple-scattering
effects may be neglected
9, 10
. Intuitively, this assumption is more likely to be fulfilled for
thinner layers.
In the case of a multilayer, another type of kinematical approximation can be
introduced. In this case, one neglects the scattering of a ray by more than one interface, but
does consider multiple scattering by centers (electrons) within each layer. Thus, both
absorption and refraction within a particular layer are taken into account
11, 12
. This
approximation is also called the single reflection approximation (SRA). Similar to the true
kinematical approach, the SRA is valid if the absolute values of the Fresnel reflection
coefficients are very small. The SRA allows a clearer treatment of scattering, from which
general conclusions can be drawn more easily. This theory is applicable if the reflectivity at
each boundary, and hence the total reflectivity of the structure, is small. This means that SRA
breaks down in the region of the critical angle and in the region of the Bragg peak. Using the
SRA, the reflectivity from a periodic multilayer structure can be written as
13

2
2 2
2
2 1 2 2 2 2 2
0
) (
1 ) (
1 ) )( 1 (
N
B A BS
B A
B
N
B A A B A AB A
P P r
P P
P P P P P P r r
R +
+ +
=

(2.17 (a))
where ) exp(
A
A
k A
t ik P = and ) exp(
B
B
k
t ik P =
B
are phase factors in the layer A and B,
respectively. Fresnel reflection relations hold, i.e.,
BA AB
r r = , and r
oA
and r
BS
are the Fresnel
coefficients of the vacuum/layer A interface and the layer B/substrate interface, respectively.
Chapter-2
40
A maximum of the second term on the right hand side of this eq
n
. is obtained if
2
) (
B A
P P =1.
Substituting for P
A
and P
B
and writing 1 on the RHS in the exponential form and equating,
one gets m t k t k
B
B
z A
A
z
2 = + , where m is an integer. Let the averaged z-component
momentum vector
d
t k t k
k
B
B
z A
A
z
z
+
= make an angle
t
with the internal surface. The
condition for maximum reflectivity is

m dn = sin 2 m d
c m
=
2 2
sin sin 2 (2.17 (b))

This is the modified Braggs law. Therefore, optical reflection from a periodic multilayer can
be considered as diffraction from a one-dimensional crystal. The Bragg eq.
n
(2.17 (b)) is
corrected by the refraction of x-rays in an averaged medium that replaces the actual
multilayer structure. This equation can also be written in the simplified form, if the angle of
incidence is sufficiently small. Using the approximations, sin and
c
2 ,
we get
2
2 2
2
2 |
.
|
\
|
+ =
d
m
m

(2.17 (c))
Now, the angular distance between two successive Bragg peaks gives the periodicity of the
multilayer, i.e.,
d 2
= . In the multilayer structure, the structure factor or the form factor

corresponds to the amplitude of scattered radiation from one period of the structure, where
each period might consist of several arbitrary layers. The amplitude of the Bragg peaks is
influenced by the individual thicknesses of the period, i.e., t
A
and t
B
. The structure factor of
the multilayer gives that the m
th
Bragg peak vanishes if the layer thicknesses t
A
and t
B
obey
the following relation
14

= =
+
=
p
d
t
p
t t
t
p
m
B
B A
B
(2.17 (d))
where t
B
is the thickness of high-Z material. The factor is the ratio of thickness of the
high-Z material to the total period thickness, and p is an integer. For example, if =0.5, 2
nd

order Bragg peak and its multiples will vanish. Now, let us consider the first and the second
terms in the eq.
n
(2.17 (a)). Invoking the maximum condition, the angular spacing between
satellite peaks (Kiessig oscillations) is inversely proportional to the total thickness, Nd, of the
Chapter-2
41
multilayer structure, and is given by
Nd 2
= . These Kiessig oscillations arise due to

interference of the waves reflected from the vacuum/top layer and bottom layer/substrate
interfaces. If multilayer structure has N periods, then, N-2 Kiessig oscillations are observed
between two successive Bragg peaks.

2.2.6.2 Dynamical Theory

The dynamical theory takes into account the multiple scatterings at each boundary. Hence,
it gives exact values near the critical angle and the Bragg angles, where the reflectivity is
maximum in period structures. The dynamical calculation of reflectivity was first formulated
by Abeles
15
in 1950 using the matrix method. Equivalent to this formalism is a recursive
approach first described by Parratt
16
in 1954, which is the most cited work in the field of x-
ray reflectometry. This formalism is an exact solution of the Helmholtz wave equation for N
discrete homogeneous layers. Parratts formalism takes into account the multiple reflection,
absorption, and refraction effects. Using this recursive method, the reflectivity of a multilayer
structure is computed as follows.
As shown in Fig. 2.12, the Fresnel reflection coefficient for the interface between j
and j+1 layers for s-polarized reflection can be written as

z j z j
z j z j
j
R
j
j j
q q
q q
E
E
f
, 1 ,
, 1 ,
1 ,
+
+
+
= = (2.18(a))
where,.
j
E and
R
j
E are the amplitude of the electric vector of the incident and reflected
waves at the interfaces j and j+1 and in medium j. The normal component of wave vector in
the jth layer is ( )
2
1
2 2
,
cos
2
=
j z j
n q . is the wavelength of the incident x-ray and
j
n is
the refractive index of the j
th
layer; this is discussed more fully in Chapter 6. Taking the
boundary condition that the tangential component of electric field as continuous, the
recursion relation for the reflected amplitude can be written according to Parratts formalism
for a continuous medium as

|
|
.
|
\
|
+
+
=
+ + +
+ + +
+
1 , 2 , 1
1 , 2 , 1
2
1 ,
1
j j j j
j j j j
j j j
f r
f r
b r (2.18(b))
Chapter-2
42
where ( )
j z j j
t iq b
,
exp = is the amplitude factor for half the perpendicular distance
j
t , the
thickness of j
th
layer. For a multilayer system with N layer pairs on a semi-infinite substrate,
the recursion relation starts from the bottom
layer with the assumption that 0
1 ,
=
+ N N
r , as
the thickness of this medium is infinity. One
also notes that b
1
=1 and, therefore,
1
1
2 , 1
E
E
r
R
= . If
R
I and
o
I are respectively the
reflected and incident intensities, then the
final reflectivity is

2
2 , 1
r
I
I
R
o
R
= = (2.18(c))
Similarly, for p-polarization, the Fresnel
reflection coefficient in Eq.
n
2.18 (a) is to
be modified as

z j j z j j
z j j z j j
j
P R
j
j j
q n q n
q n q n
E
E
f
, 1
2
,
2
1
, 1
2
,
2
1
) (
1 ,
+ +
+ +
+
= = (2.18(d))
In this thesis, all the simulations and data fitting has been done based on Parratts
formalism. The simulated reflectivity patterns of Mo/Si multilayers of periodicity 100 for
different values of and N are shown in Fig.2.13, with calculations done for the wavelength
54 . 1 = . In Fig. 2.13(a), there are three Kiessig oscillations (N-2) between the two
successive Bragg peaks. In this case, since =0.4, 1/ is not an integer. Therefore, no
preferential attenuation of Bragg peak intensity occurs. In Fig. 2.13 (b), the 2
nd
order Bragg
peak and its multiples are suppressed because, in this case, 1/ is an integer (Eq.
n
2.17 (e)).
Comparing (b) and (c) in Fig. 2.13, we see that is fixed but N changes. Therefore, in (b),
there are eight Kiessig oscillation between successive Bragg peaks, and the 2
nd
Bragg peak
and its multiples are suppressed.

Fig. 2.13 The simulated reflectivity of
Mo/Si multilayers with periodicity
100 at 54 . 1 = .
(a) N=5 and =0.4
(b) N=5 and =0.5
and (c) N=10 and =0.5
0 1 2 3 4 5
10
-8
10
-6
10
-4
10
-2
10
0
10
2
10
4
10
6
(a)
(b)
R
e
f
l
e
c
t
i
v
i
t
y
(theta)
(c)
Chapter-2
43
2.2.7 Imperfect Boundaries
2.2.7.1 Reduction of Reflectivity

The reflected amplitude in eq.
n
(2.18(b)) is for the ideal case of smooth interfaces, where
the density variation is abrupt at the interfaces. But in real system, the interfaces are not
perfectly smooth and have some finite width, caused by some imperfection at the interfaces
due to roughness, interdiffusion, and chemical reactivity. Contributions from different depths
within a boundary will add amplitudes with different phases, resulting in a reduced
reflectivity. Therefore, the reflectivity profile calculated for perfectly smooth interfaces
disagrees with measured the reflectivity profile. Thus, for good agreement between the
measured and theoretical reflectivity profiles, an appropriate model has to be considered to
take into account the imperfect boundaries. In a real ML system, instead of abrupt change of
density at the boundaries between two media, one has to consider an appropriate model of
gradual variation in density profile at the interfaces. A typical density variation at the
boundaries between two different materials for ideal and real system is shown in Fig. 2.14, as
function of depth in the ML system. The interface boundaries can be modeled by a series of
thin films with very small steps of density from one film to the next
17
. The Fourier transfer
method is a more elegant method and often permits one to obtain an analytical result. At the
interfaces, the variation of density, (z), can be taken have the form of an error function;
corresponding gradient of (z) will be a Gaussian distribution of width . Such a Gaussian
distribution is always obtained when a large number of different periodic amplitudes with
random phases are added
2
. The damping of reflectivity due to imperfection at the boundaries
was taken into account by multiplying the final reflectivity in eq.
n
(2.18 (c)) by a static
Debye-Waller factor, viz., exp (-q
2
2
), where is the root mean square value of interface
a
b
(z)
z
(z)
z
Fig. 2.14 Schematics of density variation at the interfaces (a) For ideal
case where density varies as step function at interfaces and (b) The real system
where density varies gradually at the interfaces.
Chapter-2
44
roughness. This term is analogous to the Debye-Waller factor used for thermal disorder effect
of crystal lattice vibration in conventional x-ray diffraction profiles
18
. This model is
applicable to ML systems in which all the interfaces have identical interfacial width. If the
interfaces have different widths, the best result is obtained by multiplying the Fresnel
reflection co-efficient in eq.
n
2.18 (a) by a factor
|
|
.
|
\
|
2
exp
2 2
j
q
. In this model, the
momentum transfer vector q is considered constant within the transition into the material.
But, in fact, the changing refractive index changes the propagation angle and the value of q
changes within the transition layer from a value q
j
in the j
th
layer to q
j+1
in the (j+1)
th
layer.
The change in the scattering vector at the interfaces is incorporated in a model given by
Nevot and Croce
19
. Replacing q
2
in the damping factor by the geometric average q
j
q
j+1
for
the boundary between j
th
and (j+1)
th
layer, gives a good approximation. Therefore, according
to Nevot and Croce, the reduction of Fresnel reflection co-efficient from a rough interface
with roughness between the j
th
and the (j+1)
th
layers is given by

|
|
.
|
\
|
=
+
+ +
2
exp
~
2
, 1 ,
1 , 1 ,
j z j z j
j j j j
q q
r r

(2.19)
The interface imperfection obtained from specular reflectivity gives information about the
surface roughness across the depth of the ML (i.e., not lateral). In the specular case, the
scattering wave vector transfer is normal to the sample surface and, hence, only the variation
of the electron density along the depth of the ML is obtained, and is laterally averaged over
the coherence area of the incident beam. The lateral surface roughness information is
obtained from diffuse scattering
20
.

2.2.7.2 Different Types of Roughness

In the case of a single layer film, the interface roughness scales as the function of film
thickness
21
. But, in ML structures, interfaces roughness may or may not be same at all the
interfaces because of the different growth modes of different materials. Therefore, the
interfacial roughness in a multilayer structure can be classified into four different types
13
, as
shown in Fig.2.15.

Chapter-2
45

(a) Uncorrelated type: In this case, no replication occurs towards the free surface. The
roughness profiles of different interfaces are independent.
(b) Cumulative type: The roughness increases towards the free surface, indicating
roughening during growth. The increasing roughness towards the free surface can be
accounted for by the relation
( ) ( ) j N
s j
+ =
2 2 2
(2.20)
where
s
is the initial roughness, is roughness replication parameter, N is the
total number of interfaces, and j decreases from the bottom to top the layer.
(c) Partially correlated type: The roughness of a layer is partially transmitted towards
free surfaces.
(d) Correlated type: All the interfaces have identical roughness

2.2.7.3 Effect of Roughness on Reflectivity

(i) From a Single Surface

Fig. 2.16 shows the calculated profile as a function of the momentum transfer vector
normal to the sample surface, for silicon surfaces with different surface roughness. The
reflectivity is calculated at the Cu-K
wavelength (=1.542 ). Roughness being zero means

an ideal case with a sharp boundary. As the roughness increases, the reflectivity decreases.
The effect of roughness is greater at higher values of the momentum transfer vector q
z
(eq.
n

2.19).
Fig.2.15 Schematic model of four different types of interface roughness in
multilayer structures: (a) uncorrelated, (b) cumulative, (c) partially correlated, and
(d) correlated
a b c
d
Chapter-2
46

(ii) From a Thin Film (A Single Layer on a Substrate)

Let us consider a thin film deposited on a substrate. The amplitude of interference
oscillations in x-rays reflected from it depends on the electron density contrast between the
substrate and the thin film. Let
f
and
s
be the roughness of the vacuum/film interface and
the film/substrate interfaces, respectively. If surface roughness is to be taken into account,
then, the Fresnel reflection co-efficients r
12
and r
23
(eq.
n
(2.15)) will be multiplied by the
Debye-Waller factor given in eq.
n
(2.19). Fig. 2.17 shows the effect of both the roughness
factors (
f
and
s
) on the reflectivity of a 200 Mo film on Si substrate. The reflectivity is
calculated at Cu-K
wavelength. The effect of different combinations of

f
and
s
on the
reflectivity profile is as follows.
Fig.2.16 Calculated x-ray reflectivity at Cu-K
wavelength (=1.542 ) of Si
surfaces of density 2.33 gm/cm
3
, for different values of surface roughness, .
The inset shows the scattering length density profile across film depth.
Chapter-2
47

Case1: When
f
=0 and
s
=0 : this corresponds to the ideal case.
Case 2: When
f
=5 and
s
=0 : as the top interface roughness increases, the reflectivity
decreases as function of q
z
.
Case 3: When
f
=0 and
s
=5 : as substrate interface roughness increases, the envelope is
not affected, but the amplitude of oscillation is reduced.
Case 4: When
f
=5 and
s
=5 : in this case, reduction in reflectivity is greater, but the
amplitude of oscillations is still significant at higher q
z
. This is because the reduction in the
Fresnel reflected amplitude from the two interfaces is the same and, hence. these two waves
preserve their coherence after interference. The effect of roughness on the reflectivity of
multilayer structures is discussed further in Chapter 5. Let us consider the procedure
employed in this thesis for the analysis of x-ray reflectivity data.

2.2.8 Method of Data Analysis

The measured reflectivity as function of incidence angle or momentum transfer vector
contains information in reciprocal space. Therefore, it is necessary to derive the real space
Fig.2.17 Calculated x-ray reflectivity at Cu K
wavelength (=1.542 ) for a

200 Mo film on Si substrate, for different combinations of substrate roughness
s
and film roughness
f
. The inset shows the in-depth scattering length density profile.
0.0 0.1 0.2 0.3
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
10
1
0 100 200
0.00
3.50x10
-5
7.00x10
-5
S
c
a
t
t
e
r
i
n
g

l
e
n
g
t
h

d
e
n
s
i
t
y

(
r
h
o
/
2
)
In depth ()

s
=0,
f
=0

s
=0,
f
=5

s
=5,
f
=0

s
=5,
f
=5
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
Chapter-2
48
information of the physical system. In principle, this can be done by inverse Fourier
transformation. But the problem in scattering experiments is that phase information is lost, as
only intensity is measured
13
. This is analogous to the phase problem in crystallography
22
.
Thus, real space information is generally obtained using model-dependent methods. Sinha et
al.
23
have proposed a model-independent method to obtain density profiles using anomalous
reflectivity. In this method, in order to get phase information, reflectivity is measured at two
wavelengths; one is at the absorption edge, and the other is far from absorption edges of the
constituent elements.
The most commonly used method for analyzing the XRR data is the modeling of the
electron density profile as a function of depth in the material. In this method, reflectivity is
numerically calculated using a standard method and then compared to the measured data. In
this work, the reflectivity profile is calculated numerically using a computer program based
on Parratts recursive formalism (discussed earlier). The initial gausses needed for this
program are thickness, optical constants, and rms roughness. These parameters are varied and
fitted to the measured data, using a non-linear square fitting procedure by minimizing the
value of
2
,

( ) ( ) | |

=
j j
j
c
j
m
s
R R
2
2
2

(2.21)
where, ( )
j
m
R and ( )
j
c
R are the measured and calculated reflectivity, respectively, at an
angle
j
, and s
j
is the statistical error associated with measurement of the data at
j
.
However, the initial guess of the layer thickness of deposited materials can also be
obtained by applying Fourier Transformation (FT) method
24
on the experimental x-ray
reflectivity data as follows. The reflectivity from a multilayer structure can be calculated by
the Fourier method
17
, which allows an analytical solution, based on the transformation of the
depth-dependent electron density. In this method, reflectivity can be written in the Born
approximation, which is valid at wave vector q
z
larger than the critical value q
c
, as
( ) ( )
2
2
2
exp
16
+

= dz z iq
dz
d
q
q R
z
z
z

(2.22)
Chapter-2
49
where the reflectivity ( )
z
q R is the FT of the electron density gradient
dz
d
. Therefore, Fourier
inversion gives rise to the auto-correlation function (ACF) of the electron density gradient,
which is defined as
25
( ) { } ( )
2
4
exp ) (

+

=
z z z z
dq z iq q q R z ACF (2.23)
Eq.
n
2.23 gives the position of the interfaces directly. For this, the measured reflectivity data
should be multiplied by
4
z
q and the momentum transfer vector in vacuum should be
converted to the medium by the relation
2 2 '
c z z
q q q = . Then, prior to the FT, the data range
c z
q q is removed, since the Born approximation is not valid in that range. In the case of
ideal multilayer structures, the peaks in FT are delta functions, because of abrupt interfaces.
But, in real ML structures, broadening of the peaks occurs due to the finite interfacial width.

2.2.8.1 Instrumental Resolution

Instrumental resolution depends on the
incident and exit slit widths, the
monochromator, the detector area, and the
radius of the goniometer used. Therefore, in
reflectivity experiments, the measured
scattered intensity is collected over a range
of angles around the specular direction,
which is hence smeared due to finite
instrumental line width. Therefore, there is
always a discrepancy between measured
and calculated spectra and one has to
correct this discrepancy during data fitting.
Generally, this discrepancy is removed by
convoluting the calculated spectrum with
the known instrumental, width using a Gaussian instrumental resolution function
26
( ) ( ) ( )
=
"
" "
.
z
q
z z z z
q S q q R q R (2.24)
Fig. 2.18 Calculated X-ray reflectivity
spectra of [Mo (25)/Si 65 ]
5
ML on Si-
substrate at Cu K
wavelength showing
effect of instrumental resolution with a
Gaussian width of 0.003
-1
in q
z
.
0.0 0.1 0.2 0.3 0.4
10
-4
10
-2
10
0
With out instrumental resolution
Resolution of 0.003
-1
in q
z

R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
Chapter-2
50
where ( )
z
q R is the calculated reflectivity and ( )
z
q S is the instrumental resolution function.
Fig. 2.18 shows the calculated reflectivity from an ideal [Mo (25)/Si 65 ]
5
multilayer
structure with and without instrumental resolution

2.2.9 Experimental Set-up

Hard x-ray reflectivity measurements were carried out on a reflectometer developed in
house
27
on a sealed tube, with a Cu target (=0.154 ), as shown in Fig. 2.19. Two slits of
width 0.1 mm and 0.05 mm were used at the source for collimation, to obtain a near-parallel
beam. To control axial divergence, a Soller slit with of 0.4
0
divergence was used. A Pt/C
multilayer monochromator was used at an angle of incidence of 1.2
0
to
monochromatize the incident beam. The final beam size in the axial direction was 5 mm and
the beam divergence 0.025
0
. A scintillation counter was used to detect the reflected x-ray
intensity. The measurements were performed with a step size of 0.01
0
in the theta axis. The
experimental conditions for soft x-ray reflectivity measurements are discussed in Chapter 6.

2.3 Electrical Conduction in Thin Films
2.3.1 General Considerations

Electrical conduction in thin films has long been of practical importance and of theoretical
interest as well. In situ resistivity measurement has been used to identify the discontinuous-
to-continuous transition in thin films, for analyzing variations of electrical properties of thin
films with thickness
28
, and to obtain a clearer understanding of the nature of film growth
29
.
Regardless of the class of material involved, its physical state, and whether it is in bulk or
film form, an electric current of density J (amps/cm
2
) is said to flow when a concentration of
carriers n (number/cm
3
) with charge q moves with velocity v (cm/sec) past a given reference
Fig. 2.19 A Schematics of hard x-ray reflectivity experimental
t
X-ray
Tube
Divergence
Slit
Pt/C multilayer
monochromator
Soller Slits
Sample
Detector
Chapter-2
51
plane in response to an applied electric field E (V/cm). A simple relation expresses the
magnitude of the current flow
J=nqv=nq E (2.25 (a))
=e/m. is the charge carrier mobility. From Ohms law (J=E), the conductivity or
resistivity is given by
=1/= nq (2.25 (b))

Much of what is already known about bulk conduction provides a good basis for
understanding thin film behavior. But, there are important differences that give thin films
unique characteristics, and these are as follows: (i) size effects or phenomena that arise from
the small physical dimensions involved, (ii) method of film preparation, (iii) degree of film
continuity, (iv) existence of high electric field conduction phenomena, and (v) high chemical
reactivity. In virtually all cases, thin metal films are more resistive, while insulating films are
more conductive, than their respective bulk counterparts.

2.3.2 Measurement of Thin Film Resistivity

A very common way to report values of thin film resistivity is in terms of sheet resistance,
with the unit ohms per square. To understand sheet resistance and the unit involved,
consider a thin film of length l, width w, and
thickness d, as shown in Fig. 2.20. If the film
resistivity is , then the resistance of film
measured in a direction parallel to the film
surface is given
30
by R=l/wd. In the special
case of a square film (l=w),
R=R
s
=/d ohm/, (2.26)
where R
s
is independent of film dimensions
other than thickness. Any square, irrespective
of size, would have same sheet resistance. The commonly used techniques for the electrical
resistivity measurement of thin films are the four-point probe method and the van der Pauw
method, which are described below.

l
d
current
w
Fig. 2.20 Thin-film conductor with
length l, width w, and thickness d
Chapter-2
52
2.3.2.1 Four Probe-Point Method

The most convenient way to measure the sheet resistance of a film is to lightly press a
four-point metal-tip probe assembly into the
surface is shown in Fig.2.21. A current is
passed through the two outer terminals, and
the resulting voltage is measured across the
two inner terminals. The sheet resistance is
given by
27
,
R
s
=KV/I (2.27)
where K is a constant that depends on the
configuration and spacing of the contacts. If
the film is large in extent compared with the
probe assembly, and the probe spacings are
large compared to the film thickness, then K=/ln2. Otherwise, appropriate correction needs
to be applied
27
. The four-point probe only uses a regular specimen, whereas the van der Pauw
method is applied for any configuration of sample, and is discussed below.

2.3.2.2 The van der Pauw Method

van der Pauw
31
has devised a general method to determine the resistivity of a film having
an arbitrary in shape, as shown in Fig. 2.22. The validity of the van der Pauw method
requires that the sample be flat, homogeneous, and isotropic, and have line electrodes on the
periphery, projecting to point contacts on the
surface, or true point contacts on the surface.
As shown in Fig. 2.22, a current I
AB
flows
between contacts A and B, and a voltage V
CD
is
measured between contacts C and D. Then, the
resistance R
AB, CD
=V
CD
/I
AB
. Similarly, by passing
current through B and C, and measuring the
potential across D and A, the resistance R
BC,
DA
=V
DA
/I
BC
is obtained. The resistivity is calculated as follows:
f
R R
d
DA BC CD AB
|
|
.
|
\
| +
=
2 2 ln
, ,
(2.28)
V
I
4 point probes
Film surface
Voltage
Current
Fig. 2.21 The Four probe
method for measuring resistivity of
fil
V
CD
I
AB
C
B
A
D
Fig.2.22 Van der Pauw method
for measuring resistivity of arbitrarily
shaped film
Chapter-2
53
where f is van der Pauw correction factor, whose value depends on the resistivity in-
homogeneity in the sample
31
. David et al.
32
have given the detailed aspects of finite-contact-
size corrections.

2.4 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) is based on the photoelectron effect, and was
developed by Siegbahn
33
in the mid-1960s. When an x-ray photon is incident on materials, it
ionizes the atom and ejects a core level electron. The kinetic energy (KE) distribution of the
emitted photoelectrons can be measured using an appropriate electron energy analyzer and a
photoelectron spectrum can thus be recorded.

The KE of the ejected electron can be written as
34
=
B k
E h E (2.29)
where, E
k
is the KE of the ejected electron, h the energy of the incident x-ray photon, E
B
the
binding energy (BE) of the electron in the solid, and the work function (whose value
depends on both the sample and the spectrometer). The ejected electron KE is measured by
the spectrometer. The basic XPS phenomenon is shown graphically in Fig. 2.23 (a). For each
and every element, there is a characteristic BE associated with each core atomic orbital, i.e.,
each element gives rise to a characteristic set of peaks in the photoelectron spectrum, at
kinetic energies determined by the photon energy and the respective binding energies. The
Initial state
Vacuum level

Core level
h
Final state
Vacuum level

Core level
h
BE
KE
(a)
(b)
Fig. 2.23 The x-ray photoelectron spectroscopy process and instrumentation.
(a) the basic phenomenon of XPS and (b) schematic of instrumentation
Chapter-2
54
presence of peaks at particular energies therefore indicates the presence of a specific element
in the sample under study - furthermore, the intensity of the peaks is related to the
concentration of the element within the sampled region. The area under the core-level peaks
in the spectrum provides a quantitative estimate of the composition. Therefore, the technique
provides a quantitative analysis of the surface composition. Thus, XPS is a technique widely
used to investigate the chemical composition of surfaces.
The basic requirements for x-ray photoelectron measurements are shown in Fig. 2. 23 (b).
These include (i) an x-ray source of fixed-energy, (ii) an electron energy analyzer (which can
disperse the emitted electrons according to their KE, and thereby measure the flux of emitted
electrons of a particular energy) and (iii) a high vacuum environment (to enable the emitted
photoelectrons to be analyzed without interference from gas phase collisions).

2.4.1 X-ray Photoelectron Spectral Interpretation

In illustrate the interpretation of XPS data, a typical wide-scan spectrum of a clean silver
surface using Mg-K
radiation is shown in Fig. 2.24 (Reproduced from Ref. 34). The peaks
in Fig. 2.24 can be grouped into three basic types: peaks due to photoemission from core
levels and valence levels, and peaks due to x-ray excited Auger emission. These are
discussed below.

Fig. 2.24 A typical wide scan of x-ray photoelectron spectrum of silver materials
with Mg K
radiation.
Chapter-2
55
Core Levels:
Mg K
radiation (1253.6 eV) is energetic enough to probe the core levels of silver up to the
3s shell. Fig. 2.24 clearly shows that the core levels have variable intensities and widths, and
that non-s level peaks are doublets.
Doublets: The doublets arise due to spin-orbit coupling. The magnitude of this energy
separation is proportional to the spin-orbit coupling constant, which depends on the
expectation value
3
1
r
for the particular orbital. The separation can be many electron
volts. Therefore, this energy separation is expected to increase as Z increases for a given sub-
shell (constant n, l), or to increase as l decreases for constant n (for example, splitting of 3p>
3d, as shown in Fig. 2.24). The relative intensity of the doublet peaks is given by the ratio of
their respective degeneracies, (2j+1).
Relative intensities: The basic parameter which governs the relative intensities of core-
level peaks is the atomic photoemission cross-section, . Variation in the exciting energy
also affects values of .
Peak widths: The peak width, defined as the full width at half-maximum, E , is a
convolution of several contributions:
( )
2
1
2 2 2
a p n
E E E E + + = (2.30)
where
n
E is the natural or inherent width of the core level,
p
E is the width of the photon
source (X-ray line), and
a
E is the analyzer resolution. The inherent line width of a core
level is a direct reflection of the uncertainty in the lifetime of the ion state remaining after
photoemission. Since the lifetime of 3s is smaller than 3d, the 3s peak is wider than the 3d
peak, etc.. It should be noted that the line widths of the principal light element core levels
(1s, 2p) increases systematically with increasing atomic number.
Core-level chemical shifts: The non-equivalent atoms of the same element in a solid gave
rise to core-level peaks with measurably different BE. This BE difference is known as
chemical shift. Non-equivalence of atoms can arise in several ways: difference in formal
oxidation state, difference in molecular environment, difference in lattice site, and so on.
Using the charge potential model
33
, the physical basis of the chemical shift effect can be
illustrated by a relatively simple model as

+ + =
j i ij
i
i i i
r
q
kq E E
0
(2.34)
Chapter-2
56
where E
i
is the BE of a particular core level on atom I,
0
i
E is an energy reference, q
i
is the
charge on atom I, with the final term of eq
n
. (2.34) summing the potential at atom I due to
point charge on surrounding atoms j. It is to be noted that the BE increases with an increase
in the formal oxidation number. In situations where the formal oxidation states are same, the
general rule is that the core-level BE of the central atom increases as the electro-negativity of
attached atoms or groups increases.

Valence Levels

Valence levels are those occupied by electrons of low binding energy (say 0-20 eV),
which are involved in de-localized, or bonding orbitals. The spectrum in this region consists
of many closely spaced levels giving rise to a band structure.

Auger Series
The incident X-ray excites the core electron, and the outer electron falls to the core level,
giving rise to a photon. That photon again excites secondary electrons, which are known as
Auger electrons. Auger electrons have low KE and come out only from the top surfaces of
solids (5-10 ). Auger electrons can give rise to dominant peaks in the photoemission
spectrum on the low KE side, as shown in Fig. 2.24. A series of peaks are observed on a
background which generally increases to low KE, but which also shows step-like increases
on the low KE side of each significant peak, known as Auger peaks. It is to be noted that the
background step to low KE of the photoelectron peaks is due to inelastic photoemission.

2.5 Glancing Incidence X-ray Fluorescence

In GIXRF, x-ray fluorescence intensity is measured as function of incidence angle. The
GIXRF provides information about the depth distribution of fluorescent atoms in a
quantitative way. By changing the angle of incidence, the depth sensitivity inside a thin-
layered material can be analyzed on the nanometer scale, as demonstrated first by Becker et
al
35
. Since then, the GIXRF technique has been used for a variety of applications
36
such as
the determination of the depth profile of impurity atoms, layers on a substrate, etc. In this
study, GIXRF measurements were performed on a total x-ray fluorescence spectrometer
developed in house
37
. The fluorescence data were taken using a Peltier- cooled solid-state
detector having an energy resolution of 250 eV at 5.9 keV. A well-collimated primary beam
from a line-focus Cu x-ray tube was used as the excitation source.
Chapter-2
57

2.6 References

1.
D. Attwood, Soft X-rays and Extreme Ultraviolet Radiation: Principles and
Applications, Cambridge University Press (2000).

2.
E. Spiller Soft X-ray Optics SPIE, Washington (1994).

3.
M. Jergel, V. Holy, E. Majkova, S. Luby, R. Senderak, J. Phys. D.: Appl. Phys. 28,
A241 (1995).

4.
A. Compton and S. K. Allison, X-rays in Theory and Experiments (Van Nostrand,
New York, 1935).

5.
M. Tolan X-ray Scattering from Soft-Matter Thin Films (Springer, New York,
1999).

6.
M. Born and E. Wolf, Principle of Optics (Pergamon, New York, 1980).

7.
H. Kiessig, Ann. Phys. 10, 715 and 769 (1931).

8.
J.M. Cowley, Diffraction Physics (North-Holland, Amsterdam, 1975).

9.
L.V. Azaroff, R. Kaplow, N. Kato, R. J. Weiss, A.J.C. Wilson, R.A. Young, X-ray
Diffraction (McGraw-Hill, New York, (1974).

10.
S.K. Sinha, E.B. Sirota, S. Garoft and H.B. Stanley, Phys. Rev. B, 38, 2297 (1988).

11.
M. Born and E. Wolf; Principles of Optics, Pergamon, Oxford (1993); I.W.
Hamley and J.S. Pedersen, J. Appl. Cryst. 27, 29 (1994).

12.
I.W. Hamley and J.S. Pedersen, J. Appl. Crst. 27, 29 (1994).

13.
V. Holy, U. Pietsch and T. Baumbach High-Resolution x-ray Scattering From Thin
Films and Multilayers (Springer, Germany, 1999).

14.
E. Ziegler, Opt. Eng., 34, 445 (1995).

15.
F. Abeles, Ann. Physique (Paris) 5, 596 (1950).

16.

17.
M. Tolan X-ray Scattering From Soft-Matter Thin Films (Springer, Berlin, 1999).

18.
C. Giacovazzo, H.L. Monaco, S.D. Viterbo, Fundamentals of Crystallography
(Oxford Science Publications, UK, 1991).

19.
L. Nevot and P. Croce, Revue. Phys. Appl. 15, 761 (1980).

20.
V. Holy and T. Baumbach, Phys. Rev. B. 49, 10668 (1994).

21.
M. Oehring, The Materials Science of Thin Films (Academic Press Inc., New
York, 1992).

Chapter-2
58
22.
B.E. Warren, X-ray Diffraction (Addison-Wesley, MA, 1969).

23.
S.K. Sinha, M.K. Sanyal, B.L. Carvalho, M. Rafailovich, J. Sokolov, X. Zhao, W.
Zho in Resonant Anomalous X-ray Scattering: Theory and Applications, edited by
G. Materlik, C.J. Sark and K. Fischer, (North-Holland, Netherlands, 1994) p. 421.

24.
F. Bridou and B. Pardo, J. Phys. III, 4, 1523 (1994).

25.
O.H. Seeck, I.D. Kaendler, M. Tolan, K. Shin, M.H. Rafailovich, J. Sokolov, R. Kolb,
Appl. Phys. Lett. 76, 2713 (2000).

26.
W.H. Press, S.A. Teukolsky, W.T. Vellerling, B.P. Flannery, Numerical Recipes in
FORTRAN (Cambridge University Press, 1992).

27.
S. Rai, M. K. Tiwari, G. S. Lodha, M. H. Modi, M. K. Chattopadhyay, S. Majumdar,
S. Gardelis, Z. Viskadourakis, J. Giapintzakis, R. V. Nandedkar, S. B. Roy and P.
Chaddah, Phys. Rev. B 73, 035417 (2006).

28.
M. Veldkamp, H. Zabel, Ch. Morawe, J. Appl. Phys. 83, 155 (1998).

29.
V. Korobov, M. Leibovitch and Y. Shapira, Appl. Phys. Lett. 65, 2290 (1994).

30.
L.I. Maissel, Handbook of Thin Film Technology edited by L.I. Maissel and R.
Glang (McGraw-Hill, New York, 1983) p.13-5.

31.
L.J. van der Pauw, Philips Res. Repts. 13, 1 (1958).

32.
J.M. David and MG. Buehler, Solid-state Electron. 20, 539 (1977).

33.
K. Siegbahn, Et. Al. Nova Acta Regiae Soc. Sci., Ser. IV, Vol. 20 (1967).

34.
M.P. Seah and D. Briggs in Practical Surface Analysis edited by D. Briggs and
M.P. Seah (John Wiley & Sons, New York, 1983).

35.
R.S. Becker, J.A. Golovchenko and J.R. Patel, Phys. Rev. Lett. 50, 153 (1983).

36.
R. Klockenkamper, Total Reflection X-ray Fluorescence Analysis (John Wiley and
Sons, New York, 1997).

37.
M.K. Tiwari, K.J.S. Sawhney, B. Gowri Sankar, V.K. Raghuvanshi and R.V.
Nandedkar, Spectrochemica Acta Part B, 59, 1141 (2004).

Chapter-3
59
CHAPTER-3

Design and Development of an UHV Electron Beam Evaporation System

This Chapter describes the details of an ultra high vacuum electron beam evaporation
system developed especially for the fabrication of thin films and x-ray multilayer optical
elements for a variety of applications on the Indus-1 and Indus-2 synchrotron radiation
sources. Different deposition techniques commonly used for the fabrication of x-ray optical
elements are discussed. The design, fabrication, and implementation of the e-beam
evaporation system for the fabrication of thin films and high quality of x-ray multilayers are
then discussed. Finally, the suitability of the evaporation system for the incorporation of in
situ characterization facilities is illustrated by carrying out in situ electrical measurements
on thin films at different substrate temperatures.

Chapter-3
60
3.1 Introduction

The fabrication of multilayer (ML) mirrors for the soft x-ray/extreme ultra-violet (XUV) part
of the spectrum is a challenging task. Compared to coatings for visible light, the thicknesses
of the layers required and the permitted thickness errors are about a factor of 100 smaller,
and the number of layers is typically a factor of 10 higher for the XUV region. Therefore,
tighter control of the film deposition process is required. In ML mirror coatings, layer
thickness, and the quality of surfaces and interfaces, are the parameters crucial to acceptable
performance. The coatings should be largely amorphous (or to some degree polycrystalline)
within individual layers, and reflections must conform to Braggs law for a periodicity d
equal to the thickness of one bi-layer pair, typically measured in tens of atomic monolayers.
Boundaries (interfaces) have to be sharp within 1/10 of the ML period to obtain the required
specular reflectivity. Therefore, the search for deposition technique, which produces sharp
boundaries, is a major effort in the development of coatings for XUV region.
The basic requirements for the fabrication of XUV ML mirror of high quality, by any
technique, are: (a) Precise control (angstrom level) over layer thickness and uniformity over a
large substrate area, (b) minimum contamination and stress in the layers, and (c) atomically
sharp, well-defined interface boundaries. Recent technological advances in thin film
deposition techniques have made it possible to produce MLs of the quality desired for
intended applications. Any method for the deposition of thin films can be used for the
fabrication of coatings for XUV mirrors, provided it fulfils the criteria listed above. Various
deposition techniques, such as electron beam evaporation
1
, sputtering
2-4
(magnetron and ion
beam), laser ablation
5
, and chemical vapor deposition
6
have been used to fabricate XUV
MLs. The choice of the deposition technique depends mainly on the specific goals and the
type of materials used for deposition. Electron beam evaporation and sputtering are the most
widely used techniques for the fabrication of XUV ML mirrors, as reported in the literature.
These techniques are briefly discussed in the following sections with specially emphasis on
electron beam evaporation systems of the type indigenously developed and used in the
present study. The main motivation for developing this specially designed electron beam
system is the fabrication of large-sized (20 cm 10 cm) x-ray multilayer optics of high
quality for the Indus-1 and Indus-2 synchrotron radiation sources.

Chapter-3
61
3.2 Deposition Techniques
3.2.1 Sputter Deposition

When a solid surface is bombarded with energetic particles such as accelerated gas ions
(usually inert gas), atoms on the solid surface get ejected due to collisions between the
surface atoms and the energetic ions. This phenomenon is known as sputtering
7
, shown
schematically in Fig. 3.1. The atoms ejected impinge on the substrate placed in the chamber
and condense to form a film. The energetic ions may be produced either by a glow discharge
or by a separate ion source.

Based on the method by which the discharge is produced and sustained, sputtering can be
classified into DC, RF, and magnetron (DC & RF), and ion-beam sputtering. The different
sputtering processes are described in the following sections.

3.2.1.1 DC Sputtering

This is the simplest configuration of sputter deposition, as shown in Fig. 3.2. It consists
of a pair of planar electrodes. One of the electrodes is a cold cathode and the other is the
anode (substrate). The front surface of cathode is covered with the target material to be
deposited, the back side which is water-cooled. The substrates are placed on the anode.
Argon is the most commonly used sputtering gas, at a pressure of the order of 0.1 torr. The
power supply is simply a high-voltage DC source of several kilo-Volts. To initiate the glow
discharge, a DC voltage (a few kV) is applied between the electrodes. The Ar
+
ions resulting
from the discharge are accelerated towards cathode and sputter the target material. A typical
ion current density to the cathode is ~ 1 mA/cm
2
. The sputtered atoms get deposited on the
substrate surface (placed on the anode), resulting a thin film of the target material. In the case
of insulating targets, the sputtering discharge cannot be sustained because of the immediate
Target surface
Incident ion Sputtered atom
Target atom
Fig. 3.1 Physical sputtering process
Chapter-3
62
build-up of a surface charge of positive ions on the insulator, which repels any further
positive ion bombardment.

3.2.1.2 RF Sputtering

The problem of depositing insulating materials is overcome by the RF sputtering
method. A RF voltage commonly of 13.56 MHz is applied between the electrodes. Therefore,
the target is alternately bombarded with positive ions and electrons, thus neutralizing the
space charge. The RF electric field increases the probability of collisions between the gas
molecules and secondary electrons. Therefore, a lower sputtering gas pressure and higher
deposition rates obtain than in DC sputtering.
3.2.1.3 Magnetron Sputtering

In simple diode sputtering (DC and RF) techniques, the deposition rate is low and a
high level of contamination of deposited films occurs due to the relatively high working
pressure. Moreover, the discharge envelopes the entire chamber in DC and RF sputtering,
increasing the substrate temperature substantially. To overcome these problems, the
magnetron configuration is used. A typical planar DC magnetron configuration is shown in
Ar
+
Substrate (anode)
Target
(cathode)
Argon gas
Vacuum
pump
Power
supply
Vacuum
chamber
Fig.3.2 A typical DC sputtering system. Ions strike the target (cathode), which is
the deposition source. The sputtered particles are deposited on the
substrate.(anode).
Chapter-3
63
Fig. 3.3. The electric field and magnetic fields are perpendicular to each other and are
superimposed such that glow discharge is parallel to the cathode surface. A transverse
magnetic field of a few hundred Gauss is superimposed on the glow discharge such that drift
paths for electrons form a closed loop. This electron trapping effect increases the collision
rate between the electrons and the sputtering gas molecules. This results in a lower working
gas pressure, as low as 10
-4
torr. In this magnetron configuration, the magnetic field increases
the plasma density, which leads to an increased current density at the target which, in turn,
enhances sputtering rate from the target. The sputtered particles traverse the discharge space
without collision because of the low working gas pressure, which results in effectively higher
deposition rates with low contamination of the films deposited. In Fig. 3.3, there are two
cathodes mounted 180
0
apart for deposition of two materials alternately, for the fabrication of
multilayer structures
8
. The materials to be deposited are mounted on the cathodes. A quartz
crystal is mounted at the same height as the substrate so as to monitor the thickness of the
film being deposited. The substrate is mounted on the anode with spinning arrangements, to
obtain uniform deposition. The substrate platter is rotated over the magnetron source for
alternating deposition of materials in the multilayer structure.

To vacuum
Vac. chamber
S

N
S

N
S

N
S

N
Spinning arrangement
Substrate
Quartz crystal
Target
Magnetron source
DC power supplies
& Control system
Gas flow
Substrate platter
Fig.3.3 Schematic of a twin platter magnetron assembly, showing substrate platter
arrangements generally adopted for the alternating deposition of two materials.
Chapter-3
64
3.2.2 Ion Beam Sputtering

In this technique, an ion source is physically isolated from the substrate deposited
upon. The greatest advantage of these ion sources is that, usually, the ion energy and flux can
be controlled independently and that the sample has no effect on the ion-producing discharge.
A separate ion source is used to produce a focused ion beam, which is incident on the target
material to be deposited. The ion source used for thin film deposition is generally a
Kaufmann broad-beam ion source
9
. In such a source, ions are confined magnetically to a
discharge chamber. Then ions are extracted through a multi-aperture accelerating system. To
avoid defocusing the beam due to space charge effects, electrons are mixed with the extracted
ions to neutralize the beam. The energetic ions then bombard a target made of the material to
be deposited. A typical ion beam deposition system is shown in Fig. 3.4. In this case, the
deposition yields can be varied by varying the ion beam energy and the angle of incidence on
the target. Depositions can be carried out at much lower pressures (<10
-6
Torr) by arranging
differential pumping in front of the ion source. Although the deposition rate is substantially
lower than in magnetron sputtering, enhanced control and cleanliness of the deposited
material can be achieved in ion beam sputtering. The main disadvantage is the lack of
uniformity over large surface areas, due to the directionality of the ion beam.

Fig. 3.4 A typical ion beam sputtering system
for the deposition of multilayer coatings
Si Mo
Flow controller
Ar gas
Power supply
Thickness controller
PC
Stepper motor
Ion gun
Quartz crystal
Substrate holder
Chapter-3
65
3.2.3 Electron Beam Evaporation

The basic parts in the electron beam gun are the thermionic emitter, electron optics,
the high voltage DC power supply, a permanent magnet, and the crucible containing the
evaporant. The electrons are produced by resistive heating of a wire made of a high-
temperature metal, usually tungsten or tantalum. The electrons are accelerated and focused
through the electron optics (beam former and accelerating plate). With the help of a magnetic
deflection system, the electron beam is then bent by 270 onto a water-cooled copper crucible
filled with the evaporation materials, as shown in Fig. 3.5.

On impact with the material, the kinetic energy of electrons is transformed into heat.
This corresponds to the potential of the DC supply, typically several kV. The energy of the
evaporated particles is of the order of the source temperature, which is typically in the range
of 0.2 eV - 0.5 eV. The kinetic energy of the adatoms (those adsorbed on the substrate
surface) can be enhanced through ion beam- assisted deposition
10
. The electron gun supply
Beam former
evaporant
Magnet
Accelerating plate
Crucible
Filament
heating
voltage
High
voltage
Fig. 3.5 A 270
0
bend electron beam gun showing the filament beam former,
accelerating plate, magnet, and the crucible with the evaporant material.
Chapter-3
66
voltage for the deposition of different materials ranges from 3 to 10 kV, with emission
currents ranging from a few tens of mA to a few hundred mA. All non-toxic elements and
alloys (including high melting point materials) can be deposited using this technique, as also
some selected compound materials. To maintain a constant deposition rate, a highly stable
power supply is required because vapor pressure is strongly temperature-dependent. This
technique provides high power density and, hence, control over a wide range of evaporation
rates, from low (~0.1 /min.) to very high (~10 /s) is possible. The evaporation of materials
can be carried out in a clean ultra-high vacuum system (~10
-10
Torr), which reduces
contamination of the film. Since the evaporant is contained in a water-cooled crucible, only
its surface reaches high temperatures.

Table 3.1 Advantages and disadvantages of sputtering and electron beam evaporation
techniques.

Sputtering Evaporation
Good uniformity of deposited film over
substrate requires a large cathode target

Film surfaces are bombarded by sputter gas
ions.
Film may be contaminated

Large volume of sputtered materials
required

For ML deposition, substrate holder rotates
over sputter sources.

Deposition can run unattended at high
deposition rates (15 /s).

Kinetic energy of the sputtered atoms
usually larger than ~10 eV, and may be
adjusted

Films of compound materials can be
deposited.
Good uniformity over large area

Smoother boundaries.

Film not contaminated because of
UHV conditions.
Small volume of evaporated material
required

Shutter moves in front of e-beam
sources.

Speed limited by thickness
monitoring system.

Kinetic energy of evapourated atom
smaller than ~0.5 eV and not
adjustable.

Compounds of high-Z elements
cannot be deposited, as the
stoichiometry of source material will
not be maintained in the film.
Chapter-3
67
Metallurgical reactions between the crucible and the evaporant leading to film contamination
are therefore eliminated. Under UHV conditions, the mean free path of the evaporant atoms
is very large compared to chamber dimensions. Therefore, the source-to-substrate distance
can be kept large to obtain uniform deposition over a large area substrate (~30-40 cm dia.),
assisted by substrate rotation. Thickness is controlled using a quartz crystal oscillator. To
eliminate problems associated with thermal drift in quartz crystal oscillators
11
,

thickness may
also be monitored using a sophisticated system with in situ x-ray reflectivity measurement.
The advantages and disadvantages of the sputtering and evaporation techniques are
summarized in Table 3.1.

3.3 Developed of UHV Electron Beam Evaporation System

The main motivation for the development of the e-beam system is the study of the
physics of multilayers and the fabrication of ML x-ray optical systems for various
applications, especially for the Indus-1 and Indus-2 synchrotron sources. All parameters
critical to the deposition of MLs were considered while designing the system. The UHV
chamber, the substrate holder and masking arrangements, and the load-lock system were
designed accordingly. The choice of pumping systems, and provision for incorporating
various features were made with similar considerations. These are described below, together
with a performance evaluation of the system.

3.3.1 Design of the UHV Chamber

For simplicity in machining, and to provide good mechanical strength, we have chosen
the cylindrical shape for the chamber. The total height of the chamber, made of SS 304 L, is
1200 mm and the diameter 700 mm, to accommodate large substrates. The chamber consists
of two parts, i.e., a bottom part 400 mm high and a top part 800 mm high. These two parts
are joined by a Viton O-ring placed in a groove made in the bottom part. We have chosen
UHV-compatible materials for construction. All demountable flange joints are of the
Conflat type. Oxygen-free-high-conductivity (OFHC) copper gaskets seal the flanges. The
chamber has various ports to accommodate different essential components such as electron
guns, the quartz crystal thickness monitor, the substrate holder with heating and cooling
arrangements, shutters, masking arrangements, the load-lock system, pumping systems,
residual gas analyzer, vacuum gauges, electrical and motion feed-throughs, and in situ
Chapter-3
68
resistivity measurements. The number of extra ports has been taken into account for the
incorporation of ion guns for ion assisted deposition, and the mounting of another multi-
crucible electron gun for co-deposition. The provision for other additional ports has been
made for any future requirements, such that the system would be versatile. Apart from these
ports, some symmetrical ports oriented 45
0
w.r.t the substrate enable in situ reflectivity
measurements. Another pair of symmetrical ports located in the upper part of the chamber
allows for visual adjustment of the focusing of the electron beam on the source material. The
mechanical design of the deposition system is shown in Fig.3.3. Special care has been taken
in chemical cleaning and polishing the internal surfaces to remove gases adsorbed on the
inner chamber walls during machining. The outer wall of the chamber is water-cooled
through welded copper tubing to prevent deterioration of vacuum due to degassing from the
inner walls during deposition.

3.3.2 Pumping System and Gauges

The selection of the pumping system is crucial to get a low base pressure, which is
important to obtain a tolerable contamination of the growing film deposited in a system
Fig. 3.6 The mechanical design of the UHV e-beam evaporation system. The
different accessories mounted at different ports are as follows: A-electron gun, B-
O-ring joint, C-TMP, D- port for in-situ reflectivity, E-load lock, F-SIP, G-quartz
crystal and masking arrangements, H- sample rotation, I-sample loading
Chapter-3
69
designed as above. The net gas load on the pumping system is the combination of out-
gassing, leaks, and the gas load due to the volume of the evaporator. Proper cleaning and
polishing the internal surfaces of the system minimizes the out-gassing in the system. A leak
rate of <10
-10
Torr-litre/s was achieved by employing proper sealing and welding techniques.
The system is pumped by a combination of a turbo molecular pump (TMP) of capacity 500
litres/s and a sputter ion pump (SIP) of capacity 250 litres/s. The TMP is connected to the
upper part of the deposition system through a rotating conflat flange and is isolated by a
UHV gate valve. The SIP, too, is mounted on the main chamber through a rotating conflat
flange and isolated by a UHV gate valve. The pump-down time, i.e., the time required to take
the deposition system from atmospheric pressure to 10
-3
Torr using a rotary pump of
pumping speed 25 m
3
/h is 17 minutes. The time required to evacuate the complete system to
a pressure of 110
-6
Torr using the TMP is approximately 3 hours. The base pressure attained
after baking the system for 12 hours is 110
-8
Torr, using the TMP. A combination with a
SIP of capacity 250 litres/s is used to obtain the ultimate pressure of 110
-9
Torr. The
vacuum in the chamber is measured by different pressure gauges i.e., one Pirani gauge and
two cold cathode gauges (MKS make). A residual gas analyzer is mounted on the upper part
of the deposition chamber to check the quality of its ambient (the partial pressure of various
gases) prior to a deposition run.

3.3.3 Electron Guns and Shutters

The bottom part houses a linear, three-crucible 6 kW electron gun (Telemark, model
568), which is UHV-compatible and has a 270
0
bend. The volume of each crucible is 15 cm
3
.
This e-gun is mounted on a conflat flange along with feedthroughs for cooling water and
electrical high voltage supply. The required water flow rate is 4 to 7 litres/min, the water
temperature being maintained at 17
0
C-25
0
C by a chiller unit. The gun is provided with an
auto-controlled linear shutter arrangement placed above it. Water flow, vacuum gauges, and
the high voltage rack door are all inter-locked, such that if any one is not functioning well,
the power supply is shut down. The electron beam can be swept laterally and longitudinally
to focus on the source materials and can be visually inspected though one of the angled port
in the top part of the deposition chamber.

Chapter-3
70
3.3.4 Substrate Holder and Masking Arrangements

The substrate holder is positioned ~75 cm (can also be varied) from the electron gun to
ensure uniform deposition, with thickness uniformity of 1% over a 20 cm x 10 cm substrate.
Rotation of the substrate is done through a rotary feedthrough arrangement located at top of
the chamber. A separate substrate holder was designed so that it could either be cooled by
liquid nitrogen or heated. A large number (up to eighty) of small substrates can be loaded at a
time with separate masking arrangements made through linear and rotary feedthroughs.
Therefore, the system is capable of depositing films of different thickness in a single run. The
substrate loading is done though a load-lock system, which is described below

3.3.5 Load-Lock System for Sample Transfer

A load-lock chamber with a quick access door was also designed and fabricated. The
load- lock is connected to the top part of deposition system through a conflat flange isolated
by a UHV gate valve. The load-lock chamber is also cylindrical, 210 mm in length and 220
mm in diameter, made of SS 304L, and is mounted horizontally. The load-lock system
consists of a port for pumping, a port for pressure measurement, a port for loading samples, a
view port, and a port for the magnetic transfer rod. The pumping is done by a TMP (250
litre/s), backed by an oil-free diaphragm pump to achieve a base pressure of 110
-8
Torr. A
cold-cathode gauge is used for pressure measurement. The sample is transferred from the
load-lock chamber to the main chamber through a magnetic sample transfer rod with a 70 cm
travel, thereby protecting the vacuum in the main chamber.

3.3.6 Automation of Thickness Control

The thickness of the film being deposited can be monitored using a quartz crystal
oscillator coupled to a frequency counter. The quartz crystal is mounted in the same plane
and close to the substrate. To minimize thermal drift due to radiant heat during deposition,
the quartz crystal is cooled by water at ~18
0
C flowing at the rate of ~ 5 litre/min. The
frequency counter is coupled to a personal computer (PC) through a RS 232 interface. Film
thickness is monitored live through a PC. A photograph of the assembled system along
with all the electronics is shown in Fig. 3.7. All the samples used in this present thesis were
fabricated using this e-beam evaporation system. The detailed performance testing of the
deposition system was carried out as described below.
Chapter-3
71

3.4 System Performance and Results

After fabrication and chemical cleaning of deposition system, the accessories were
installed one by one with proper care during installation to obtain UHV conditions. This was
done first by vacuum testing accompanied by residual gas analysis. Then, thin films of
various materials were deposited to realise optimum deposition conditions, such as
deposition rate control and precise thickness control. Next, thickness uniformity was checked
over a large area. Finally, bi-layer and multilayer samples were deposited successfully.
Further, the versatility of the system was checked through in situ resistivity measurement, the
facility designed to study thin film growth, which is important for the optimization of x-ray
multilayer performance

3.4.1 Vacuum Testing

During the assembly of the e-beam system, helium leak test was carried out using a a
leak detector. The total leak rate was reduced to ~110
-10
Torr-litre/sec. After leak testing,
Fig. 3.7 A photograph of the indigenously- developed e-beam evaporation system, along
with the associated electronics.
Chapter-3
72
the actual vacuum testing of the deposition system was done using a 500 litre/s TMP. The
system was pumped by the TMP for 24 hours, and a pressure of ~110
-7
Torr was reached.
At this stage, the RGA spectrum showed a negligible partial pressure of O
2
and N
2
, which
indicated that the system had no leaks. The RGA also showed that the major constituent
among the residual gases was water vapor, due to moisture adsorbed on the inner wall of the
deposition chamber. To remove water vapor, the system was baked thoroughly at 200
0
C for
24 hours, except the O ring portion, which was kept at 120
0
C. When the system was
brought to ambient temperature, the chamber pressure was ~810
-9
Torr. The SIP pump was
then turned on. Ten hours later, the ultimate pressure in the system was ~110
-9
Torr, with
the RGA showing negligible hydrocarbon contamination (<10
-10
Torr), H
2
and helium being
the major residual constituents present in the system.

3.4.2 Deposition of Thin Films of Various Materials

The actual performance of the e-beam system was tested with the deposition of thin
films, bi-layers, and multilayers of various materials. First, a number of trial depositions of
thin films of Au, Pt, Mo, Fe, C etc. were carried out. For these depositions, ultrasonically
cleaned float glass substrates of size ~3 cm
3 cm were used. Prior to film deposition,
the root mean square (rms) surface
roughness of these substrates was
characterized using hard x-ray reflectivity
(XRR). The source-to-substrate distance
was kept constant at ~ 75 cm for all
depositions. The source materials were
thoroughly degassed prior to deposition, to
reduce gas load. Film thickness was
monitored using a quartz crystal oscillator,
keeping the deposition rate constant at ~ 5
/min. The samples were characterized by
hard XRR using a Cu-K
source
(wavelength =1.542 ). Figure 3.8 shows
the measured and fitted XRR spectra of
Fig. 3.8 X-ray reflectivity spectra of
(a) Au films, (b) Pt film and (c) Mo
films, measured at Cu K
wavelength
0.0 0.1 0.2 0.3 0.4 0.5 0.6
10
-7
10
-4
10
-1
10
2
Au 200

q
z
(
-1
)
Au 100
(a)
10
-7
10
-4
10
-1
Measured
Fitted
(b)
Pt 100

R
e
f
l
e
c
t
e
d

I
n
t
e
n
s
i
t
y

(
a
.
u
)
10
-8
10
-5
10
-2
10
1
10
4
(c)
Mo 300
Mo 100

Chapter-3
73
Au, Pt, and Mo films of different thicknesses. The reflected intensity is plotted (log scale)
against momentum transfer perpendicular to sample surface (q
z
=4sin/). The XRR spectra
were fitted using Parratts formalism
12
. The measured spectra show a clear oscillation pattern
due to interference of x-rays from reflected the vacuum/film interface and the substrate/film
interface. The measured data provide a good fit. The microstructural parameters obtained
from the best-fit results are tabulated in Table 3.2. Au films and Pt films can be fitted well
with a single-layer model, whereas Mo films need a low-density layer of thickness 13 at
the top to provide a good fit. This low-density film arises because of the formation of a Mo
oxide when exposed to air. To verify this, Mo films of the same thicknesses were fabricated
with C as a capping layer. The results are given in the next section. Finally, film thickness
uniformity was checked by depositing on smaller substrates placed at different positions over
an area measuring 20 cm 10 cm. The variation of thickness over this area was found to be
less than 1%.

Table 3.2 Best-fit results of Au, Pt, and Mo thin films using X-ray reflectivity measurements.
t and indicates thickness and root mean square roughness of the layer, respectively.
f

stands for electron density of the layer as a percentage of bulk density.

Fitted Parameters Film Deposited
thickness ()
Layer t () ()
f
Au 100 8 96% of bulk Au (film 1)

100
Float glass

5
Au 199.5 10 96% of bulk Au (film 2) 200
Float glass

5
Pt 99.5 6.5 97% of bulk Pt 100
Float glass

3.8
Mo oxide 13 7 60% of Mo
Mo 94 8 97% of bulk
Mo (film1) 100
Float glass

5
Mo oxide 13 8 60% of Mo
Mo 293 10 97% of bulk
Mo (film2) 300
Float glass

5
Chapter-3
74
3.4.3 Deposition of Bi-layers

The measured and fitted XRR spectra of C-on-Mo bi-layer structures with varying Mo
thickness are shown in Fig. 3.9. The XRR spectra are well-fitted without low-density Mo
oxide layers on top of Mo layers, necessary to fit XRR data of a bare Mo film. The best-fit
results of the C/Mo bi-layer structures are tabulated in Table 3.3. The fitted thickness values
are well-matched with the deposited ones. These results indicate that thickness errors in the
films deposited are very small (<0.2%). It is thus concluded that the e-beam system
developed in house is capable of providing high quality ML structures with minimum
thickness errors.

Table 3.3 Best-fit results of C-on-Mo bi-layers films using X-ray reflectivity measurements.
Fitted Parameters Film Deposited
thickness ()
Layer t () ()
f
C 50 6 96% of C
Mo 100 8 97% of bulk
Mo/C (film1) 100/50
Float glass

5
C 50 6 96% of Mo
Mo 299.5 10 97% of bulk
Mo/C (film2) 300/50
Float glass

5
Fig. 3.9 X-ray reflectivity spectra of
(a) Mo (100)/C(50) bi-layer and

0.00 0.07 0.14 0.21 0.28 0.35
10
-5
10
-3
10
-1

(b)
R
e
f
l
e
c
t
e
d

I
n
t
e
n
s
i
t
y

(
a
.

u
.
)
q
z
(
-1
)
(a)
Mo (100)/C(50)
10
-6
10
-4
10
-2
10
0
Measured
Fitted
Mo (300)/C(50)
Chapter-3
75
3.4.4 Deposition of Multilayers

The capability of the e-beam system to yield x-ray multilayer structures (MLS) was
tested by fabricating various MLS, such as Mo/Si, Mo/C, Mo/Y, W/C, Fe/Si, and Fe/B
4
C.
XRR data on MLS of [Mo(30 )/Si(60)]
5
and [Mo(25 )/C(35)]
10
are shown in Fig. 3.10.
The MLS show higher order Bragg peaks, along with distinct N-2 Kiessig oscillations
indicating that high quality in terms of rms roughness and thickness errors.

The XRR data were fitted assuming varying layer thickness, densities of layers, and
roughness, and assuming the formation of interlayers. The structural parameters of the ML
structures obtained from best-fit results are tabulated in Table 3.4. In Mo/Si ML, the best fit
is obtained assuming an interlayer at both the Mo/Si and Si/Mo interfaces. The ML is thus
fitted best using a four-layer model (discussed in detail in Chapter 4). The two interlayers are
asymmetric in thickness, i.e., the thickness of the Si-on-Mo interface is smaller than that of
the Mo-on-Si interface. In the case of the Mo/C ML, no interlayer is formed, and the ML is
well-fitted using a two-layer model. The fitted thicknesses of layers match well readings
from the thickness monitors.
Fig. 3.10 X-ray reflectivity spectra of (a) Mo/Si multilayer and (b)
Mo/C multilayer, measured at Cu K
wavelength
0.0 0.1 0.2 0.3 0.4 0.5 0.6
10
-8
10
-6
10
-4
10
-2
10
0

[Mo(30)/Si(60)]
5
R
e
f
l
e
c
t
e
d

I
n
t
e
n
s
i
t
y

(
a
.

u
.
)
q
z
(
-1
)
[Mo(25)/C(35)]
10
(a)
0.0 0.1 0.2 0.3 0.4
10
-8
10
-6
10
-4
10
-2
10
0
Measured
Fitted
(b)
Chapter-3
76

Fitted Parameters Film Deposited thickness
()
Layer t () ()
f
Si 48 5 96% of bulk
Interlayer
(Si-on-Mo)
8 5 65% of Mo
Mo 22 6 97% of bulk
Interlayer
(Mo-on-Si)
10 5 65% of Mo
[Mo/Si]
5
ML Mo(30 )/Si(60)
Float glass

3
C 35 7 96% of Mo
Mo 25 8 97% of bulk
[Mo/C]
10
ML Mo(25 )/C(35)
Float glass

4

Table 3.4 Best-fit results [Mo/Si]
5
ML and [Mo/C]
10
ML structures using X-ray
reflectivity measurements.

3.4.5 In situ Resistivity Measurement Facility

The versatility of the e-beam system is that it has built-in characterization facilities, such as
for making in situ resistivity measurements at different substrate temperatures. For low
temperatures, the substrate is cooled by liquid nitrogen, and its temperature is measured by a
thermocouple. The thickness dependence of film sheet resistance is measured in situ, during
film deposition, by the van der Pauw method, using pressure contacts. All the necessary
precautions were taken to maintain UHV compatibility, by making electrical connections via
a feedthrough welded to a conflat flange. To obtain the average resistivity of the film and to
ensure its uniformity, a pair of terminals is alternately used for current passage and voltage
measurement. The details of the arrangements and results on molybdenum thin films are
given in Chapter 4.

Chapter-3
77
3.5 Conclusions

A versatile ultra high vacuum (~110
-9
Torr) electron beam evaporation system, with a
load-lock system arrangement has been designed and fabricated to study physics of x-ray
multilayers and the development of x-ray optical elements, especially for the Indus-1 and
Indus-2 synchrotron radiation facilities. Since the system was indigenously developed, it has
a large number of extra ports to accommodate different accessories for different thin film-
related experiments. The system has a 6 kW multi-pocket electron gun (3 crucible each 15
cm
3
in volume) with an automated shutter arrangement that allows the deposition of
multilayer structures. The system has the facility for substrate rotation during film deposition
to ensure better thickness uniformity. Automated control of shutter movement and thickness
monitoring are done through a PC. Thin films and multilayers have been successfully
fabricated using the clean environment achieved in the deposition chamber. The analysis of
the thin films of different materials deposited in the system confirm that films of desired
thickness can be deposited in it with good control over the deposition rate. The system is able
to coat substrates as large as 20 cm 10 cm, with thickness non-uniformity less than 1%,
when the source-to-substrate distance is about 75 cm. The results of x-ray reflectivity
measurements on trial multilayer structures of Mo/Si and Mo/C indicate that the system is
capable of fabricating multilayer structure with good control over individual layer thickness
and hence the period thickness. An in situ electrical measurement set-up is designed and
mounted in the system to carry out electrical measurements during film growth.
Chapter-3
78

3.6 References

1.
M.B. Stearns, C.H. Chang, D.G. Stearns, J. Appl. Phys. 71, 187 (1992).

2.
A.K. Petford-Long, M.B. Stearns, C-H. Chang, S.R. Nutt, D.G. Stearns, N.M. Ceglio,
A.M. Hawryluk, J. Appl. Phys. 61, 1422 (1987).

3.
Y.I. Jdiyaou, M. Azizan, E.L. Ameziane, M. Brunel, T.A. Nguyen Tan, Appl. Surface
science 55, 165 (1992).

4.
A. Ulyanenkov, R. Matsuo, K. Omote, K. Inaba, J. Harada, J. Appl. Phys. 87, 7255
(2000).

5.
S.V. Gapanov, F.V. Garin, S.A. Gusev, A.V. Kochemasov, Y.Y. Platonov, N.N.
Salashenko, E.S. Gluskin, Nucl. Instrum and Meth., 208, 227 (1983).

6.
S. Masataka, S. Seki, Y. Isino, H. Nagata, Y. Suzuki, Jpn. J. Appl. Phys. 31, 1219
(1992).

7.
J. E. Mahan, Physical Vapor Deposition of Thin Films (Wiley, New York, 2000)
p.199.

8.
D.L. Windt, R. Hull, W.K. Waskiewiez, J. Appl. Phys. 71, 2675 (1992).

9.
H.R. Kaufman and R.S. Robinson Operation of Broad Beam on Sources,
(Commonwealth Scientific, Alexandria, Virginia, USA, 1987).

10.
J.J. Cuomo, S.M. Rossnagel, H.R. Kaufman (eds.), Handbook of ion beam
processing Technology, (Noyes Publications, USA, 1989).

11.
E. Spiller, Soft X-ray Optics (SPIE. Washington, USA, 1994).

12.

Chapter4
79

Chapter-4

A Study of Thin Film Growth

In this Chapter, a detailed study of the growth of ultra-thin molybdenum films on
insulating substrate at different substrate temperature is described. The nature of thin film
growth and a microscopic picture at different growth stages are deduced from modeling of
sheet resistance data obtained in situ, during film growth.
A detailed study of the nucleation, island growth, discontinuous-to-continuous
transition, and amorphous-to-crystalline transition of molybdenum films is presented. The
various conduction mechanisms extant have been identified: tunneling between isolated Mo
islands for the thinnest (discontinuous) films, surface scattering for intermediate thickness
(amorphous continuous film) and contribution of grain boundaries in thick films
(crystalline). In the island growth region, the conductance (G) has two exponential
dependencies on thickness ( ) t : t B G
1
log and t B G
2
log , explained as reflecting
anisotropic and isotropic growth of islands, respectively. In the insulator-to-metal transition
region, the conductance of the films follows the scaling law ( )
q
c
t t G , where q and t
c
are
the critical exponent and the critical thickness, respectively. The value of q agrees well with
the theoretically predicted values for critical exponent of conductivity in a two-dimensional
(2D) percolating system. The critical fractional coverage of the films, 5 . 0 =
c
x further
confirms that film growth is predominantly 2D. The microstructural analysis of film growth
derived from in situ sheet resistance measurements is in good agreement with that predicted
by percolation theory. We also provide experimental evidence for the universality of the
conductivity critical exponent in 2D percolating system by demonstrating the identity of the
conductivity critical exponent in the two geometries studied.
Chapter4
80
4.1 Introduction

Much attention has been paid over the last several decades to study the early stages of
thin film growth, especially to understand the nucleation and growth of metal islands that
undergo coalescence that results in the percolation of a continuous thin film
1
. Of particular
interest is the structural information on island films, such as island density, island size, and
fractional surface coverage. This information is invaluable for the understanding of the
kinetics of thin film formation, which strongly affects electrical characteristics of these films.
Recently, there has been a great interest in the percolative characteristics of island films.
2
For
example, the determination of the critical exponent q associated with percolation
conductivity for the growth of thin metallic films on insulating substrates. The percolation
theory has been discussed for two dimensions
3
(2D), three dimensions
4
(3D) and, recently,
Travenee and Markos
5
have studied critical conductance distribution in various dimensions
theoretically. Dubson and Garland
6
have used the analog simulation technique to study the
electrical conductivity transition involving 2D model percolation systems. They have found
the critical exponent to be 03 . 0 29 . 1 = q for site percolation on a square lattice, and
07 . 0 34 . 1 = q for random void continuum percolation, in agreement with the theoretical
prediction that the conductivity exponents for these two systems are the same. Theory
7,8

predicts that the critical exponent of conductivity q is, in general, non-universal, but that it is
universal for 2D growth. Using a computer-controlled plotter, Han et al.
9
have fabricated a
well-defined 2D anisotropic percolation system consisting of insulating random ellipsoids on
a thin aluminum film. They have measured the conductivity exponent q and found that q for
an anisotropic percolation system approaches the same value as it does for the isotropic case
near the critical region. They proved the universality concept, that the critical exponents
should be independent of the detailed microstructure, as predicted by Lobb, Frank and
Tinkham
10
theoretically. The universality of the critical exponent has been verified
experimentally .
11

The early stages in the growth of metal films on an amorphous substrate consist of the
nucleation of small metal clusters. The clusters grow into islands of the condensed phase. As
the metal content increases, the islands grow in size and get interconnected to form infinite
chains of metal, enabling continuous current paths extending through the material which "fill
in to form a continuous metal film ( 1 = x and
min
t t = ). The electrical properties of such
Chapter4
81
systems vary continuously as the metal content is changed. The variation of conductance as a
function of thickness is related to the fractional surface coverage via the growth mode of
islands as influenced by substrate temperature.
12
The percolation threshold
c
t (also known as
critical thickness) marks an insulator-to-metal (I-M) transition, and is accompanied by a
sharp drop in dc resistivity.
13
The effective thickness at which the resistance begins to fall is
different for different substrates and substrate conditions, e.g., substrate temperature.
14
The
mechanism of electrical conduction in discontinuous metal films on insulating substrates has
been the subject of several papers, which have been reviewed by Morries and Coutts.
15
The
main conduction mechanism is electron tunneling between metal islands, as assumed in
existing theory. Most theoretical treatments assume a homogeneous system. In practice,
ultra-thin evaporated metal films on insulating substrates lend themselves quite well for the
study of the conduction mechanism of isolated metal particles. Particles of almost any
desired size can be prepared by simple changing experimental parameters, such as average
film thickness and substrate temperature. Uniform deposition over the substrate is obtained
by keeping a large source-to-substrate distance. In evaporated metal films which are
extremely thin and deposited at substrate temperature ~300 K.
16
, the particle size is
reasonably uniform.
The microstructure of the percolating system has been studied theoretically by Amar et
al.
17
in terms of island density and island size distribution, as a function of coverage. They
have presented the anisotropic island growth of a sub-monolayer film grown by molecular-
beam-epitaxy at low temperature. Bruschi et al.
18
have examined the conformity of
anisotropic island structure at low temperature. They have studied temperature dependence of
thin metal film nucleation, growth, and percolation on amorphous substrates, by Monte Carlo
simulations. They observed that the size and shape of islands change with temperature. They
show theoretically that the transition temperature from anisotropic islands to isotropic islands
occurs in the temperature interval 37-200 K. The microstructure of the percolating system
has also been studied experimentally. Examples are experiments on gold films using
transmission electron microscopy (TEM).
19
indium oxide films using in situ conductivity
12

measurements, and antimony films using both TEM and in situ conductivity methods.
20
Thus, percolating systems can be studied experimentally using both electron microscopy and
in situ resistance measurements. The determination of the insulating-to-metallic transition
region, fractional surface coverage, and hence critical coverage, using electron microscopy
Chapter4
82
may be erroneous due to contamination of ultra-thin film when exposed to air. The in situ
resistance measurement method is a sensitive one for studying percolating systems
experimentally. This method has been applied to thin films of various materials, for example,
silver,
11
indium oxide,
12
platinum,
14
antimony,
20
nickel,
21
chromium
22
and bismuth.
23
One of
the difficulties in the experimental determination of the critical exponent of conductivity is
the determination of the critical fractional coverage
c
x and the I-M transition region. As per
previous work, the I-M transition region and the critical exponent of conductivity have been
determined by the least-square fit method
21,22
of resistance data obtained by in situ
measurements.
In this chapter, we present a simple method for the experimental determination of the
I-M transition region and of the critical exponent of electrical conductivity in percolating
systems. We apply this method to molybdenum films. This is, to our knowledge, the first
direct experimental determination of the I-M transition region and of the critical exponent of
conductivity for this material. We also studied the discontinuous- to-continuous transition
and observed the amorphous-to-crystalline transition in molybdenum films. We demonstrate
the determination of structural parameters such as the relative average nucleation density,
island area, and fractional surface coverage of island Mo films. The microscopic picture of
island films growth derived from in situ sheet resistance measurements is found to be in good
agreement with that predicted by percolation theory. We also provide experimental evidence
for the universality of critical exponents of conductivity in 2D percolating systems.

4.2 Theoretical Approach

We present a brief theoretical background to the nucleation and growth of thin films,
and to their electrical transport properties, before proceeding to a description
of our experimental results and their discussion. In this review, we first
discuss the different stages of film growth, followed by that of the factors
affecting the film growth process. This is followed by a survey of the different
mechanisms for electron transport in discontinuous and continuous metallic
thin films.

4.2.1 Different Stages of Film Growth

Chapter4
83
Three different types of film growth viz., island growth, layer-by-layer growth, and
layer-followed-by-island growth, have been presented pictorially in Chapter 1. Since the type
of growth involved in this present thesis is island growth mode, we provide below a general
picture of the sequence of the nucleation and growth steps involved in the formation of a
continuous film in the island growth mode.
The overall film growth process may be divided into four different stages: (a) the
nucleation and island growth stage, (b) the coalescence process, (c) the porous network stage
(channels and holes stage) and, finally, (d) the continuous film. A schematic of the various
stages of film formation is shown in Fig. 4.1. The deposited atoms (adatoms), on impinging
the substrate, get adsorbed on the surface of the substrate. The adsorbed species are not in
thermal equilibrium with the substrate initially and move over the substrate surface, and
interact among themselves, forming bigger clusters. The clusters, also known as nuclei, are
thermodynamically unstable and may tend to desorb in time. If the deposition parameters are
such that a cluster collides with other adsorbed adatoms before getting desorbed, it starts
growing in size. After reaching a certain critical size, the cluster becomes thermodynamically
stable and the nucleation barrier is said to have been overcome. This stage involving the
formation of stable nuclei is known as the nucleation stage. To calculate the smallest stable
nuclei size, one has to calculate the Gibbs free energy as follows. Consider a homogeneous
nucleation of a spherical solid phase of radius r from a pre-existing supersaturated vapor. In
such a process, the gas-to-solid transformation results in a reduction of the chemical free
energy of the system given by
V
G r
3
3
4
, where
V
G corresponds to the change in chemical
free energy per unit volume. Generally
V
G is negative (i.e., free energy decreases).
Fig.4.1 Schematic of different stages of film formation in island growth mode
Nucleation Island grow Coalescence Channels Holes
C
o
n
t
i
n
u
o
u
s

F
i
l
m

Chapter4
84
Simultaneously, new surfaces and interfaces form. This results in an increase in the surface
free energy of the system given by
2
4 r , where is the surface energy per unit area. The
total free-energy change in forming the nucleus is thus given by

2 3
4
3
4
r G r G
V
+ = , (4.1)
and minimization of G with respect to r yields the equilibrium size (critical size) of
*
r r = .
Thus,
0 = G
dr
d

V
G
r
=
2
*
(4.2)
Substituting in equation 4.1 gives

( )
2
3
*
3
16
V
G
G
=

, where G
*
G(r
*
) (4.3)
The variation of the Gibbs free energy with the radius of nuclei is shown in Fig. 4.2.
*
G
represents an energy barrier to the nucleation process. Nuclei with radius larger than r
*
have
surmounted the nucleation energy barrier and are stable. They tend to grow larger, while
lowering the energy of the system. The smallest detected nuclei have a size of 10 to 30.
24

The nucleation density and the average nucleus size depend on a number of parameters such
as the energy and rate of the impinging species, the activation energy of the adsorption,
thermal diffusion, topography, chemical nature of the substrate and the temperature. A
nucleus grows in size to form islands. A nucleus can grow both parallel to the substrate by
surface diffusion of the adsorbed species, and perpendicular to it by direct impingement of
Fig. 4.2 Gibbs free-energy change as a function of radius of nucleus. r
*
is the
critical nucleus size, and
*
G is the critical free-energy barrier for nucleation.
Chapter4
85
the adatoms. However, in general, the rate of lateral growth is much higher than in the
perpendicular direction.
The next stage involves merging of the islands by the coalescence phenomenon.
Coalescence decreases the island density, and is characterized by a reduction in the total
projected area of the islands on the substrate. Coalescence continues until a connected
network with unfilled channels in between develops. Continuing further deposition, the
channels fill in and shrink, leaving isolated voids behind. Therefore, this is also known as the
hole stage. Finally, with further deposition, the hole surface area gets filled with deposited
material, and the film is said to be continuous. This collective set of events occurs during the
early stages of film formation. Since, in x-ray multilayers, each layer should be continuous to
form a sharp boundary, it is important to study the thin film growth process for achieving
optimum performance in such multilayer structures.

4.2.2 Factors affecting the Film Growth Process

The topographical and microstructural details of a thin film of a given material
depend on the kinetics of growth and hence on the substrate temperature, the source and
energy of impurity species, the chemical nature and the topography of the substrate etc.
These parameters influence the surface mobility of the adsorbed species: kinetic energy of
incident species, deposition rate, supersaturation, sticking co-efficient, and the level of
impurities. Therefore, the physical structure is affected by these factors. Increasing the
kinetic energy of incident species increases surface mobility. However, at sufficiently high
kinetic energies, surface mobility is reduced, due to the penetration of the incident species
into the substrate. Surface mobility can also be enhanced by increasing the substrate
temperature. Under conditions of a low nucleation barrier and high super saturation, the
initial nucleation density is high and the size of the critical nucleus is small. This results in
fine-grained, smooth deposits, which become continuous at small thicknesses. High surface
mobility, in general, increases the surface smoothness of the films by filling in the
concavities. Density is an important parameter of the physical structure of thin films. A
general behavior observed in thin films is a reduction in film density with decreasing film
thickness. The adhesion of a film to the substrate is strongly dependent on the chemical
nature, cleanliness, and the microscopic topography of the substrate surface. The adhesion of
the films is better for higher values of (a) kinetic energy of incident species, (b) initial
Chapter4
86
nucleation density, and (c) adsorption energy of the deposit. The presence of contamination
on the substrate surface may increase or decrease the adhesion, depending on whether the
adsorption energy is increased or decreased, respectively.

4.2.3 Electron Transport in Metallic Films

Metal films usually grow by the island
25
growth mode (Volmer-Weber) and films are
discontinuous in nature in early stages of growth. Metal films, hence, do not provide a
continuous bridge for electric current in the island stage. The conduction mechanisms at this
stage are, of course, different from metallic conduction. The basic principles of the electron
transport mechanisms in each growth stage are briefly discussed in the following sections.

4.2.3.1 Conduction in Discontinuous Films

At the initial stage of growth, the film is discontinuous, as it is comprised of discrete nuclei
which, with increasing duration of deposition, lead to metal island formation. As the
deposition process is continued, these nuclei grow in size, thus reducing the effective
distance of separation them, as discussed in Chapter 1. Conduction electrons from one island
will not be able to move freely to the nearest neighbor, because of the gap between them.
Hence conductivity of such a film cannot be explained on the basis of continuous movement
of electrons throughout the film as in the case of continuous films. Since islands are
separated by gaps with some dielectric in between, say air or vacuum, there will be a
potential barrier between two islands and electrons have to cross or overcome this barrier to
exhibit conductivity under an applied electric field. During the early 1960s, the electrical
properties of discontinuous films were widely interpreted in terms of thermionic emission
26
.
Following Neugebauer and Webb
27
and many researchers proposed a tunneling model
28, 29
.
Most work published since then has been based to some extent on this fundamental concept,
with each variation dominating at different temperatures and in different geometrical
situations i.e., for different island sizes and inter-island spacings.
30

4.2.3.1.1 Transport by Thermionic Emission

As per the report of Minn
31
, the barrier between particles is taken to be lower than the
work function of the bulk material by a factor of q
2
/d due to image forces, where is a
Chapter4
87
function of the size of the islands and the distance d between them. The conductivity is given
as
|
|
.
|
\
|
=
d
q
KT k
BdqT
2
1
exp

(4.4)
where is the work function for bulk material and B is a constant (characteristic of each
film). This mechanism dominates for d 100 and T>300 K.
32

4.2.3.1.2 Transport by Tunneling

Tunneling cannot occur classically but, rather, is a quantum mechanical effect. As a
quantum effect, it requires the solution of Schrodingers equation for the boundary conditions
appropriate to island structure. Inside the island, the wave function is approximated by plane
waves, outside by damped ones. Applying a field displaces the relative position of the Fermi
levels of neighboring particles and the tunneling probability increases for a transition of
charge from island to another in the direction of the field. The conductivity due to quantum
mechanical tunneling is given as
30

|
.
|
\
|
|
.
|
\
|
=
KT
E
B m
h
d
D h
m A
exp 2
4
exp
2
2

, (4.5)
where A and B are constants, is the potential barrier between islands, is the activation
energy, m is the mass of the electron, d is the inter-island distance, h is the Plancks constant,
and K is the Boltzmanns constant. This equation predicts that the conductivity of a film
consisting of discrete islands is independent of the applied electric field.
For a given temperature, the film conductance due to quantum mechanical tunneling
increases exponentially as the inter-island distance shrinks
33

d
e

(4.6)
1
is the characteristic tunneling distance of the electron wave function.

4.2.3.2 Conduction in Continuous Films

When the bulk thickness is reduced gradually to a stage where it is of the order of the
mean free path of electrons (mean velocity relaxation time), then the conduction electrons
will be more frequently scattered by the two surface boundaries of the film. Consequently its
conductivity is reduced relative to its bulk value. This effect due to the reduction of material
Chapter4
88
size in thin films is called the size effect. Such a thin bulk may be considered a thin
continuous film, and conduction phenomena in a continuous film are discussed in below.

4.2.3.2.1 Matthiessens Rule

While Matthiessens Rule was originally suggested for electrical conduction in bulk
metals, it is also valid for thin metal films. It states that the various electron scattering
processes that contribute to the total resistivity (
T
) of a metal do so independently and
additively; that is,
T
=
Th
+
I
+
D
(4.7)

where
Th
,
I
, and
D
are the thermal, impurity, and defect contributions, respectively, to the
total resistivity
T
. Electron collisions with vibrating atoms (phonon) displaced from their
equlibrium lattice positions are the source of the thermal or phonon (
T
) contribution, which
increases as T
5
at low temperatures
34
and linearly at high temperatures, as shown in Fig.4.3.
This is the source of the well-known positive temperature co-efficient of resistivity in metals.
Impurity atoms and defects locally disrupt the periodic electric potential of the lattice, and
electrons are effectively scattered. However, the contributions
I
and
D
are temperature-
independent, if the concentration of impurities and defects is low. An alternative formulation
of Matthiessens rule in terms of the respective mean free-electron length () between
collisions has the form

1/
T
=1/
Th
+1/
I
+1/
D
(4.8)

Fig.4.3 Schematic electrical resistivity dependence as a function of temperature for
a metal. Thermal, impurity, and defect contributions are shown
Chapter4
89
T
can be hundred to thousands of angstroms: lengths difficult to reconcile with classical
models of vibrating atoms. Clearly, quantum effects are to be invoked here. In films of
thickness d, yet sufficiently thin so that d<
T
, the new source of film surface scattering arises,
and can increase the measured film resistivity. This is discussed in below.

4.2.3.2.2 Electron Scattering from Film Surfaces

When an electron fails to traverse the full mean-free path length because its motion is
interrupted through collision with the film surface, the electrons undergo either specular or
diffuse scattering. In specular or elastic scattering, the electron is reflected much the same
way a photon is from a mirror. In this case, one can imagine removing the surface and
doubling the film thickness (or tripling it for two surface reflections, etc.). The electron now
continues on an imaginary straight path to complete the path it would have in bulk, finally
scattering at a point interior to the extended surface. If this happens, the film resistivity is the
same as in the bulk, and there is no film thickness effect on resistivity. When the scattering is
totally nonspecular, or diffuse (inelastic), the electron mean free path is terminated by its
impinging on the film surface. After a surface collision, the electron trajectory is independent
of the impingement direction, and the subsequent scattering angle is random. The resistivity
arises because fewer electrons flow through the reference plane, thus registering a lower
current flow.
Reducing the geometry in one dimension creates an additional film contribution, i.e.,
f
, to be added to the other in the equation above. If
f
is the smallest, then, it dominates
the effective film resistivity. A quantum theory of thin film conductivity was developed by
Fuchs
35
and Sondheimer
36
, and is commonly known as the F-S theory. According to the F-S
theory, thin film conductivity for pure diffuse scattering, and for very thin films (
o
>>d), is
|
.
|
\
|
+ = 423 . 0
1
ln
4
3
k
k
f
o
o
f
, (4.9)
where k=d/
o
. Clearly, as d or k shrinks,
f
is diminished and
f
rises; a size effect is
exhibited.
A more realistic model of scattering which includes an admixture of both specular and
diffuse contribution, with P being the fraction of surface scattering events that are specular,
has been given for thin films
30
as
Chapter4
90
( )
|
.
|
\
|
+ + = 423 . 0
1
ln 2 1
4
3
k
P k
f
o
o
f
(4.10)

4.2.3.2.3 Grain-Boundary Scattering

Grain boundaries (GB) are expected to be effective electron scatterers and behave like
film interfaces in this regard. As the crystallite size becomes smaller than the electron mean-
free path, a contribution to the film resistivity from GB scattering arises. Mayadas and
Shatzkes (M-S) have treated this effect
37
; the average grain diameter is equal to the film
thickness, and only GBs whose normals lie in the film plane contribute to the scattering. The
theory predicts that GB resistivity
gb
is given by
)
`
|
|
.
|
\
|
+ + =
1
1 ln
2 3
1
3
3 2
gb
o
(4.11)
The new parameter, , related to the relaxation time for GB scattering, depends on the grain
geometry and its scattering power,
R
R
D
o
=
1
(4.12)
where D is the average grain diameter and R (0<R<1) is a GB electron reflection parameter.
In the M-S theory, R plays a role similar to that of P in the F-S equations. If R=1, the film
resistivity is infinite because electrons are reflected back to where they originated. In
addition, R depends on the type of GB, impurity adsorption at GBs, and the intergrain
geometry.

4.3 Growth of Molybdenum Thin Films

Mo is an important material for the metallisation of large-scale integration/very large-
scale integrated circuits, because of its low resistivity and high melting point.
38
Mo is also
used as the back contact for advanced solar cell assemblies.
39
Alternating layers of Mo with
low-Z spacer elements such as Si, C, Be, and B
4
C are used as x-ray multilayer mirrors
40-43

and are finding extensive applications as optical elements in extreme ultraviolet lithography,
plasma diagnostics, spectroscopy, and astronomy. For such a real multilayer to exhibit its
theoretically calculated reflectivity, it is essential that the Mo layers are continuous, with
sharply defined boundaries: this requires the film growth mechanism to be predominantly
two-dimensional. Also the reflectivity would be enhanced by minimizing the absorption at
Chapter4
91
soft x-ray wavelengths that is always appreciable: this requires the thickness of absorbing
layers to be just above the t
min
for which the film is continuous.
44, 45
Only a clear
understanding of the nucleation and growth mechanisms of Mo layers in such structures
would allow one to make the right choice of optimal technological parameters. Therefore, it
is important to investigate growth mode and the discontinuous-to-continuous transition in
ultra-thin Mo films. The film growth process has been studied using in situ sheet resistance
measurements made during the growth of the film. Detailed microscopic pictures at different
stages of film growth are derived from a modeling of the sheet resistance data so obtained.

4.3.1 Experimental

Molybdenum films were deposited on smooth (rms surface roughness 3 ) float glass
substrates at 300 K and 100 K, using the indigenously developed ultra-high vacuum electron
beam evaporation system, which has been discussed in detail in Chapter 3. The substrate was
cooled by liquid nitrogen and its temperature measured by a tungsten-rhenium thermocouple.
The base pressure in the chamber was ~5.9 10
-7
Pa during the deposition when the substrate
temperature was 300 K (film 1), and ~1.2 10
-7
Pa with the substrate at 100 K (film 2). The
source-to-substrate distance was kept at 0.5 m to ensure uniform deposition. The purity of
the evaporated molybdenum was 99.9 %. The thicknesses were measured by a quartz crystal
coupled to a programmable frequency counter/timer. The frequency counter in turn was
coupled to a personal computer through the RS232 interface. To minimize the thermal drift
in the quartz crystal, it was cooled by flowing water (~20
0
C). A proper shutter was arranged
in front of the substrate through a rotary feed-through to reduce error in the measured
thickness. The deposition rate was kept at 0.1 /min and was monitored using the water-
cooled quartz crystal oscillator.
There are two possible types of errors in the measurement of film thickness: (a)
Random errors and (b) Relative errors due to thermal drift in quartz crystal. The random
errors were estimated by repeating the experiments 7-8 times and % variation in thicknesses
from one run to another was ~0.7%. The relative error was obtained by fabricating
multilayers of molybdenum and silicon with varying molybdenum thickness from 10 to 50 ,
with a period of 100 , and totaling 10 layer pairs. This way, the inner layers are not exposed
to air, and are thus uncontaminated. The individual layer thicknesses in the ML structure
were checked by hard x-ray reflectivity measurement using Cu-K
radiation. A relative error

Chapter4
92
of ~0.2% was thus established for the thickness measurement. The total error in the
measurement of thickness was estimated to be 0.7%.
The thickness dependence of film sheet resistance ( )
s
R was measured in situ by the van
der Pauw method, using pressure contacts, as shown in Fig. 4.4. Ohmic contacts to the gold
film (1000 ) pre-deposited at four points on the substrate are made by a gold tip with
diameter 3 mm. The electrodes were spaced 20 mm apart. The sheet resistance was measured
during the film growth by passing current from a current source with accuracy and stability of
5%. The voltage drop due to this current was measured by a digital multimeter with a
resolution of 10 V and a maximum error of 50 V. The current and the voltage ranges
during measurements were from 10 nA to 2 mA, and 80 V to 60 mV, respectively. The current
and voltage probes were interchanged during measurement to obtain the average sheet
resistance of film, and ascertain film uniformity. As shown in Fig.4.4, the resistance R
AC, BD
obtained by measuring potential difference between B and D when unit current is passed
through contacts A and C is given by
AC
D B
BD AC
I
V V
R

=
,
. Similarly,
AD
C B
BC AD
I
V V
R

=
,
is
measured simultaneously. The average resistance was calculated by

2
, , BD AD BD AC
R R
R
+
= (4.13)
The sheet resistance was calculated using the equation (2.28) in Chapter 2, and taking into
account van der Pauw correction factors and the geometrical factors discussed in detail in
Chapter 2.

substrat
Pre-deposited gold
Fig.4.4. Schematic diagram of the van der Pauw method for in situ sheet
resistance measurement.
Chapter4
93
4.3.2 Results and Discussion

In Fig. 4.5, we show the sheet resistance as a function of Mo film thickness for film 1 and
film 2. Four distinct regions are clearly discernible for both the films. For example, for film
1: (A) The initial region, 01 . 0 t to 0.35 nm, where the sheet resistance
s
R does not change
significantly, and is predominantly that of the substrate surface. (B) the thickness interval
35 . 0 t nm to 0.9 nm, where
s
R decreases with thickness significantly. (C) In the 3
rd
region
9 . 0 t nm to 1.8 nm,
s
R decreases by several orders of magnitude with thickness. (D) In the
fourth region ~ > t 1.8 nm, where
s
R decreases very slowly with thickness. As the film grows,
its resistivity decreases from
7
10 1 -m to 2.42 -m Fig.4.5 shows
s
R vs. t for Mo films
of thickness up to 2.2 nm, with better resolution of the experimental data for the early stages
of film growth.

The experimental points display a distinct point of inflection at 9 . 0 t nm and a less
prominent change of slope t 1.1 nm indicate the existence of critical point for films 1 and 2,
respectively. We discuss these distinct regions of film growth briefly in the subsequent
sections.
Fig. 4.5 The sheet resistance R
s
vs. thickness t for Mo thin films. The solid
triangles () denote film 1 deposited at 300 K, and the solid circles () film 2
deposited at 100 K. The inset shows the four different regions of film growth.
0 5 10 15 20
10
0
10
2
10
4
10
6
10
8
10
10
0.0 0.5 1.0 1.5 2.0
10
1
10
3
10
5
10
7
10
9
R
s

(
/

)
t (nm)
A B C D
film1
film2
R
s

(

/

)
t (nm)
Chapter4
94

4.3.2.1 Nucleation and Growth

For the convenience of data analysis, the experimental data are plotted in terms of
conductance vs. thickness in Fig. 4.6. The region (A), prior to the onset of conduction, may
be attributed to the nucleation stage. The onset of conduction occurs earlier in film 2 than in
film 1, indicating a higher nucleation density in film 2 than in film 1 (discussed in the next
paragraph). The difference in nucleation density for the two different deposition conditions is
because nucleation rate depends on the ability of the adsorbed atoms to diffuse and collide
with one another.
30
The available activation for surface diffusion is small for film 2, due to
the low deposition temperature. This implies the nuclei grow only from material received by
direct impingement from the vapor phase and because of stronger binding of adsorbed single
atoms to the substrate.
For the evaporated ultra-thin films, which are of interest here, we assume a reasonably
uniform island size.
16
In the present case, the large source-to-substrate distance ensures
uniform deposition over the substrate. In a homogeneous system, electronic conduction in a
discontinuous metal film on an insulating substrate takes place through tunneling from one
metal island to the neighboring island, because of the overlap of electron wave functions at
neighboring islands. According to this model, film conductance increases exponentially as
inter-island distances shrink
12,46

Fig. 4.6 The conductance G vs. t plots of sheet resistance measurements. The solid
triangles () denote film 1 and solid circles () denote film 2.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
10
-11
10
-9
10
-7
10
-5
10
-3
10
-1
10
1
B
1
B
1
B
2
B
1
C A
x10
3

film1
film2
Theoretical
G

(

-
1

)
t (nm)
D B
Chapter4
95

r
o
e G G

= , wherein (4.14)

1
is the characteristic tunneling distance of the electron wave function, r is the distance
between the islands. Fig. 4.7 shows a schematic diagram of island size and separation. From
this figure, the distance r between the islands can be written as = l r , where is the
characteristic linear dimension of the islands, and l is the average distance between island
centers, which may be expressed in terms of the density N of the islands per unit area;
N
l
1
= . We have used Eq. (4.14) to fit the experimental data as shown by the solid curves
in regime B. This growth stage is best fitted by two exponential dependencies of G on t:
t B G
1
log and t B G
2
log . We have attempted to interpret the difference between
these dependencies in terms of isotropic and anisotropic growth of islands as follows. The
nominal thickness t is given by the thickness monitor reading based on t m = , where
m is the increment of mass per unit area of the material deposited on the quartz oscillator
and is the material density. At this stage, we assume that the island growth is primarily
2D. The relation between island area (a
i
) and the measured t thickness according to thickness
monitor, may be written as
i i
h a N kt = , where the product a
i
h
i
is the volume of an average
isolated island; k is the ratio of the sticking coefficient of Mo for the substrate and for the
quartz crystal (thickness monitor). Thus, a simple relation between a
i
and t can be obtained as

t
h
k l
a
i
i
2
=
(4.15)

r
l
Fig. 4.7 Schematic diagram of island size and separation
Chapter4
96
Case 1, anisotropy:
If islands grow in a single predominant direction, we denote the growth as anisotropic. In this
case, a
i
may be expressed as a product
i
c , where c
i
is a characteristic width of the average
island, which is relatively constant at this stage, and ( ) t is linear
t
c h
k l
i i
2
= (4.16 a)
Case 2, isotropy:
If islands grow comparably in both the directions, we denote the growth as isotropic. In this
case, a
i
may be expressed as
2
, and ( ) t is has a square root dependence:
t
h
k l
i
2
= (4.16 b)
Thus, the ( ) t dependence is different for isotropic and anisotropic island growth.
Using Eq. 4.14, the growth region (region B) is best fitted by two exponential
dependences of G on t : t B G
1
log up to thickness 0.5 nm and t B G
2
log up to
thickness 0.9 nm for film 1, t B G
1
log up to thickness 1.1 nm for film 2. The
dependence t B G
1
log and t B G
2
log may be attributed to anisotropic and
isotropic growth of islands, respectively, as discussed above. The term anisotropic applies
to islands of irregular shape, and the term isotropic is used for nearly circular islands. The
microstructure change from anisotropic to isotropic, at thickness 0.5 nm, for film 1 is due to
the coalescence of islands. The coalescence process occurs for film 2 at thickness 0.7 nm and
is indicated by sudden slope change during island growth. The average area and density of
islands are calculated from the slope of straight line before coalescence of islands (the height
of isolated islands is relatively constant assuming 2D film growth). Between Film 1 and film
2, the ratio of the average island area is 1.15, and the ratio of the average island density is
0.86. This shows that, in the early stages of growth, the islands are more compact and smaller
in size in film 2 than in film 1. The lower density of nucleation in film 1 is reflected in the
earlier onset of conduction in film 2 than in film 1 (Fig. 4.6). During coalescence of islands, a
transition from anisotropic to isotropic growth occurs for film 1 while, for film 2, the growth
mode does not change. The island shape changes during coalescence are due to a driving
force
47
for film 1. This driving force arises from the thermal energy at the substrate surface.
At the lower temperature, the island growth occurs by direct impingement of adatoms,
Chapter4
97
resulting in anisotropic islands while, at the higher temperature, island growth occurs by
surface diffusion of adatoms. The larger islands formed by incorporation of small mobile
islands behave in a liquid-like manner, to minimize surface free energy, resulting in isotropic
islands. At a particular substrate temperature, the variation of conductance with film
thickness is due to variation of island distribution that accompanies increasing film thickness.
As the substrate temperature changes, the nucleation density and the growth mode of islands
changes, as we have observed for the two films. Similar microstructure evolution in film
growth has been observed experimentally for other materials. Examples are: isotropic growth
of gold films,
19
and isotropic and anisotropic growth of indium oxide films.
12
The
microstructure of growing Mo films deduced by us from in situ sheet resistance
measurements is consistent with the theoretical simulations by Bruschi et al.
18
for isotropic
and anisotropic growth of thin metal films on amorphous substrates in percolating systems.
Our observation of transition in the growth mode, i.e., isotropic to anisotropic growth of Mo
films is supported by the scanning tunneling microscopy study of Ag films by Caspersen et
al.
48
. They have observed the existence of a transition in the growth mode as a function of the
substrate temperature.

4.3.2.2 Study of Infinite Clusters

In the percolating stage or the I-M transition region (C), the large islands which are
formed during coalescence process grow further and join to form an infinite cluster metallic
network structure, in which the deposited material is separated by long irregular chains of
metal atoms. As the deposition continues, secondary nucleation occurs in these channels. The
secondary islands grow and touch the side of the channels. At the same time, the channels are
bridged and filled rapidly in a liquid-like manner. As the deposition continues, most of the
channels are eliminated, but many small irregular holes remain.
30
Secondary nucleation takes
place on the substrate within these holes and the growing nuclei are incorporated in a liquid-
like manner to form a continuous film. The thickness at which the infinite cluster is formed is
called the critical thickness
c
t . The minimum film thickness below which film is
discontinuous is identified as
min
t . The film undergoes I-M transition in the region
min
t t t
c
< < . In the I-M transition region, the conductance of film is described by the scaling
law
4,48,50

Chapter4
98
( ) ( )
q
c
x x x

when
c
x x < (4.17 a)
( )
q
c
x x
c
x x > (4.17 b)

where x is the occupation probability of a conducting area on the insulating substrate (also
known as fractional coverage),
c
x is the critical value of fractional coverage and q is the
critical exponent of the electrical conductivity. The value of q is a consequence of the film
growth mode. With the assumption
21
that ( )
c
x x is proportional to ( )
c
t t near
c
x , the
conductance should scale as ( )
c
t t .

The upper boundary limit point of the I-M transition region marks the onset of a
continuous film at the thickness
min
t . The thickness
min
t is identified from the minimum in
the R
s
t
2
vs. t graph, as shown in Fig. 4.8. The R
s
t
2
vs. t graph has been discussed in more
detail in the next section (4.3.2.3) for the calculation of the average grain diameter. The inset
0 5 10 15 20
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
2 3 4 5 6
10
1
10
2
10
3

t (nm)
d
(
R
s
t
2
)
/
d
t
film 1
film 2

R
s
t
2

(

n
m
2
)
t (nm)
t<t
min
, x<1
t=t
min
, x=1
t>t
min
, x=1
Fig. 4.8 The R
s
2
t vs. t plot of sheet resistance measurements. Solid triangles
() are for film 1 and solid circles () are for film 2. The inset shows the
derivative trace i.e., ( )
2
t R
dt
d
s
vs. t.
Chapter4
99
in Fig. 4.8 shows the derivative trace i.e., ( )
2
t R
dt
d
s
vs. t . The
min
t obtained from derivative
trace near t= t
min
is 1.8 nm and 2.2 nm for films 1 and 2, respectively. Rycroft and Evans
51
have identified
min
t from minimum in the Rt
2
vs. t plot for Fe, Pt and Cu films. The higher
min
t value for film 2 is due to low surface mobility. We have discussed it in more detail by
correlating it with x-ray reflectivity analysis.
52
Veldkamp et al.
45
sputter- deposited Mo on
sapphire substrate and obtained 5 . 2
min
= t nm at 300 K, while in our case, 8 . 1
min
= t nm is
obtained, for Mo films evaporated onto float glass at 300 K. The difference in t
min
value may
be attributed to the different substrates used and/or to different deposition techniques
adopted, both resulting in different surface mobilities of Mo atoms. We deposited Mo on
float glass and Veldkamp et al. have deposited on sapphire resulting difference in t
min
value
due to different surface mobilities. Moreover, atomic clusters of Mo may be incorporated in
sputter-deposited films of Veldkamp et al. Such atom clusters have lower surface mobility
than single atoms. Thus a sputtered film would require more atoms to be deposited to form
continuous films; hence a higher t
min
value. (The adhesion of the deposited films to the
substrate is strongly dependent on the chemical nature, microscopic topography, and
cleanliness of the substrate surface.
53
)
The conduction mechanism is dominated by tunneling below
c
t , as discussed earlier. At
c
t , metallic conduction starts contributing to conduction. The experimental deviation of
tunneling current occurs at thickness approximately 9 . 0
c
t nm and 1.1 nm for films 1 and 2,
respectively (Fig. 4.6). To obtain q and actual values of
c
t , Eq. (4.17b) is used with a starting
guess value of
c
t . The experimental data are plotted as G vs.
c
t t for different values of
c
t
as shown in Fig. 4.9. Despite the variation of
c
t over a narrow range, only the curve with
critical thickness 9 . 0 =
c
t nm and 1.1 nm come close to a straight line for films 1 and 2,
respectively. A least-squares fit to this curve yields the critical exponent, q= 1.29 0.02 and
1.3 0.02 for films 1 and 2, respectively. We have obtained data over eight separate
evaporations that yield an average for t
c
=0.9 0.01 nm and 1.1 0.02 nm, with the critical
exponent q=1.290.02 and 1.30.02 for films 1 and 2, respectively. The slope change in Fig.
4.9 is due to the amorphous-to-crystalline phase transition in the films, and has been
discussed in more detail in the next section (4.3.2.3).
Chapter4
100

The calculated critical exponent of conductivity agrees well with the 2D theoretical model
calculations by Dubson and Garland, within experimental error. The error of 0.01 in
c
t and
q is attributable to the overlapping of different conduction mechanisms at the two ends of
0.1 1 10
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
0.1 1 10
10
-4
10
-3
10
-2
10
-1

q=1.30.02
t
c
=1.2 nm
t
c
=1.1 nm
t
c
=1.0 nm
fitted
G

(
-
1
)
t-t
c
nm
b)
t
c
=1.0 nm
t
c
=0.9 nm
t
c
=0.8 nm
fitted
G

(
-
1
)
t-t
c
nm
q=1.290.02
a)
Fig. 4.9 The G vs. t-t
c
plot for three different values of t
c
as indicated in the
figure, (a) for film 1 and (b) for film 2. The solid line corresponds to the linear fit
of the points for which t-t
c
<1.6 nm and 2.4 nm for films 1 and 2, respectively,
where the island growth should be still parallel to the substrate surface.
Chapter4
101
the I-M region. Suresh
54
has plotted sheet resistance vs. thickness for Mo thin films at 300 K.
We have calculated the critical exponent of conductivity, as outlined here, by re-plotting his
data. The critical exponent of conductivity obtained from the data obtained by Suresh
54

is 02 . 0 3 . 1 = q , consistent with our results. This corroborates the simple method reported
here to obtain the critical exponent of electrical conductivity in the percolating system.
The critical fractional coverage
c
x is obtained from the proportionality constant in Eq.
(4.17). In 2D growth, the fractional coverage is related to thickness
21,51
as kt x = . The
parameter k is obtained with the boundary conditions
51
viz., at 0 = t ,
D
G G = and at
min
t t = ,
M
G G = . where
D
G and
M
G are the conductance of dielectric substrate and at the
onset of continuous film, respectively.
Chapter4
102
The ordinate intercept in Fig.4.9 yields the parameter, 55 . 0 = k nm
-1
and 0.45 nm
-1
for films
1 and 2, respectively. Experimental
M
G values are taken for the calculation of k as
3
10 1 . 4

=
M
G
1
and
3
10 3 . 8

1
for films 1 and 2, respectively. The critical fractional
coverage obtained, 5 . 0 = =
c c
kt x , for both the films. This corresponds well to 2D film
growth
13
. The measured data are plotted with fractional coverage in Fig. 4.10. The
experimental data were theoretically simulated using Eq. (2.17) for different values of q . The
best fit is obtained for 29 . 1 = q and 1.3 for films 1 and 2, respectively. The theoretical
0.2 0.4 0.6 0.8 1.0
10
-9
10
-7
10
-5
10
-3
10
-1
0.2 0.4 0.6 0.8 1.0
10
-11
10
-9
10
-7
10
-5
10
-3
10
-1

Experimental
Theoreical (q=0.5)
q=1
q=1.3
q=2
q=3
x
G

(
-
1
)
b)
q
Experimental
Theoretical (q=0.5)
q=1
q=1.29
q=2
q=3

x
G

(
-
1
)
a)
q
Fig. 4.10 Conductance G vs. fractional coverage x. (a) solid triangles () for film 1
and (b) solid circles () for film 2. The dotted and solid line obtained from best fit to
Eq. 2 for different values of q, as indicated in the figure. The solid line indicates the
best-fit data of q =1.290.02 and 1.30.02 for films 1 and 2, respectively.
Chapter4
103
simulation does not fit below critical thickness for films 1 and 2, as shown in Figs.4.10 (a)
and (b), respectively. This implies that, for our case, the relation ( ) ( )
q
c
x x x

is not
valid below critical thickness. The measured discrepancy could be explained by the
important role electron tunneling between particles plays when metal islands with small
separations are formed (for high melting temperature metals
14
).

We show the comparison of our results with the least square method used earlier by
various authors
21,22
for the calculation of the I-M transition region and the critical exponent
of conductivity. A sliding least-square fit using Eq. (4.17) was done on 3.5 nm segments of
experimental data of Fig. 4.5. The starting thickness
s
t specifies each segment. For each
value of
c
t and q , the corresponding
2
value is obtained. The least square fit results are
Fig. 4.11 The critical thickness t
c
, the critical exponent of conductivity q, and
2
vs. the starting thickness t
s
, obtained from the least square fit method.
(a) for film 1 and (b) for film 2.
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0
1
2
3
4
5
6
0.0 0.5 1.0 1.5 2.0 2.5 3.0
1
2
3
4
10
3
10
5
10
7
10
9
10
11
t
c
q

q
,

t
c

(
n
m
)
t
s
(nm)
b)
10
1
10
3
10
5
10
7
10
9
10
11

2
2
t
c
q
q
,

t
c

(
n
m
)
t
s
(nm)
a)
2

2
Chapter4
104
shown in Figs. 4.11 (a) and (b) for film 1 and 2, respectively. The
2
value is minimum for
01 . 0 9 . 0 =
c
t , 02 . 0 29 . 1 = q for film 1, and 02 . 0 1 . 1 =
c
t , 02 . 0 3 . 1 = q for film 2. The
values of
c
t and q are nearly constant in the starting thickness range 8 . 1 9 . 0
s
t nm and
2 . 2 1 . 1
s
t nm for films 1 and 2, respectively. The same procedure was repeated for the
segment widths 3 nm, 4 nm, 4.5 nm and 5.5 nm, giving identical results, implying that the I-
M transition region and q values obtained by two different methods are same. Confidence in
the experimental values of t
c
, t
min
and q was generated by the repetition of the experiments
over eight separate runs. The data are reproducible from run to run, yielding consistent data
such as that of Fig. 4.5. Although the critical thickness is different, the critical exponents of
conductivity are identical within experimental error in the two geometries studied. This
provides an evidence of existence of universality of critical exponents for conductivity in 2D
percolating systems. This is, to our knowledge, the first determination for universality of
critical exponent of conductivity for Mo films deposited on float glass, rather than in a model
system.

4.3.2.3 Amorphous-to-Crystalline Transition

In the thickness region
min
t t > , the scattering of electrons by thin film surfaces and grain
boundaries are used to model the continuous film conductivity. The film sheet resistance in
the metallic continuous region D (Fig. 4.5) has been fitted using the relation,
55

3
3 1 0
t A t A A Rt + + = , (4.18)

where the 2
nd
and 3
rd
terms are contributions to conduction from grain boundary scattering
and roughness scattering, respectively. The fitted plots of metallic regions ( )
min
t t > are
shown in Fig.4.12. The best-fit results are tabulated in Table I. It is clear that the grain
boundary contribution (A
1
) is zero up to thickness 2.5 nm and 3.3 nm for films 1 and 2,
respectively. This implies that the films are amorphous up to the thickness mentioned above.
The non-zero A
1
values for thicknesses greater than 2.5 nm and 3.3 nm for films 1 and 2,
respectively, imply that there is a contribution of grain boundaries to the conductance in the
films. The films undergo amorphous-to-crystalline transition at thickness of 2.5 and 3.3 nm
for films 1 and 2, respectively.
Chapter4
105

TABLE I. The best-fit results to equation 4.18 obtained in the continuous region t>t
min
for
films 1 and 2.

Parameter
Film 1
Thickness (nm) range
1.8-2.5 2.5-20
Film 2
Thickness (nm) range
2.2-3.3 3.3-20
A
0
(-nm)
A
1
(-nm
2
)
A
3
(-nm
4
)
298.06 223.34
0 223.34
598.73 1583.68
188.06 73.3
0 208.09
1093.19 3846.74
5 10 15 20
10
-2
10
-1
10
0
10
1
2 3 4 5
10
-2
10
-1
G

(
-
1
)
t (nm)
X10
film 1
film 2
Theoretical
G

(
-
1
)
t (nm)
Fig. 4.12 G vs. t measurements in the region D (t>t
min.
). The solid triangles () for
film 1 and solid circles () are for film 2. The solid line was obtained from best fit to
Eq. 3. The inset shows clearly the conduction jump at thickness t= 2.5 nm and 3.3
nm in the films 1 and 2, respectively.
Chapter4
106
This transition can be visualized from the cusp observed in the conduction (inset in Fig.
4.12). This amorphous-to-crystalline transition is the cause of slope change in Fig.4.9 (a) and
(b) for films 1 and 2, respectively. A similar amorphous-to-crystalline transition has been
observed for antimony films (Ref. 20). In Table I, A
1
value is higher for film 1 than for film
2, indicating that grain boundary scattering is greater in film 1. The in situ sheet resistance
data are used to calculate the average grain diameter
g
D . The average grain diameter is
related to resistivity as
51
( )
1
0 0 0

=
g
D , where
0
is electron mean free path in the
material and is given as
0
2
0
2 / ne hk
f
= . Here
f
k is the wave vector at the Fermi
surface, n is the electron density, e is the electron charge,
is the resistivity of an
infinitely thick film, and
0
is the resistivity of the bulk material. For convenient analysis in
terms of size effects,
56
the measured sheet resistance data have been plotted in Fig. 4.8 in
terms of R
s
t
2
as a function of film thickness. The film resistivity is linearized when
2
t R
s
is
plotted with film thickness for
min
t t >> . The slope of the linear region yields the physical
parameter
, whereas an ordinate intercept yields the specular parameter. The values of

are 242 . 0 -m and 0.037 -m for films 1 and 2, respectively. The values of
0
are
5.710
-2
-m and 9.210
-3
-m at 300 K and 100 K, respectively.
57
Thus, the average
grain diameter
g
D obtained for films 1 and 2 are 3.1 nm and 6.5 nm, respectively (assuming
b.c.c. Mo, and a spherical Fermi surface). The roughness scattering is greater for film 2 than
for film 1. This is reflected in the higher value of A
3
for film 2 than for film 1 (Table I). The
greater degree of roughness scattering in film 2 is due to the lower atomic surface mobility
which, in turn, increases the rms surface roughness.
We have also checked the density of our films using x-ray reflectivity.
52
The density
obtained from reflectivity are 99% and 97% of bulk density for films 1 and 2, respectively.
The high density indicates a low concentration of primary voids in our films, and points to
the high quality of the ultra-thin Mo films deposited in the indigenously developed e-beam
evaporator.

Chapter4
107
4.4 Conclusions

In this Chapter, we have presented a study of thin film growth (island growth),
providing the theoretical background on the different stages of film growth, and factors
effecting film growths process. This is followed by a review of the different conduction
mechanisms for electron transport in discontinuous and continuous metallic thin films. A
detailed experimental study of the growth of ultra-thin Mo films on insulating substrates at
different substrate temperatures, using the in situ sheet resistance measurement technique, is
presented. A detailed microscopic picture of the different stages of film growth has been
deduced from the sheet resistance data obtained during film growth. In the island growth
region, the conductance (G) has two exponential dependencies on thickness ( ) t :
t B G
1
log and t B G
2
log , attributed to anisotropic and isotropic growth of islands,
respectively. We have demonstrated a direct experimental method for the determination of
the insulating-to-metal transition region, and of the critical exponent of electrical
conductivity for a percolating system. The upper limit of I-M transition is obtained from the
minimum in the R
s
t
2
vs. t plot. The deviation of conduction mechanism from tunneling gives
the lower limit. The critical exponent of conductivity is found to be q= 1.290.02 and
1.30.02, with critical thickness t
c
= 0.9 0.01 and 1.1 0.02 nm, for Mo films deposited at
300 K and 100 K, respectively. These values agree well with those obtained by using a least-
square fit method. The critical fractional coverage
c
x is 0.5 for both the films. The island
growth is isotropic for Mo films deposited at 300 K, and anisotropic for the film deposited at
100 K, after coalescence takes place. The minimum thickness for which the film becomes
continuous is 1.8 nm and 2.2 nm for Mo films deposited at 300 K and 100K, respectively.
Information on relative average nucleation density, island area, and fractional surface
coverage, has been derived from in situ sheet resistance measurements. This makes possible
the physical interpretation of a percolating thin film system. The study shows that Mo films
grown on float glass substrates are predominantly 2D. The microscopic pictures of
molybdenum films deposited at two different temperatures are different but the critical
exponents of their electrical conductivity have identical values. This provides evidence for
the universality of conductivity exponents in 2D percolating systems. From the conductivity
jump observed, we have identified the crystalline transition in Mo films.

Chapter4
108
4,5 References
1.
J.A. Venables, Philos. Mag. 27, 693 (1973).

2.
K. Seal, M.A. Nelson, Z.C. Ying, D.A. Genov, A.K. Sarychev, V.M. Shalaev, Phys.
Rev. B 67, 035318 (2003).

3.
C.J. Lobb and D.J. Frank, Phys. Rev. B 30, 4090 (1984).

4.
S. Kirkpatrick, Rev. Mod. Phys. 45, 574 (1973).

5.
I. Travenee and P. Markos, Phys. Rev. B 65, 113109 (2002).

6.
M.A. Dubson and J.C. Garland, Phys. Rev. B 32, 7621 (1985).

7.
B.I. Halperin and S. Feng, Phys. Rev. Lett. 54, 2391 (1985).

8.
S. Feng, B.I. Halperin, P.N. Sen, Phys. Rev. B 35, 197 (1987).

9.
K.H. Han, J.O. Lee, S.-Ik. Lee, Phys. Rev. B 44, 6791 (1991).

10.
C.J. Lobb, D.J. Frank, M. Tinkham, Phys. Rev. B 23, 2262 (1981).

11.
M. Octavio, G. Gutierrez, J. Aponte, Phys. Rev. B 36, 2461 (1987).

12.
V. Korobov, M. Leibovitch, Y. Shapira, Appl. Phys. Lett. 65, 2290 (1994).

13.
D.J. Bergman and D. Stroud, Solid State Phys. 46, 147 (1992).

14.
I. Ostadal and R.M. Hill, Phys. Rev. B 64, 033404 (2001).

15.
J.E. Morris and T.J. Coutts, Thin Solid Films 47, 3 (1977).

16.
M. Bumer and H.-J. Freund, Prog. Surf. Sci. 61, 127 (1999).

17.
J.G. Amar, F. Family, P.M. Lam, Phys. Rev. B 50, 8781 (1994).

18.
P. Bruschi, P. Cagnoni, A. Nannini, Phys. Rev B 55, 7955 (1997).

19.
R.F. Voss, R.B. Laibowitz, E.I. Alessandrini, Phys. Rev. Lett. 49, 1441 (1982).

20.
P. Jensen, P. Melinon, M. Treilleux, J.X. Hu, J. Dumas, A. Hoareau, B. Cabaud,
Phys. Rev. B 47, 5008 (1993).

21.
L. Cheriet, H.H. Helbig, S. Arajs, Phys. Rev. B 39, 9828 (1989).

22.
J.A.J. Lourens, S. Arajs, H.F. Helbig, El-Sayed A. Mehanna, L. Cheriet, Phys. Rev B
37, 5423 (1988).

23.
J. Schmelzer, Jr., S.A. Brown, A. Wurl, M. Hyslop, R.J. Blaikie, Phys. Rev. Lett. 88,
226802 (2002).

24.
C. A. Neugebauer in Handbook of Thin Film Technology, edited by L.I. Maissel
and R. Glang, McGraw-Hill, New York (1983) p8-32.

25.
K. Wasa and S. Hayakawa, inHandbook of Sputter Deposition Technology, Noyes,
Westwood, New Jersey, USA (1992).

Chapter4
109
26.
S.S. Minn. and J. Rech, Centre Natl. Rech. Sci. Lab. Bellevue Paris, 51, 131 (1960).

27.
C.A. Neugebauer and M.B. Webb, J. Appl. Phy. 33, 74 (1962).

28.
C.J. Adkins, J. Phys. C: Solid State Phys., 15, 7143 (1982).

29.
P. Sheng and J. Klafter, Phys. Rev. B 27, 2583 (1983).

30.
L.I. Maissel in Handbook of Thin Film Technology edited by L. I. Maissel and R.
Glang, McGraw-Hill, New York (1983) p13-18.

31.
S.S. Minn and J. Rech, Centre Natl. Rech. Sci., Bellevue, Paris, 51, 131 (1960).

32.
A. Wagendristel and Y. Wang in An Introduction to the Physics and Technology of
Thin Films World Scientific, Singapore (1994) p.53.

33.
V. Korobov, M. Leibovitch, Y. Shapira, Appl. Phys. Lett. 65, 2290 (1994).

34.
M.S. Rogalski and S.B. Palmer, in Solid State Physics, Gordon and Breach
Science, Netherlands (2000) p.188.

35.
K. Fuchs, Proc. Cambridge Phil. Soc. 34, 100 (1938).

36.
E.H. Sondheimer, Adv. Phys. 1, 1 (1951).

37.
A.F. Mayadas and M. Shatzkes, Phys. Rev. B1, 1382 (1970).

38.
T P. Drusedau, K. Koppenhagen, J. Blasing, T.-M. John, J. Appl. Phys. A 72, 541
(2001).

39.
M.A. Martinez and C. Guillen, J. Material Processing Tecnology 143-144, 326
(2003)).

40.
E. Spiller in Soft X-ray Optics, SPIE Optical Engineering Press, Washington, USA
(1994).

41.
T.W. Barbee, Jr. S. Mrowka and M.C. Hettrick, Appl. Opt. 24, 883, (1995).

42.
F. Puterro-Vuaroqueaux, B. Vidal, J. Phys.: Condensed Matter 13, 3969 (2001).

43.
M. Toyoda, Y. Shitami, M. Yanagihara, T. Ejima, M. Yamamoto, M. Wanatabe, Jpn.
J. Appl. Phys., Part 1, 39, 1926 (2000).

44.
M. Lohmann, F. Klabunde, J. Blasing, P. Veit, T. Drusedau, Thin Solid Films 342,
127 (1999).

45.
M. Veldkamp, H. Zabel, Ch. Morawe, J. Appl. Phys. 83, 155 (1998).

46.
H. Bottger and V.V. Bryksin, inHopping Conduction in Solids, Academic, Berlin,
(1985).

47.
D.W. Pashley, Proc. Phys. Soc. (London) A64, 1113 (1951).

Chapter4
110
48.
K.J. Caspersen, C.R. Stoldt, A.R. Layson, M.C. Bartelt. P.A. Thiel, J.W. Evans, Phys.
Rev. B 63, 085401 (2001).

49.
A.L. Efros and B.I. Shklovskii, Phys. Status Solids, B 76, 475 (1976).

50.
B. Abeles, H.L. Pinch, J.I. Gittleman, Phys. Rev. Lett. 35, 247 (1975).

51.
I.M. Rycroft and B.L. Evans, Thin Solid Films 283, 290 (1996).

52.
M. Nayak, G.S. Lodha, R.V. Nandedkar, Solid State Physics 2003, Proceedings of the
DAE Solid State Physics Symposium, Gwalior, India, Dec. 26-30, p. 449 (2003).

53.
K. Wasa and S. Hayakawa, in Handbook of Sputter Deposition Technology, Noyes,
Westwood, New Jersey, USA, p 18 (1992).

54.
N. Suresh, Ph. D. Thesis, Devi Ahilya University, Indore, India (1999).

55.
H.-U. Finzel and P.Wissmann, Ann. Phys. 498, 5 (1986).

56.
C.R. Tellier, A.J. Tosser and G. Siddall (Ed.), Size Effects in Thin Films (Thin Film
Science and Technology, 2), Elsevier, p.51 (1982).

57.
H.P.R. Frederikse and D.E. Gray (Eds.), American Institute of Physics Handbook,
Mc-Graw Hill, New York, (1982) p 9-41.

Chapter 5
111

Chapter-5

Study of Soft X-ray Multilayers

This Chapter describes a systematic study of the surfaces and interfaces in Mo/Si,
Fe/B
4
C, and Mo/Y multilayers (MLs) for application as soft X-ray polarization elements in
the Indus-1 synchrotron radiation source. In Mo/Si MLs, the inter-diffusion of the two
materials at the interfaces leads to the formation of interlayers. The nature of these
interlayers has been studied using hard x-ray reflectivity (XRR) and depth-graded x-ray
photoelectron spectroscopy (XPS). For the analysis of the XRR data, a four-layer model is
used to account for the interlayers. We present a systematic simulation of the effect of
interfacial roughness and the thickness of the interlayers on the reflectivity profile. The
interfacial roughness reduces the intensity of individual peaks, while the thickness of the
interlayers redistributes the reflectivity profile. This model discerns the asymmetry in
interlayer thickness, at the two distinct types of interfaces, if the interfacial roughness is
small compared to interlayer thickness. Mo/Si MLs with varying periodicity, varying number
of layer pairs, and different ratios of the thickness of the high-Z layer to the period of the ML,
have been fabricated using a UHV e-beam evaporation system. In all the Mo/Si MLs, the
interlayer is found to be asymmetric. A mechanism for this asymmetric formation is proposed
based on the different heats of sublimation of Mo and Si. The composition of the interlayer is
determined using depth-graded XPS, revealing the formation of MoSi
2
at both the interfaces.
The experimental results agree well with theoretical calculations based on solid-state
amorphization, stemming from the large heat of mixing. The effective heat of formation
model reveals the formation of MoSi
2
as the first phase. Finally, Mo/Si ML based soft x-ray
reflection polarizing elements are fabricated with the desired period, to operate in the
wavelength range 12.5 nm-16.0 nm (above the Si L-edge). The reflectivity performance in the
s-polarization geometry has been tested on the Indus-1 reflectometry beam line. At 13 nm, a
reflectivity of 45% at the Brewsters angle has been achieved.
An attempt has been made to understand the interface characteristics in Fe/B
4
C.
Using XRR and glancing angle x-ray fluorescence techniques, a study of the surfaces and
interfaces in B
4
C-on-Fe and Fe-on-B
4
C bi-layers, and Fe- B
4
C-Fe tri-layers, has been
Chapter 5
112
carried out, with a systematic variation in Fe and B
4
C thicknesses. It is observed that there is
sharp interface in Fe-on-B
4
C. In the case of B
4
C-on-Fe, there is a low density (w.r.t. Fe)
interlayer at the interface. As the bottom Fe layer thickness increases, the interlayer
thickness increases, but its density decreases. The nature of interlayer is independent of the
thickness of the top B
4
C layer. The formation of this interlayer is neither due to interdiffusion
nor interlayer compound formation, but due to physical roughness at the interface. Finally, a
surface/interface study of Fe/B
4
C MLs has been carried out for their potential polarizer
application above the wavelength of the boron K-edge. Surfaces/ interface studies of bi-
layers, tri-layers and MLs of the Mo/Y system have also been carried out for potential
polarizer application in the wavelength range =81 to 124 (above the M-edge of
Yttrium).
Chapter 5
113
5.1 Introduction

The main goal of soft x-ray multilayer coatings (or mirrors) is the enhancement of
reflectivity at wavelengths and for angles of incidence where single surfaces with useful
reflectivity are not available. These ML mirrors are formed by depositing (on suitable
substrates) alternating layers of high-Z and low-Z materials that form sharp interfaces stable
over the long term, with individual layers of thickness of the order of nanometers. In these
ML structures (MLS), the amplitude reflected from successive interfaces add in phase to give
enhanced peak reflectivity in the Bragg condition
1,2
. The MLs permit high normal incidence
reflectivity and moderate energy bandwidth (10<E/E<100) in the soft x-ray/EUV region.
The composition of individual layers and the periodicity of the ML can be tailored according
to the desired wavelength and angle of incidence. These MLs have wide-ranging applications
in both science and technology, as discussed in Chapter 1. The actual performance of the
MLS depends both on the design parameters of ML structure
3
and on the quality of surfaces
and interfaces in the ML fabricated
4
. The performance of an MLS depends largely on
whether there exists a combination of materials having the right refractive indices (good
contrast and low absorption), whether these materials can be deposited in smooth thin layers,
and whether they remain stable (low reactivity and low diffusion) in the deposited state. The
roughness of the interfaces also plays an important role: rms roughness of the same order of
magnitude as the layer thicknesses will spoil the performance of ML coatings. If d is the
period of a MLS, the roughness tolerable
3
is ~d/10. Thickness errors in the MLS are also of
prime importance. The effect of roughness, interlayer formation, and thickness errors on the
reflectivity performance has been discussed in Chapters 1and 2 of this thesis. On this basis, it
may be stated that an understanding of the properties of the surfaces and interfaces of these
MLs is crucial to optimization of the actual performance of such structures.
This Chapter is arranged as follows: First, a theoretical background on the design
criteria for ML structures is given. This is followed by a detailed description of Mo/Si MLs:
their fabrication, surfaces and interfaces, and an evaluation of their performance as
polarizers, using the Indus-1 SR source. This is followed by the Fe/B
4
C system: fabrication,
and a study of the surfaces and interfaces of bi-layers, tri-layers, and MLs. Finally, the Mo/Y
system: fabrication, and a study of the surfaces and interfaces of bi-layers, tri-layers, and
MLs.

Chapter 5
114
5.2 Design Criteria of ML Structures

The main goal of ML coatings is to enhance the near-normal and/or normal-angle reflectivity
of soft x-ray /EUV radiation, and to achieve useful reflectivity despite the fact that all
materials absorb them. There are two possible types of designs of ML structures, which
would give high reflectivities: (a) quarter wave stack and (b) Bragg crystal-type
5
.

Quarter wave stack
Fig. 5.1 (a) shows the quarter wave stack design of ML structure (Reproduced from
Ref. 3), where two non-absorbing materials with different refractive indices, high and low (H
and L), are deposited alternately, with each layer thickness equal to /4. Each layer thickness
extends from a node to an antinode of the standing wave, which is generated within the ML
by the coherent superposition of the incident and reflected waves
6
.

Fig. 5.1 Two possible designs for high reflectivity mirrors. (a) Quarter wave
stack-type showing standing wave field inside a ML. L and H indicates low-Z
and high-Z material, respectively. (b) Bragg crystal-type showing the position of
high-Z material at the node of standing wave, minimizing absorption losses.
Chapter 5
115
In this design, reflectivities from successive boundaries add in phase and, for a sufficiently
large number of layer pairs, the reflectivity can approach unity for absorption-free materials.
The actual number of layer pairs required is obtained from a calculation of the Fresnel
reflection coefficient. However, if one of the materials becomes slightly absorbing, the
performance of such a stack deteriorates very fast
7
. This is because high intensity near the
antinodes produces strong absorption losses if one of the materials becomes absorbing.
Unfortunately, all materials become absorbing
8
for wavelengths below =1100 . Therefore,
it is impossible to use the quarter wave stack design to obtain high reflectivity in the soft x-
ray/EUV spectral range.

Bragg Crystal-type
Fig. 5.1 (b) shows a Bragg crystal-type design, in which very thin, strongly absorbing
layers (of high-Z materials also known as the absorber) are positioned at the nodes of a
standing wave field, and the remaining space is filled with a material with very low
absorption (low-Z materials also known as spacer). Since the absorber layers are
positioned at the node positions where the intensity is a minimum, absorption losses will be a
minimum. In this design, the small contribution of reflected intensity from all the individual
scattering layers add in phase to the reflected wave, and in the limit of very thin absorber
layers, the absorption losses approach zero. Reflectivity approaching 100% can be obtained
in the limit of a very large number of layers. For a comparison with the quarter wave stack-
type, it is noted that the reflectivity of a single quarter-wave thick film changes only slightly
(quadratically) when one reduces its thickness, while the amount of absorbing material
decreases proportionally with thickness. Therefore, ML design with a reduced thickness of
the absorbing layer, and a corresponding increase in the thickness of the spacer layer will
have reduced absorption losses.
As discussed above, the Bragg crystal type of multilayer structure (MLS) is
asymmetric in low-Z and high-Z material thicknesses. Therefore, in designing of ML
coatings an important parameter is the ratio of high-Z material thickness to the bi-layer
period, generally denoted by (gamma), i.e.,

d
t
t t
t
H
L H
H
=
+
= (5.1)
Chapter 5
116
where t
H
, t
L
, and d indicate the high-Z material thickness, the low-Z material thickness and
the period of the MLS, respectively. The low-Z material acts simply as a spacer, with as
little absorption as possible. Therefore, the optical constants of the low-Z materials,
L
and
L

(reported in Chapter 6), should be as small as possible, to provide the highest possible
refractive index contrast at the interfaces. In approximating to that limit, the high-Z layers
provide both scattering and absorption. The tradeoff then is to obtain sufficiently strong
scattering, to first approximation through refractive index contrast at the interfaces, while
minimizing the absorption by reducing the thickness of the high-Z layer at the nodal position.
To obtain sufficiently strong scattering through refractive index contrast at the interfaces, the
layer should be continuous. An analytical formula for the optimum design of periodic MLs
with a large number of periods is given by Vinogradov and Zeldovich
9
. They have obtained
the optimum value of from a solution of the equation
( )
(
+ =
L L H H
L L
opt opt

tan (5.2)
For
H L
<< an approximate solution for small values of
opt

is

3
1
3 1
|
|
.
|
\
|

H H
L L
opt

(5.3)
For coatings with small d-spacing, due to practical reasons, the typically achieved values of
are more generally in the range
8
of 0.3 to 0.5.

Quasi-periodic
In the soft x-ray/EUV region, no absorption-free materials exist, and an optimized
design is between the two types of design discussed above. Therefore, an optimum design for
a fixed number of layers should not be periodic but quasi-periodic
10, 11
. The optimum
thickness of the absorbing layer will depend on the position of the layer in the stack. Near the
bottom of the ML, and for boundaries with low reflectivities, the reflected wave is very weak
and hence there exists only a very little contrast in the standing-wave field. Therefore, the
initial layer thicknesses near the bottom should correspond to a quarter-wave stack. As the
number of layers increases, the contrast of standing-wave field increases and the thick
absorbing layer in the quarter-wave stack increases absorption at the antinode. Therefore, one
has to switch over to the Bragg crystal design with thin absorber layers and increase in spacer
layer thickness to maintain the periodicity in the stack.
Chapter 5
117

An optimally designed coating should therefore have the absorbing layer thickness
decreasing continuously from the bottom (quarter-wave type) to the top (minimum thickness
to get continuous film). The optimum thickness t
opt
of the absorber layer that gives the
highest reflectivity is shown in Fig. 5.2 for a finite number of layers
3
, where a non-absorbing
spacer material is alternated with an absorbing material with absorption co-efficient k=0.5.

5.2.1 High Reflectivity Multilayers

The goal is to enhance the low reflectivity from a single boundary by adding
reflectivities in phase from many boundaries. In the x-ray region, the normal incidence
reflectivity from a single boundary is R < 10
-4
(Fig. 2.6 in Chapter 2). Hence, the reflected
amplitude r is 10
-2
. Therefore, 10
2
amplitudes have to be added to obtain a total reflectivity
approaching one. i.e., the minimum number of periods required is

r
N
2
1
min
= (5.4)
Fig. 5.2 Optimum design for normal incidence reflectivity of periodic ML
structure with periodicity d=/2 vs. the number of periods in the structure. The
dashed curve gives the optimum thicknesses of the absorber layer to obtained
highest reflectivity from a periodic ML with number of layer pairs N
(Reproduced from Ref. 3).
Chapter 5
118
The factor 2 is because of two boundaries per period. At an angle away from the critical
angle,
4
1
q
R , where momentum transfer vector
sin 4
= q (Chapter 2). Therefore, the
minimum number of layer pairs can be written from eq.
n
(5.4) as

2
2
2
min
sin

q N (5.5)
Enhancement in reflectivity can only be realized when the absorption of the materials is
sufficiently small, such that x-rays can reach all the deposited layers. But, due to absorption
in the layers, the maximum number of periods that can be penetrated by x-ray to contribute in
reflectivity can be written as
3

2
sin
2
max
= N (5.6)
where is the effective average absorption index of the two materials in the ML. In the
limiting case, with a low absorption spacer material and if the layer of strong absorber
material is very thin, would approach the absorption index of the spacer material. The
number of layer pairs is plotted against reflectivity in Fig. 5.2, for different values
(different high-Z thickness keeping periodicity constant) in the MLS. Reflectivity increases
with N
2
and becomes saturated at different numbers of periods, depending on the absorption
in a period. At very short wavelengths, the
absorption decreases faster than the reflected
amplitude and N
max
increases to a very large
value. Similarly, at any particular
wavelength, as the angle of incidence
decreases, the N
max
get reduced. These
discussions can be translated into materials
selection rules to find the best materials
combinations for optimum performance at
different wavelength as discussed in Chapter
1 (section 1.4.2.2). For the best reflectivity
performance of a MLS, the ratio N
max
/N
min

should be as high as possible. This is
achieved with the following rules: (i) Spacer
Fig. 5.3 Theoretical calculated
reflectivity of ideal MLs (with N=50
and =0.4) from best material
combinations in the wavelength range
55 to 160 , at incident angle of 45
0

60 80 100 120 140 160
0
15
30
45
60
75
Mo/B
Mo/Sr
Mo/Be
Fe/B
4
C
Mo/B
4
C
Mo/Y
Mo/Si
R
e
f
l
e
c
t
i
v
i
t
y

(
%
)
Wavelength ()
Chapter 5
119
materials with lowest absorption to get largest value of N
max
, and (ii) Second materials with
largest possible reflection co-efficient at the boundary with the spacer to get a smaller value
of N
min
. The first rule is the most important for selection of materials for x-ray MLs. Low-Z
elements at the longer wavelength side of their K- or L-absorption edges are prime choices.
The second rule will usually give a larger choice of materials with similar performance. The
calculated theoretical reflectivity from ideal ML at Brewsters angle (45
0
) for the best
material pairs in the wavelength range from 55 to 160 is shown in Fig. 5.3. It may be noted
that the absorption edge of the low-Z element in this wavelength range is as follows: boron-K
(66 ), yttrium-M (80 ), beryllium-K (111 ) and silicon-L (124 ). It is clear that low-Z
elements at the longer wavelength side of their K-, L- and M-absorption edges give the best
performance in reflectivity when Mo is the high-Z material, as shown in Fig. 5.3. Although
Mo/Be and Mo/Sr MLs have high theoretical reflectivity, their practical application is
difficult due to the health hazard of Be and Sr. Above the boron K-edge, boron has low
absorption, and boron-based MLs have high reflectivities. However, MLs with boron are less
stable
3
. Since B
4
C offers very stable interfaces, B
4
C is generally used instead of boron,
although B
4
C-based MLs have a lower theoretical reflectivity than boron-based MLs
12
.
Therefore, it is clear from Fig. 5.3 that, for practical reasons, the best material combinations
for ML in the wavelength range 67 to 80, 81 to 124 , and 125 to 160 are Fe/ B
4
C,
Mo/Y, and Mo/Si, respectively. Therefore, this thesis focuses on a study of the surfaces and
interfaces of these three systems, i.e., Fe/ B
4
C, Mo/Si and Mo/Y, for polarizer application in
the wavelength range from 67 to 160 on the Indus-1 synchrotron source. In the remaining
sections, we present a detailed study of the Mo/Si system. This is followed by our study of
the Fe/ B
4
C system, and then the Mo/Y system.
Chapter 5
120
5.3 Mo/Si Multilayer Structures

Mo/Si MLs are efficient mirrors
13,14
in the soft x-ray region (125-300), important to
technological applications such as lithography, astronomy, x-ray microscopy and
spectroscopy
14, 15-17
. Mo/Si MLs are also used in soft x-ray polarimeters
18-20
. As noted earlier,
the nature of the interfaces plays an important role in achieving optimum performance of
MLs for x-ray optics. Central to these issues is an understanding of the structure and growth
of the layers and interfaces. Therefore, understanding the growth mechanism that affects the
microscopic interface structure is very important to improving this technology. The
formation of molybdenum silicides is also important to the semiconductor community due to
their potential application in the VLSI technology, viz., dual-gate CMOS circuits
21
and self-
aligned silicide stacks
22
. The Mo/Si system is also attractive to researchers due to its
superconducting properties
23, 24
. Nakajima et al.
23
have attributed superconductivity in Mo/Si
MLs to the amorphous Mo-Si phase formed in the interfacial region of the layers. For all
these reasons, it is important to investigate the nature of interfaces in Mo/Si MLs.
In their binary phase diagrams with silicon, all transition metals have a few silicide
phases
25
. This implies high chemical reactivity between the elements at the interfaces, even
at room temperature. This results in the automatic formation of an interfacial compound layer
at the interfaces, due to the solid state amorphization reaction at metal/Si interfaces, such as
in Ir/Si, Co/Si and Ti/Si
26-28
. A characteristic feature of the Mo/Si ML is that, the interlayer
region is asymmetric. The Mo-on-Si interface is thicker than the Si-on-Mo interface. This
asymmetry, was first observed by Petford-Long et al.
29
, is an intrinsic property of Mo/Si
MLs, and has been observed in MLs deposited by sputtering
29, 30
and evaporation methods
31,
32
. Various research groups disagree over the sharpness, reactivity, and the growth mode of
Mo/Si interfaces. For example: in evaporated Mo/Si MLs, Stearn et al.
31
have observed an
Mo-on-Si interface of (153 ) and Si-on-Mo interface of (63 ), using high resolution
TEM, and suggested that the interlayers form due to penetration and mixing at interfaces;
Yakshin et al.
33
, using small angle x-ray scattering, have observed an Mo-on-Si interface of 8
and Si-on-Mo interface of 8 , and suggested that the interlayer comprises Mo
5
Si
3
+MoSi
2
;
Modi et al.
34
have observed and Mo-on-Si interface of 10 and Si-on-Mo interface of 8 ,
using soft-XRR, and suggested that the interlayers are composed of MoSi
2
formed by a
thermally activated process. Recently, Kesseis et al.
35
have used an analytical method in
which they calibrate the intensity scale of cross-sectional TEM using XRR, to determine that
Chapter 5
121
the Mo-on-Si interlayer thickness is 12 and that the Si-on-Mo interlayer thickness is 11 .
They attempted also to determine the composition of interlayer using this method, suggesting
that the Si-on-Mo interface comprises MoSi
2
, whereas the Mo-on-Si interface comprises
Mo
3
Si
5
. Thus, the final word on the interlayer asymmetry and composition in Mo/Si MLs
still remains elusive.
The microstructural parameters of the interlayers, such as interlayer thickness and
interface roughness, are generally determined using non-destructive methods such as hard
XRR, and destructive methods, viz., cross-sectional TEM. The observation of a minor
asymmetry (~2 ) at the two interfaces using TEM may be difficult due to poor image
contrast between the interlayer and the pure elements. Moreover, artifacts may be introduced
during the TEM sample preparation. The XRR technique provides a non-destructive
characterization of the internal interfaces of thin film multilayers. Specular x-ray reflectivity
at glancing angles of incidence has been utilized effectively to characterize the interface
morphology on the atomic scale. XRR yields a density profile perpendicular to the sample
surface, as given by Parratt
36
and modified by Novet and Croce
37
for real ML structures.
Therefore, specular XRR gives depth-graded information about the interlayer (i.e., interlayer
thickness, roughness, and density) present at the interfaces of ML structures. Some authors
have also used XRR method for Mo/Si data fitting the using the tri-layer
32
and four-layer
38-40

models to fit experimental data, without detailed studies of the effect of nature of interlayer
on the reflectivity profile and the effect of roughness on small interlayer asymmetry. Kim et
al.
39
have reported, using the four-layer model, of a small asymmetry in thickness of 2.6
(Mo-on-Si interlayer 4.6 and Si-on-Mo 2 ), when the roughness is 6 (greater than
interlayer width). This seems to be unphysical, as the interlayer thickness is smaller than the
interface roughness. The effect of the nature of the interlayers on specular XRR, and a
comparison with measured data have not been reported.
The compositions of the interlayers formed at the interfaces of MLs are generally
identified using either structural analysis (electron and x-ray diffraction), or chemical
composition analysis (x-ray photoelectron spectroscopy, Auger electron spectroscopy,
ultraviolet photoelectron spectroscopy and secondary ion mass spectroscopy etc.). The
composition at interfaces can also be determined using the soft x-ray resonant reflectivity
method
41
, as will be detailed in Chapter 6. The diffraction methods have considerable
limitations if the interface compound is amorphous and very thin (~ 1 nm). Moreover,
Chapter 5
122
artifacts may be introduced during the TEM sample preparation. For example, although TEM
gives a good picture of the interlayer, it is not sensitive to phase composition, because of the
small volume examined
42
. On the other hand, chemical composition methods probe the
chemistry of the surfaces, and hence give information about changes in the chemical
environment due to interaction at the interfaces, even if the compound is amorphous and very
thin (~ 1 nm). Kessels et al.
35
have suggested that depth profiling in XPS, AES, and SIMS
may induce changes in the original profile. However, using SRIM
43
2000, we have observed
that there is no substantial change in the original profile at an incidence ion angle of 45
0
with
low ion energy of 2.5 kV. Therefore, depth profile x-ray photoelectron (XPS) method has
been adopted in this work for interlayer composition analysis. The results obtained from XPS
agree well with interlayer composition determined by soft x-ray resonant reflectivity
presented in Chapter 6. Depth- profiled XPS is considered an effective analytical technique
44

for evaluating the composition and the chemical state of materials at interfaces. Depth-
profiled XPS has been used previously for composition analysis at different interfaces such
as: La
2
O
3
/Si interfaces
45
, WN
x
/Si interfaces
46
and Pt/Si interfaces
47
. Very few reports are
available in the literature on the analysis of interactions at Mo/Si interfaces using chemical
methods. Nguyen et al.
48
have studied the interface as a function of silicon coverage on
polycrystalline Mo substrate using UPS and XPS. They have observed negligible shift of the
Si 2p energy in the monolayer range, the shift becoming ~0.3 eV in a thick deposit. Slaughter
et al.
32
used Auger depth-profiling of sputtered-deposited Mo/Si MLs. They have observed
periodic variation of Mo and Si concentration in the Auger depth profile, and the presence of
impurities in MLs. There is no quantitative report on the nature of the interlayer at the two
types of interfaces in Mo/Si MLs.
In this work, a detailed study of the surfaces and interfaces in Mo/Si MLs has been
conducted using hard XRR and XPS measurements, and the performance of the MLs for
polarizers tested using the Indus-1 synchrotron source at Brewsters angle. The experimental
data of hard XRR are fitted using a four-layer model to take into account of the interlayers
formed at the interfaces. The model clearly distinguishes the effect of interfacial roughness
and of the sub-nanometer interlayer on the x-ray reflectivity profile. The four-layer model is
able to determine the atomic scale asymmetry of interlayer thickness at the interfaces viz.,
Mo-on-Si and Si-on-Mo. The phase composition of the interlayers is identified using depth-
profiled XPS. The experimental results are explained using thermodynamical driving force
Chapter 5
123
(TDF) and effective heat of formation (EHF) models. Finally, the actual performance of the
MLs as polarizing elements is tested using the reflectometry beam line of Indus-1.

5.3.1 Experimental

Mo/Si MLs were deposited on float glass substrates using the indigenously developed
ultra-high vacuum electron beam evaporation system (described in Chapter 3). To avoid
contamination during deposition, the system was baked at 180
o
C for 12 hours, to achieve a
base pressure of ~ 2 10
-9
mbar. High quality float glass and Si(100) were used as substrates
for the deposition of MLs, after appropriate cleaning. The rms roughness of these substrates
was ~4 , as measured using XRR. A deposition rate of ~6 /min for both Mo and Si was
maintained using a quartz crystal microbalance. The purity of Si and Mo used as evaporation
sources was 99.999% and 99.95%, respectively. Before the fabrication of Mo/Si MLs, single
layers of Mo and Si, and Mo-Si bi-layers with varying thicknesses were deposited to
calibrate deposited thicknesses, layer density, and the nature of the interfaces. Some of the
XRR results of these single and bi-layer films have been discussed in Chapter 3. The MLs
were fabricated with a varying number of layer pairs (N) from 4 to 30, to study the effect on
roughness of the number of layer pairs. The values were also varied from 0.2 to 0.4 to
maximize the reflectivity. The ML periodicity was varied from 88 to 113 for
polarization application in the wavelength range 125 to 160 . Hard XRR reflectivity
measurements were carried out on the MLs in a reflectometer
49
using Cu K
radiation. A
schematic diagram (Fig. 2.19) of the hard XRR experimental set up has been given in
Chapter 2. The XRR data were fitted using by Parratts formalism
36
as modified by Novet
and Croce
37
for real ML structures. The XRR data analysis method has been detailed in
Chapter 2. XPS analysis was carried out using a PES workstation coupled to a toroidal
grating monochromator beam line installed on the Indus-1 SR source
50
. This workstation is
equipped with an OMICRON 180
0
hemispherical analyzer (model EA 125) along with a twin
anode x-ray sources (Al K
and Mg K
). The concentric hemispherical analyzer with has a

pass energy of 50 eV. The overall resolution, including the analyzer, was estimated to be ~
0.8 eV. Photoelectron take-off angle was 45
0
. Measurements were carried out at a base
pressure < 610
-10
mbar. The energy scale was calibrated using Fermi level and peak position
from the system database, with Au 4f
7/2
at 84.00.1 eV serving as the external reference.
Since the glass substrate is non-conducting, charging effect was observed. To correct the
Chapter 5
124
shifts in binding energy of core levels due to charging effect, graphitic C 1s at 284.7 eV was
used as internal reference. Core level spectra were curve-fitted and analyzed after Shirley
background subtraction. For the depth-profiling of the MLs, a2.5 kV argon ion beam at and
angle of 45
0
with the normal to the sample surface was used. The approximate sputtering
rates, 2 /min for Si and 1.5 /min for Mo, were determined from separate experiments
carried out on single layer thin films of Si and Mo with known thickness. The performance of
the MLs was tested at Brewsters angle in the 115 to 160 wavelength range, using the
CAT-TGM reflectivity beam line on the Indus- SR source
51
.

5.3.2 Results of Hard XRR Measurements and Discussion

The typical angle-dependent reflectivity spectra of Mo/Si MLs with periodicity 100
and = 0.29 for N=4, 10, and 20 are shown in Fig. 5.4. The experimental data were fitted by
a two-layer model using Parratts formalism
36
. The ML structures show Bragg peaks up to
the 5
th
order, with distinct Kiessig oscillations between the successive Bragg peaks (clearly
visible for ML with a small number of layer pairs), indicating the high quality of the MLS.
Fig. 5.4 Measured and fitted XRR spectra of Mo/Si multilayers with a 100 period
and = 0.29, for different N. (a) For N=4,
(b) N=10, and (c) N=20. For N=4, roughness of Si and Mo is 4 and 5,
respectively. For N=10, roughness of Si and Mo is 7 and 8, respectively. For
N=20, roughness of Si and Mo is 10 and 11 , respectively.
0.0 0.1 0.2 0.3 0.4
10
-8
10
-6
10
-4
10
-2
10
0
10
2
10
4
10
6
10
8
10
3
(c)
measured
fitted (two-layer model)
10
7
R
e
f
l
e
c
t
i
v
i
t
y
(a)
(b)
q
z
(
-1
)
Chapter 5
125
Fig. 5.4 also shows the comparison of the intensities of the Bragg peaks between
experimental data and fitted values, taking into account the reduction of the intensities due to
interfacial roughness in a statistical approach (two-layer model). The differences between
experimental and fitted results indicate the presence of interlayers formed by interdiffusion at
the interfaces in Mo/Si MLs. In the subsequent sub-sections, we demonstrate the effect of the
interlayers and their asymmetry on the reflectivity profile, through simulations using Parratts
formalism
36
, which incorporates a four-layer model with interlayers.

5.3.2.1 Modeling and Simulation

The interfaces of any real ML system are imperfect due to roughness, interdiffusion
and chemical reactivity of materials. The imperfect interfaces can be dealt either by a
statistical approach (two-layer model) or multiple layer models with the incorporation of
interlayers between the two-layer systems. In the statistical approach: the reflectance from a
ML structure with N layers can be calculated using a recursion formalism
36
. For an ideal ML
with sharp interfaces, the composition is defined by a -ratio. But when a compound
material is formed at the interface, the interlayer model gives a better fitting to the reflected
profile than the statistical approach. The interlayer is due to interdiffusion at the boundaries.
Fig.5.5 shows a four-layer model, which incorporates the interlayer between pure Si and Mo.
Note that the period of the ML is
2 21 12 1
d d d d d + + + = . The thickness of the interlayers
12
d
and
21
d may not be the same, and such a structure can not be described by the -ratio used
for the statistical approach. The effective -ratio
33
to take into account the formation of
interlayer with thickness
12
d and
21
d and corresponding refractive indices
12
and
21
is:

( )
d
w d w d d
eff
21 21 12 12 1
+ +
= (5.7)
where the weight factor w for compounds at different boundaries is defined as

2 1
2 12
12

= w ,
2 1
2 21
21

= w (5.8)
Chapter 5
126
1
and
2
are the refractive indices of the high-Z and low-Z material, respectively. In the
case of Mo/Si MLs, the four layers consist of the Mo and Si layers and the two silicide
interlayers. The structure of the coating can be written as [Si/Mo
x1
Si
y1
/Mo/ Mo
x2
Si
y2
]
N
+ Si +
SiO
2
. Here, N is the number of periods, and the last two layers correspond to a partially
oxidized top layer of Si. The interlayer composition is taken as MoSi
2
.

5.3.2.2 Influence of Interlayer

The significant influence of the interlayer on the reflectivity profile, compared with
statistical interfacial roughness, is shown in Fig.5.6, which is simulated at the Cu K

wavelength (=1.54 ). For simulations, a small number of layer pair was chosen to show
the clear influence of Kiessig oscillations on full the reflectivity profile. Fig. 5.6 (a) shows
the effect of interface roughness on the reflectivity profile, assuming zero interlayer thickness
(two-layer model). As the interface roughness increases, the reflectivity diminishes, without
affecting the distribution of the individual peaks. Fig. 5.6 (b) shows for zero interface
roughness the effect of interlayer thickness on reflectivity profile. As the interlayer thickness
increases, the reflectivity of Bragg peaks decreases, while the peak positions are shifted due
to the change in the refraction correction term. The figure also indicates that the decline in
the intensity of the higher order Bragg peaks due to increasing interlayer thickness is greater
than the decline due to interface roughness. This is because of the decrease in refractive
index contrast due to the formation of interlayers. It is to be noted here that the interlayer
leads to redistribution of the reflectivity curve, which is not the case when no interlayer is
Fig. 5.5 A schematic diagram of a real multilayer structure with interlayers. The
four-layer model shows imperfect boundaries, and the interlayers between two pure
Mo and Si layers.
Chapter 5
127
present. Therefore, Fig.5.6 provides clear information about the difference in the effect of
interfacial roughness and the sub-nanometer interlayer on the reflectivity profile.

For simulations, a small number of layer pair was chosen to show the clear influence of
Kiessig oscillations on full the reflectivity profile. Fig. 5.6 (a) shows the effect of interface
roughness on the reflectivity profile, assuming zero interlayer thickness (two-layer model).
As the interface roughness increases, the reflectivity diminishes, without affecting the
distribution of the individual peaks. Fig. 5.6 (b) shows for zero interface roughness the effect
of interlayer thickness on reflectivity profile. As the interlayer thickness increases, the
reflectivity of Bragg peaks decreases, while the peak positions are shifted due to the change
in the refraction correction term. The figure also indicates that the decline in the intensity of
the higher order Bragg peaks due to increasing interlayer thickness is greater than the decline
due to interface roughness. This is because of the decrease in refractive index contrast due to
the formation of interlayers. It is to be noted here that the interlayer leads to redistribution of
the reflectivity curve, which is not the case when no interlayer is present. Therefore, Fig.5.6
provides clear information about the difference in the effect of interfacial roughness and the
sub-nanometer interlayer on the reflectivity profile.

Fig. 5.6 Simulated spectra of Mo/Si ML for number of layer pairs N=4 with a
period of 100 and = 0.29, at =1.54 . (a) Effect of roughness on reflectivity
profile with zero interlayer thickness. (b) Effect of interlayers on reflectivity profile
with zero roughness.
0.0 0.1 0.2 0.3 0.4 0.5
10
-9
10
-6
10
-3
10
0
10
-9
10
-6
10
-3
10
0
(b)
With roughness= 0

R
e
f
l
e
c
t
i
v
i
t
y
Interlayer thickness
0
10

q
z
(
-1
)

(a) with interlayer thickness=0
Roughness
0
10
Chapter 5
128

A comparative simulation of the effect of roughness and interlayers on the higher order
Bragg peak reflectivity, peak position, and peak width is shown in Fig. 5.7. It is clear that the
effect of interlayers on higher order Bragg peak reflectivity, peak position, and peak width is
more significant than the effect of roughness. This is because the higher order Bragg peaks
appear at higher q
z
values.

5.3.2.3 Effect of Asymmetry

An intrinsic property of the as-deposited Mo/Si ML is that the formation of the
interlayer, Mo-on-Si and Si-on-Mo, is asymmetric. Fig. 5.8 shows the effect of atomic- scale
asymmetry in interlayer thickness for different roughness values. Fig. 5.8 (a) indicates that a
small interlayer thickness asymmetry (~2) is clearly identified in MLs with small
roughness. The effect is greater at larger q
z
values. As interfacial roughness increases from
Fig. 5.8(a) to Fig. 5.8(c), the sensitivity of interlayer asymmetry decreases. We have
observed an ambiguity in the asymmetry of interlayer for ML if interfacial roughness is
comparable to the interlayer thickness, as shown in Fig. 5.8(c). In this case, simulation gives
Fig. 5.7 Simulation of the effect of roughness and of interlayers on higher order
Bragg peaks for MLs with N=4 (a) Effect on peak reflectivity. (b) Effect on peak
position. (c) Effect on peak width.
4 8 12 16
0.4
0.8
1.2
1.6
3 6 9
0.032
0.040
0.048
0.056
0 3 6 9
32
40
48
0.3
0.6
8
10
12
14
3rd order
2nd order
1st order
(b)

p
e
a
k

p
o
s
i
t
i
o
n

(
d
e
g
r
e
e
)
Roughness ()
3rd order
2nd order
1st order
(c)
interlayer thickness ()

p
e
a
k

w
i
d
t
h

(
d
e
g
r
e
e
)
Roughness ()
Roughness () P
e
a
k

r
e
f
l
e
c
t
i
v
i
t
y

P
e
a
k

r
e
f
l
e
c
t
i
v
i
t
y
3rd order
2nd order

roughness
interlayer
(a)
1st order
Chapter 5
129
almost the same results when the two interlayer thicknesses are interchanged. This has also
been observed experimentally and is discussed in the next section.

5.3.2.4 Application of Four-layer Model to Mo/Si MLs

Fig. 5.9 shows the experimental data fitted, with the incorporation of interlayers, and using the
Fig. 5.8 Simulation spectra of the effect of asymmetric interlayer thickness on
reflectivity profile of MLs with N=4, but of different roughness. (a) For
roughness=0 . (b) For roughness=5 and (c) For roughness=10 .
10
-7
10
-4
10
-1
0.0 0.1 0.2 0.3 0.4 0.5
10
-7
10
-4
10
-1
10
-7
10
-4
10
-1
(b)

Roughness=5
t
Mo-on-Si
=10, t
Si-on-Mo
=8
t
Si-on-Mo
=10, t
Mo-on-Si
=8
(c)
q
z
(
-1
)

Roughness=10
R
e
f
l
e
c
t
i
v
i
t
y
(a)

Roughness=0
Fig. 5.9 Measured reflectivity data of fig.5.4 for Mo/Si MLs are re-plotted, along
with four-layer model fitting. (a) For N=4 (b) For N=10 and (c) For N=20. The
inset shows the distinct Kiessig oscillations up to the 2
nd
Bragg peak
0.00 0.06 0.12 0.18 0.24 0.30 0.36 0.42
10
-8
10
-6
10
-4
10
-2
10
0
10
2
10
4
10
6
10
8
0.05 0.10
10
-8
10
-6
10
-4
10
-2
10
0
10
2
10
4
10
6
10
8

(c)

R
e
f
l
e
c
t
i
v
i
t
y
(a)
(b)
q
z
(
-1
)
(c)
10
3
10
7
Measured
Fitted with interlayer
R
e
f
l
e
c
t
i
v
i
t
y
(a)
(b)
q
z
(
-1
)
Chapter 5
130
four-layer model. In the data fitting, the interlayer composition ratio used is the Mo:Si ratio
of 1:2 obtained from XPS results, corresponding to MoSi
2
. Using the four-layer model, the
experimental data could be best-fitted up to all the higher order Bragg peaks. This validates
the four-layer model, indicating real structure in the ML stacks. The inset of Fig. 5.9 clearly
shows the distinct Kiessig oscillations up to the 2
nd
Bragg peak, indicating well-defined ML
structure with good control over deposited thicknesses. A greater decline in the intensity of
the higher order peaks, than given by the simulation of the two-layer model, occurs because
of the poor refractive index contrast at the interfaces attributable to the formation of
interlayers. The values of the individual layer thickness, interface roughness, interlayer
thickness, and optical constant, deduced from the best fit are tabulated in Table 5.1. The
measured thickness of the Si and Mo layers are expected to be smaller than the nominal
values (quartz crystal microbalance), as the measurement includes the formation of
interlayers at the interfaces during deposition. The Mo-on-Si interlayer thickness is 100.5
, whereas the Si-on-Mo interlayer thickness is 80.5 .

No. of
layer pair
Layer Thickness
()
Roughness
()
Indices
10
-6

Absorption
*
10
-6

4

Si
Si-on-Mo
Mo
Mo on Si
580.5
80.5
240.5
100.5
4
4
5
4
6.3 [7.59]
19.3 [20.94]
28.3 [28.8]
19.3 [20.94]
0.15 [0.17]
1.18 [1.25]
1.4 [1.88]
1.18 [1.25]
10 Si
Si-on-Mo
Mo
Mo on Si
580.5
80.5
240.5
100.5
7
7
8
7
6.2 [7.59]
19.4 [20.94]
28.6 [28.8]
19.4 [20.94]
0.16 [0.17]
1.17 [1.25]
1.6 [1.88]
1.17 [1.25]
20 Si
Si-on-Mo
Mo
Mo on Si
580.5
80.5
240.5
100.5
10
10
11
10
6.4 [7.59]
19.5 [20.94]
28.1 [28.8]
19.5 [20.94]
0.15 [0.17]
1.15 [1.25]
1.5 [1.88]
1.15 [1.25]

In Fig. 5.10 (a), the scattering length density is plotted as a function of depth of the ML for
N=4, as obtained from the fitting of experimental data. The distance between the centers of
two successive maxima (pure Mo) or two successive minima (pure Si) corresponds to the
Table 5.1 XRR experimental results for Mo/Si MLs (Fig. 5.9) best-fitted with the four-
layer model. In square brackets are the values of optical constants (indices and
absorption) from Henke et al.
52
Chapter 5
131
period of the ML. In Fig. 5.10 (b), dashed line corresponds to the observed interface region
arising from the combination of silicide interlayers and layer roughness, whereas the dotted
line represents the calculated profile without the Gaussian roughness. The interlayer regions
are asymmetric, as marked in Fig. 5.10 (b). Further, Fig. 5.10 (b) indicates the presence of
the native Si oxide (~10) at the top of the ML.

Table 5.1 shows that the Si-on-Mo interlayer thickness is 8 0.5 , whereas the Mo-on-Si is
100.5 . The error may be due to the uncertainty in the estimation of the weight factor
for the composition of the compound at the boundaries. It is important to note that the Mo-
on-Si interlayer thickness is greater than that of Si-on-Mo, and is clearly observed for N=4
and N=10 in Fig. 5.9 (a) and Fig.5.9 (b), respectively. For N=20, Fig. 5.9 (c), we have
observed an ambiguity in the asymmetry of the interlayer because our experimentally fitted
values give almost identical results when the two interlayer values are interchanged. This is
0 15 30 45 60 75 90 105 120 135
2.0x10
-5
4.0x10
-5
6.0x10
-5
8.0x10
-5
(a)

Depth in
SiO
x
Si
Si-on-Mo
8
Mo
Mo-on-Si
10
(b)
0 100 200 300 400 500
0.0
3.0x10
-5
6.0x10
-5
9.0x10
-5
S
c
a
t
t
e
r
i
n
g

l
e
n
g
t
h

d
e
n
s
i
t
y

(
r
h
o
/
2
)
depth in
100
FIG. 5.10 Scattering length density profile obtained after fitting experimental data
using interlayers. (a) Maxima and minima correspond to layers of pure Mo and Si,
respectively. In between is the interface region arising from the formation of silicide
interlayers and layer roughness. (b) The dashed line represents actual profile used
for fitting, assuming interlayers with Gaussian roughness. The dotted line represents
the profile without Gaussian roughness.
Chapter 5
132
because, for N=20, interfacial roughness is comparable to the interlayer thickness, as shown
in Table 5.1. Therefore, when the roughness is comparable to interlayer thickness, the effect
of roughness dominates, making it difficult to observe the small asymmetry (~2 ) in the
interlayer thickness, consistent with the simulation results shown in Fig.5.8. Using the four-
layer model, Kim et al.
39
reported that Mo-on-Si interlayer thickness is 4.6 , whereas that of
Si-on-Mo is of 2 (asymmetry of 2.6 ) with the roughness of 6 (greater than interlayer
thickness). As per their best-fit XRR figure, the fitted data do not reflect the real structure in
their MLs. The authors might have used interlayer thickness values in XRR fitting from their
TEM results without any detailed fitting. Hence, we conclude that even a small asymmetry
(~2 ) in interlayer thickness can be discerned using specular XRR, when the roughness is
smaller than the thickness of the interlayer, as per our simulations and experimental results.
Our value of interlayer thickness agrees well with that observed previously
30
. However, our
results for interlayer thickness differ significantly with those from TEM and XPS
29, 53,54
.
This may be due the characterization technique being different. In TEM, the error in the
atomic scale asymmetry of interlayer thickness may be due to poor image contrast between
the interlayers and the pure elements whereas, in XPS, the error is due to poor thickness
resolution during depth profiling. The values of optical constants (indices and absorption) for
Si, Mo, and the MoSi
2
interlayers obtained from XRR best fit, given in Table 5.1, are slightly
different from the values from Henke et al.
52
, which are in square brackets. The difference is
attributable to density change.
Different authors suggest different possible mechanisms of asymmetry in as-
deposited Mo/Si MLs. Some authors
29, 30, 38
have proposed a kinetic model (different
momenta of Si and Mo) for the formation of interlayers during the deposition of Mo/Si MLs.
Windit et al.
55
proposed the initial formation of an amorphous interlayer due to Si diffusion
processes, in which the Si atoms tend to migrate, even as the Mo atoms are deposited.
Slaughter et al.
32
estimated the activation energy, using the Arrhenius analysis of variation of
silicide thickness obtained from TEM data. From the estimated activation energy, they
suggested a surface diffusion process rather than bulk diffusion process. On the other hand,
Stearns et al.
53
have suggested that Mo atoms get easily embedded into relatively open and
disordered amorphous Si than Si does into the more closely packed crystalline Mo lattice.
The same authors

proposed
31
later that interlayer thickness variation with temperature is
consistent with a process that is at least partly thermally activated. By considering the
Chapter 5
133
different thermal conductivities of Si and Mo, Liwen et al.
55
have proposed that the
asymmetry in sputter-deposited Mo/Si MLs is due to thermally activated processes. Attempts
have also been made to explain the asymmetry using molecular beam dynamics
56
, concluding
that the different degrees of penetration and interdiffusion of the adatoms during deposition
is the cause of asymmetry. If the kinetic model plays a role in the formation of interlayers at
the interfaces, then, evaporated MLs would not show interlayers. This is because, the
energies of evaporated and sputtered atoms are of ~0.2 eV and ~10 eV respectively
29, 31
,
whereas the cohesion energies of amorphous Si and Mo are ~3 eV and ~7ev, respectively
57
,
which are greater than energy of the evaporated atoms. In fact, interlayer formation in as-
deposited Mo/Si MLs is independent of the deposition process. Therefore, we feel the kinetic
model is not valid for the formation of interlayers in as-deposited Mo/Si MLs. Silicon is the
predominant diffusant in Mo-Si binary system
25
. In the Mo/Si system, an intensive atomic
intermixing takes place at the boundary during deposition because of the low activation
energy of surface diffusion for Si and the local temperature effect. The asymmetry in the
interlayer thickness is due to difference in the heat of sublimation of Mo and Si. This affects
the diffusion of Si in Si-on-Mo and Mo-on-Si cases differently. The initial intermixing would
be a thermally activated process by the negative thermodynamic heat of mixing, because any
process involving condensation from the vapor will release the latent heat. The heats of
sublimation
58
of Mo and Si are 664.5 kJ g-atom
-1
and 450.1 kJ g-atom
-1
, respectively. The
thermal conductivity of Mo being higher than that of Si, the heat produced on the Si surface
by the Mo adatoms will be larger, diffusing slowly due to the lower thermal conductivity,
than when depositing Si on to the Mo surface. This may lead to a higher local temperature in
the Mo-on-Si case, with a greater probability of surface diffusion of Si. When the interlayer
thickness reaches 100.5, it becomes a barrier for further surface diffusion of Si. Similarly,
in the Si-on-Mo case, the local rise in temperature may be lower. As Si is the dominant
diffusant in Mo/Si system, after arrival of Si on the Mo surface, Si has to diffuse via bulk
diffusion into the Mo layer. As the co-efficient of bulk diffusion for Si is very low in the
low temperature range
59
and the local rise in temperature is small, there is a lower probability
of bulk diffusion and, hence, a thin interlayer of thickness 80.5 is formed at the Si-on-Mo
interface. When the interlayer thickness reaches 80.5 , it becomes a barrier for further bulk
diffusion of Si. Therefore, interlayer thickness is independent of number of layer pairs as
given in Table 5.1. Therefore, in the interface region, the supply of metal atoms is limited
Chapter 5
134
due to thermal effects that favor the transport of the Si atoms over Mo atoms. This, in turn,
energetically favors the Si-rich stoichiometry
25
.

Application of the Four-layer Model to a second set of Mo/Si MLs

A second set of Mo/Si MLs were fabricated with periodicity of 92 and the =0.32.
The number of layer pairs was varied from 5 to 30 with Si as the top layer. Fig.5.11 shows
the measured and fitted spectra using four-layer model at the Cu-K
wavelength.

The MLs clearly show higher order Bragg peaks. N-2 distinct Kiessig oscillations are clearly
discernible in the inset of Fig.5.11 for MLs with smaller number of layer pairs.
This indicates the high quality of ML structures. According to the extinction rule, in these
MLs (=0.32), the 3
rd
order Bragg peak (and multiples) should be diminished. The presence
of the 3
rd
order Bragg peak reveals the -ratio has been changed due to interlayer formation
at the interfaces, which we have taken into account in four-layer model fitting. The best-fit
data are shown in Table 5.2. In this fitting, we assume the MoSi
2
composition at the
0.0 0.1 0.2 0.3 0.4 0.5 0.6
10
-6
10
-4
10
-2
10
0
10
2
10
4
10
6
0.00 0.04 0.08
10
-1
10
1
10
3
10
5
(d)
(c)
(b)

R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
(a)
10
6
10
4
Experimental data
Fitted data
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
(a)
(b)
(c)
(d)
10
2
Fig. 5.11 XRR spectra of Mo/Si MLs with period 92 and = 0.32, at the Cu-
K
wavelength. (a) For N=5, (b) N=10, (c) N=20, and (d) N=30.
Chapter 5
135
interfaces in the four-layer model. The Si-on-Mo interlayer thickness is 80.5 , the Mo-on-
Si interlayer thickness being 100.5 . The interlayer thickness does not change with the
number of layer pairs. The measured optical constants ( and ) are slightly different from
those of Henke et al.
52
due to the difference in density.

No. of
layer
pair
Layer Thickness
()
Roughness
()
Indices
10
-6

Absorption
*
10
-6

5

Si
Si-on-Mo
Mo
Mo on Si
500.5
80.5
240.5
100.5
4
4
5
4

6.1
19.2
27.9
19.2

0.14
1.18
1.5
1.18

10 Si
Si-on-Mo
Mo
Mo on Si
500.5
80.5
240.5
100.5
5
5
6
5
6.3
19.1
28.1
19.1
0.13
1.16
1.6
1.16
20 Si
Si-on-Mo
Mo
Mo on Si
500.5
80.5
240.5
100.5
7
7
8
7
6.2
19.0
28.3
19.0
0.12
1.15
1.6
1.15
30 Si
Si-on-Mo
Mo
Mo on Si
500.5
80.5
240.5
100.5
9
9
10
9
6.1
19.1
28.1
19.1
0.15
1.18
1.3
1.18

In this case, too, the roughness of the ML samples increases with increasing number of layer
pairs, due to the presence of cumulative roughness, as discussed in Chapter 2. In the next
section, the determination of the composition of the interlayers using depth-graded XPS
measurements, and their correlation with theoretical calculations, is presented.

5.3.3 Results of Depth-graded XPS Measurements

In order to investigate the chemical nature of the interfaces, depth-profiled XPS
measurements were made using an Al K
source. Fig.5.12 shows the survey scans of Mo/Si

ML sample of N=20 for different sputtering times (depth). The virgin sample (no sputtering)
Table 5.2 The best fit for XRR experimental results (Fig. 5.11), employing the
four-layer model for Mo/Si MLs.
Chapter 5
136
indicates the presence of adsorbed oxygen and carbon on the top surface. In addition to this,
peaks due to Si 2p and Si 2s are also seen, due to Si on the top surface. The survey scans of
the sample sputtered for 15 min show weaker O 1s and C 1s peaks, Si 2p and Si 2s peaks,
and a low intensity Mo 3d peak was observed. As the sputtering time increases, the intensity
of Si 2p and Si 2s peaks decreases; simultaneously, the Mo 3d peak intensity increases, and
Mo 3p and 3s peaks appear. After 30 min sputtering, the Mo 3d, 3p, and 3s peaks reach the
maximum intensity, whereas Si 2p and 2s peaks reach minimum intensity. This suggests
crossing of the Si-on-Mo interface. As sputtering time increases further, intensities of Si 2p
and Si 2s peaks increase, while those of Mo 3d, Mo 3p and Mo 3s peaks decrease. For a
sputtering time of 50 min, Mo 3p and Mo 3s peaks vanish, the Mo 3d peak intensity reaches
minimum, and Si 2p and Si 2s peaks attain maximum intensity. This suggests crossing of the
Mo-on-Si interface. On further increasing the sputtering time, Mo 3d peak intensity increases
along with the appearance of Mo 3p and Mo 3s peaks, while that of Si 2p and Si 2s peaks
decreases. These observations thus reveal that both the interfaces have been probed.
Fig. 5.13 shows narrow scans corresponding to Si 2p and Mo 3d, recorded for
different sputtering times, together with the fitted spectra. To calibrate the chemical shift at
the interface, XPS spectra of Si and Mo thin films were measured and shown at the top of
Fig. 5.13. The fitted Si 2p spectrum (Fig. 5.13 (a)) corresponding to 0 min (virgin sample)
shows two peaks at binding energy of 99.19 eV (Si) and 102.3 eV (SiO
x
). As the sputtering
time increases, the SiO
x
peak disappears, indicating that the native oxide is formed only at
1400 1200 1000 800 600 400 200 0
0.0
2.0x10
3
4.0x10
3
6.0x10
3
8.0x10
3
1.0x10
4
1.2x10
4
1.4x10
4
C
Auger O
Auger
O 1S C 1S
M
o

3
S
M
o

3
P
M
o

3
d
S
i

2
s
S
i

2
P
8.2
7.4
6.6
5.8
5
4.2
3.4
2.6
I
n
t
e
n
s
i
t
y

(
A
r
b
.

U
n
i
t
)
Binding Energy (eV)
1.8
60 Min.
55 Min.
50 Min.
42 Min.
35 Min.
30 Min.
25 Min.
20 Min.
15 Min.
Virgin
Fig. 5.12 XPS survey scans of Mo (30 )/ Si(60 ) ML with N=20, for different
sputtering times. The spectrum intensity is multiplied by the number shown with x.
Chapter 5
137
the top Si surface. As the sputtering time increases further, the contribution of Si from MoSi
2

(B.E. 99.39 eV) increases, whereas the contribution from pure Si decreases. During fitting,
all possible Mo-Si compounds were taken into account. Thus, it can be seen from the
analysis of Fig. 5.13 that MoSi
2
is the main compound at the Mo/Si interfaces, with little
contribution of Mo
5
Si
3
. We have discussed the variation of MoSi
2
intensity with sputtering
time only because the other silicides are insignificant. For 30 min sputtering, contribution of
Si from MoSi
2
is a maximum and that of pure Si a minimum. At the first interface (Si-on-
Mo), the total peak shift is ~0.2 0.04 eV towards the higher binding energy w.r.t. pure Si
(given at top of Fig. 5.13 (a)). Again, as the sputtering time increases, contribution from pure
Si increases and that due to MoSi
2
decreases. For the sputtering time of 50 min, the
contribution of Si from pure Si reaches maximum, whereas the contribution of Si from MoSi
2

reaches a minimum. Therefore, at the second interface (Mo-on-Si), the shift for Si 2p is ~0.2
0.04 eV to the higher binding energy side. As sputtering time increases further, Si
contribution from pure Si decreases, whereas that from MoSi
2
increases. The Si 2p peak shift
remains the same for both the interfaces.
The fitted Mo 3d spectrum (Fig.5.13 (b)) of the virgin sample shows no signal from
Mo, as Si is the top surface. After 15 min of sputtering, peaks at binding energy 227.5 eV and
230.7 eV are observed. These peaks are assigned to Mo 3d
5/2
and Mo 3d
3/2
of MoSi
2
,
respectively. As the sputtering time increases further, the contribution of Mo from pure Mo
(binding energy 227.9 eV and 231.02 eV for Mo 3d
5/2
and Mo 3d
3/2
, respectively, shown at
the top of Fig.5.13(b)) increases, whereas the contribution from MoSi
2
decreases. At the first
interface (Si-on-Mo), the contribution of Mo from MoSi
2
is more than that of pure Mo. The
peak shift of Mo 3d
5/2
and Mo 3d
3/2
are ~0.4 0.04 eV towards lower binding energy, w.r.t.
the pure Mo peaks. Again, as the sputtering time increases further, Mo contribution from
pure Mo decreases and that of MoSi
2
increases. At the second interface (Mo-on-Si), the
contribution of Mo from pure Mo is minimum, whereas the contribution of Mo from MoSi
2

is maximum. The shift in binding energy is again ~0.4 0.04 eV, towards lower binding
energy. As sputtering time increases still further, Mo contribution from pure Mo increases
whereas that from MoSi
2
decreases. The Mo 3d peak shifts remains the same within
experimental error for both the interfaces.
Chapter 5
138

106 104 102 100 98 96 94

Binding Energy (eV)
SiO
x
Si 2p
Si Virgin
Si
20 min.
Mo
5
Si
3
MoSi
2

15 min.
MoSi
2
Mo
5
Si
3
Si

MoSi
2
Mo
5
Si
3
Si
25 min.

MoSi
2
Mo
5
Si
3
Si
30 min.

MoSi
2
Mo
5
Si
3
Si
35 min.

I
n
t
e
n
s
i
t
y

(
a
r
b
.

u
n
i
t
)

MoSi
2
Mo
5
Si
3
Si
42 min.

MoSi
2
Mo
5
Si
3
Si 50 min.

MoSi
2
Mo
5
Si
3
Si
55 min.

MoSi
2
Mo
5
Si
3
Si
60 min.

Si
Si thin film

236 234 232 230 228 226 224

Binding Energy (eV)
Virgin
3d
3/2

15 min.
20 min.
25 min.
30 min.
35 min.
42 min.
50 min.
55 min.
60 min.
3d
5/2
Mo
Mo
5
Si
3
MoSi
2

Mo
MoSi
2
Mo
5
Si
3

MoSi
2
Mo
5
Si
3
Mo

MoSi
2
Mo
5
Si
3
Mo

I
n
t
e
n
s
i
t
y

(
a
r
b
.

u
n
i
t
)
MoSi
2
Mo
5
Si
3

MoSi
2
Mo
5
Si
3
Mo
Mo

MoSi
2
Mo
5
Si
3
Mo

MoSi
2
Mo
5
Si
3
Mo

Mo

Mo thin film
Fig. 5.13 XPS narrow scans of Mo /Si ML for different sputtering times.
(a) Si 2p spectra and (b) Mo 3d spectra
(a) (b)
Chapter 5
139
The concentration profile as function of
sputtering time is shown in Fig. 5.14. For the
virgin sample (0 min sputtering), the oxygen
concentration is ~ 8 at.% and, at 15 min sputtering
time, it falls below the detection limit, suggesting
that interfaces are not contaminated with oxygen.
After 15 minutess of sputtering, Si concentration
is maximum, whereas the Mo concentration is
minimum. After 30 min sputtering time, the Si
concentration becomes minimum and Mo
concentration is maximum. This reveals crossing
the first interface region (Si-on-Mo), marked A
(Fig. 5.14). As sputtering time increases further,
Si concentration increases and Mo concentration
decreases. After 50 minutess of sputtering, Si
concentration becomes maximum and Mo concentration becomes minimum. This suggests
crossing the second interface (Mo-on-Si) marked B (Fig. 5.14). With further increase in
sputtering time, the Si concentration decreases and Mo concentration increases. Fig. 5.14
indicates qualitatively that the Si-on-Mo interlayer thickness (region A) is smaller than the
Mo-on-Si interlayer thickness (region B). The asymmetry in Mo-Si interface is in agreement
with previously reported results as discussed earlier, suggesting that the Si-on-Mo interface is
thinner than the Mo-on-Si interface.

5.3.3.1 Discussion

The Si 2p peak shifts ~0.2 0.04 eV towards higher binding energy w.r.t. pure Si,
whereas the Mo 3d peaks shift by ~0.4 0.04 eV towards lower binding energy with respect
to pure Mo, suggesting formation of the MoSi
2
phase
60
. These asymmetrical shifts are due to
unbalanced charge transfer during Mo-Si reaction and give rise to modification/shifts in the
core levels. In Fig. 5.13, the extent of peak shift is the same at both the interfaces within
experimental error, suggesting that the MoSi
2
is the major silicide formed at both the
interfaces. The amount of Mo and Si in the silicide was determined from the intensity of Mo
3d and Si 2p arising from Mo bonded with Si to be 32.2% and 67.8%, respectively, making
Fig. 5.14 Elemental concentration
depth profile of Mo/Si ML with N=20.
The Si-on-Mo interface region
(marked A) and Mo-on-Si interface
region (marked B) are shown.
0 10 20 30 40 50 60
0
10
20
30
40
50
60
70
80
90
100
Si
Mo
O
A
t
o
m
i
c

C
o
n
c
e
n
t
r
a
t
i
o
n

(
%
)
Sputtering Time (Min.)
A B
Chapter 5
140
for MoSi
2
. The formation of the MoSi
2
phase at the interfaces is supported by the measured
concentration profile shown in Fig. 5.14. At both the interfaces, even as the interface
thickness differs, the concentration of Si varies from 92 at.% to 67 at.%. The measured
concentrations favor the formation of MoSi
2
phase at 300K thermodynamically, as seen from
the binary Mo-Si phase diagram
58
(Fig. 5.15). We conclude that only the silicide MoSi
2
is
formed at both the interfaces.

The experimentally observed MoSi
2
compound is compared with the prediction of the
tendency for solid-state amorphizing reaction (SSAR) due to interaction at the interfaces in
Mo-Si MLs, based on the thermodynamic driving force (TDF) model
61, 62
. The first phase
formed at the interface may be predicted through the effective heat of formation (EHF)
model proposed by Pretorius et al.
63
.

SSAR and First Phase Formation at Mo/Si Interfaces:

To predict the reactions at the interfaces, a thermodynamic analysis of the intermixing
of Si and Mo was carried out, as solid-state reactions involve atomic intermixing and phase
transformation. According to Schwarz et al.
61
, the basic conditions of SSAR are: (i) lowering
of the free energy of the system due to a large negative heat of mixing, and (ii) one element
diffusing faster into another, so that atomic rearrangements required for the crystalline phase
are not possible, thus resulting in an amorphous phase. The Gibbs free energy change ( ) G
as the TDF in the Mo-Si ML may be written as
Fig. 5.15 The binary Mo-Si phase diagram
T
(
K
)
97.9
(Mo)
+
Mo
3
Si
Mo
5
Si
3
+
-MoSi
2
-MoSi
2

+
(Si)
26
53
-MiSi
2

+
liq.
Atomic concentration of Si/((Mo+Si) atomic concentration)
Mo
3
Si
Mo
5
Si
3

MoSi
2
LIQUID
Chapter 5
141

M c a
Mo Mo
c a
Si Si
G G c G c G + + =

(5.9)

where
Si
c and
Mo
c are molar fractions of Si and Mo, respectively.
c a
Si
G

and
c a
Mo
G

are the
free energy changes between the crystalline and amorphous phases of Si and Mo,
respectively.
M
G stands for the mixing free energy change between amorphous phases.
Assuming that the amorphous phase is an under-cooled liquid, the free energy change
between crystalline and amorphous phases can be written using Gong and Hentzells
method
62

c a c a c a
S T H G

= ( ) ( ) , 2 2 ln
p m m f m
C T T T H T T + = (5.10)

where
c a
H

and
c a
S

are the enthalpy and entropy changes between crystalline and
amorphous phases.
m
T is the melting temperature.
p
C is the heat capacity difference
between the two phases. The mixing free energy change between the amorphous phases in
Eq
n
. (5.9) is derived from the enthalpy of mixing
M
H based on Miedemas theory
64, 65

whereas the entropy change of mixing (
M
S ) is derived for the ideal solution:

M M M
S T H G =
( ) ( ) ( ) ( )( ) { } P R n P Q e P c c g c c f
WS o Mo Si
s
Mo
s
Si
+ =
2
3 / 1
2
*
, ,
( )
Mo Mo Si Si
c c c c KT ln ln + + . (5.11)

where, ( )
s
Mo
s
Si
c c f , is the symmetrical function of the surface concentrations of Si and Mo,
( )
Mo Si
c c g , the factor associated with the molar fraction and the volume of the Wigner-Seitz
cells of Si and Mo, e the electric charge,
*
the chemical potential difference of electrons,
and
WS
n the electronic density difference of Wigner-Seitz cell. P, Q
o
and R are constants.
The parameters in Miedemas model for calculating the heat of mixing are given in refs. 64
and 65. This model has previously been applied to various metal/silicon systems, for example
Ti/Si
66
and Zr/Si
67
. We have applied this model to the Mo/Si system.
Chapter 5
142
Fig. 5.16 shows the calculated free energies vs. molar fraction plot for the Mo/Si
binary systems at 300 K. The dashed line represents a mixture of Mo and Si without taking
interaction into account. The solid line represents Mo-Si compound formation due to solid-
state reaction between Mo and Si. The free energy is substantially lowered when Mo reacts
with Si (solid curve). The necessary condition for the formation of the Si-metal bond is,
0 <
M
H . This leads to a compound with a lower free energy than the mixture represented
without taking interaction into account. The mechanism involved in SSAR is a TDF of
intermixing due to a large heat of mixing between Mo and Si in the Mo-Si binary system. Si
is the predominant diffusant
25
in Mo/Si. Therefore, SSAR leads to amorphous compound (a-
compound) formation at the interface of Mo/Si MLs at 300 K. The two dotted lines are
common tangents between pure Si and a-compound and between pure Mo and a- compound
(Fig. 5.16). The whole composition range is divided into three characteristic regions A, B,
and C w.r.t. the two common tangents. In region B (between 75 at.% and 25 at.% of Si), only
0.0 0.2 0.4 0.6 0.8 1.0
-70
-60
-50
-40
-30
-20
-10
0

G
i
b
b
s

f
r
e
e

e
n
e
r
g
y

(
k
J
/
g
-
a
t
.
)
Molar fraction
Si Mo
Si + Mo
compound
At T=300K
Si
Mo
A B C
Fig. 5.16 Gibbs free energy diagram for the Mo-Si binary system calculated
at 300 K. The mixing of Si and Mo (without interaction) is represented by the
straight dashed line. The solid line represents a-compound formation due to solid-
state interaction between Mo and Si. The dotted line represents the common tangent
between Si and the a-compound and between Mo and the a-compound.
Chapter 5
143
the Mo-Si compound has a stable state. Outside this region, thermodynamics suggests that
the Mo-Si compound should coexist with unreacted Si (region A) and with unreacted Mo
(region C). We have observed that the concentration of Si at both the interfaces varies from
~92 at.% to 67 at.%. The interlayer thickness varies from the Si side to the Mo side (Fig.
5.14). These variations in Si concentration at the interfaces suggest a mixture of Si and the
Mo-Si compound in the region near the Si layer, and merely the Mo-Si compound near the
Mo layer. The phase of Mo-Si compound can be predicted by EHF model as follows.
Phase formation at the interface through SSAR is a non-equilibrium process. The first
phase formed at the interfaces is the one with the lowest EHF. Pretorius et al.
63
have defined
EHF as

|
|
|
.
|
\
|
=
element iting of ion concentrat compound
element iting of ion concentrat effective
H H
o
lim
lim
(5.12)
The effective concentration in the Mo/Si system comes out as Mo
0.021
Si
0.979
, which is the
lowest temperature eutectic concentration
58
. Based on this EHF concept, the EHFs of all
possible chemical compounds in equilibrium phase diagram (Fig.5.15) have been calculated,
in order to predict the first phase formed in the Mo/Si system. These are shown in Table 5.3.
For all the possible compounds in Mo/Si system, Mo is the limiting element. Table 5.3 shows
that MoSi
2
has the lowest EHF. Therefore, MoSi
2
is the first phase formed at the Mo/Si
interfaces. This supports our XPS results regarding the formation of the silicide MoSi
2
at
both the interfaces.

Table 5.3 Calculated effective heats of formation for the Mo/Si system.

Our results differ from those of Kessels et al.
35
. They have calibrated the intensity
scale of cross-sectional TEM using XRR and determined the absolute intensity-to-density
conversion-scale of cross-sectional TEM. Using this method, they reported that the MoSi
2

phase is formed at the Si-on-Mo interface and Mo
3
Si
5
at the Mo-on-Si interface. As per the
Lowest eutectic=Mo
0.021
Si
0.979

Phase Compound
Concentration
Limiting
element
H
0

(kJ/metal at.)
H

(kJ/metal at.)
Mo
3
Si
Mo
5
Si
3

MoSi
2

0.75/0.25
0.625/0.375
0.333/0.667
Mo
Mo
Mo
-33.472
-56.065
-108.784
-0.937
-1.882
-6.857
Chapter 5
144
Mo-Si binary phase diagram (Fig. 5.15), Mo
3
Si
5
does not exist or could be a metastable
phase. The only possible stable phases in Mo-Si system are Mo
3
Si, Mo
5
Si
3
, and MoSi
2
. Out
of these three phases, EHF model predicts only the MoSi
2
phase, which is
thermodynamically favorable at room temperature and is observed experimentally. Kessels et
al. have used electron beams with ion polishing for the preparation of their TEM samples. In
our case, we have not used ion polishing. Therefore, we feel that, our results do not agree
with those of Kessels et al. due to the different deposition process employed and/or the error
in their calibration procedure. Yakshin et al.
33
have determined phase composition at
interfaces of Mo/Si using hard x-ray reflectivity. Normally, hard x-ray reflectivity is not
sensitive to phase composition at interfaces
35
. Yakshin et al. have pointed out that the same
quality of fit could be obtained for slight differences in the compositions of phases. They
have also reported different phase compositions for different ion energies during ion beam
polishing. This indicates that the difference in phase composition reported Yakshin et al. and
by us may be due to different techniques of deposition and/or characterization.

5.3.4 ML Performance Assessment using Indus-1 Synchrotron Radiation

Multilayer mirrors reflecting near Brewster angle of 45 are useful as reflection
polarizers
19, 68, 69
. Using polarized light from
synchrotron sources (SR) a wide range of
phenomena can be studied in materials science,
physics, chemistry and biology
19, 70
. The basic
principle of polarization has been discussed in
Chapter 2, and the applications of ML polarizers
have been discussed in Chapter 1. For the
polarization experiments on the Indus-1
synchrotron source, we have fabricated Mo/Si-
based reflection polarizing elements with
desired periodicity to operate in the wavelength
range 125 to 160. The reflectivity
performance in the s-polarization geometry was
tested using the Indus-1 reflectometry beamline,
which is discussed later.
Fig. 5.17 The wavelength scan
measurements of Mo(30)/Si(62)
MLs with different N using Indus 1
SR at incidence angle 45
0

120 125 130 135 140 145 150
0
5
10
15
20
25
30
35
40
45
N=10
N=20
N30
%

o
f

R
e
f
l
e
c
t
i
v
i
t
y
Wavelength ()
Chapter 5
145
The reflectometry beam line employs a toroidal grating monochromator for high flux
and moderate spectral resolution
71
. Further details of the beamline and the experimental
station are given in Chapter 6. The wavelength-dependent reflectivity was measured with a
XUV Si photodiode detector. A typical wavelength scan with sets of MLs with the Mo
(30)/Si (62 ) composition and N=10, 20, and 30 is shown in Fig.5.17. Here, the sample is
kept fixed at Brewsters angle (45
0
) and the wavelength is scanned continuously from 120
to 150 . As the ML period is 92 , the center of the Bragg peak measured is at 130. As N
increases, the reflectivity increases. The increase in reflectivity from N=20 to 30 is much
smaller than that from N=10 to 20, because of absorption of soft x-rays in the layers (Fig.
5.2). Since roughness increases as the number of layer increases (Table 5.2), and also due to
absorption in the layer, therefore, we terminated at N=30. We achieved 45% reflectivity at
130 with a spectral bandwidth 2.9 from the N=30 MLS. As N (the number of layer
pairs) decreases, the full-width at half maximum increases.
The Bragg peak reflectivity from MLs with N=30 was measured at different
wavelengths from 115 to 160 , using the
Indus-1 SR, as shown in Fig. 5.18. The
measured peak reflectivity decreases
drastically below the Si L-absorption edge
(~124 ) due to absorption of the radiation in
the Si layer of the MLS. The peak reflectivity
is maximum at wavelengths just above the Si
edge and then decreases slowly as the
wavelength increases. The reflectivity is
reasonably good for polarization applications
from in the wavelength range from 125 to
160 . Thus, by tailoring the periodicity, MLs
can be used as reflection polarizing elements
for the wavelength range from 125 to 160
, because of the high reflectivity at
Brewsters angle.

Fig. 5.18 Peak reflectivity of a Mo/Si
ML with N=30 for different
wavelengths, from 115 to 160 ,
measured using the Indus-1 SR.
110 120 130 140 150 160
0
10
20
30
40
50
%

o
f

p
e
a
k

R
e
f
l
e
c
t
i
v
i
t
y
Wavelength ()
Chapter 5
146
5.4 Fe/B
4
C Bi-layers, Tri-layers, and Multilayer Structures

As we have shown earlier (Fig. 5.3), boron-based MLs give high reflectivity above
the boron K-edge. However, boron-based MLs are not stable. Therefore, instead of boron,
B
4
C-based MLs are used above the boron B K-edge, as these ML structures are stable and
have reasonably high reflectivity. Some authors have shown that ML structures of B
4
C with
transition metals (such as Mo, Ru, and W) provide high reflectivity in this wavelength range
and display high thermal stability
72-74
. Calculations shown earlier (Fig. 5.3) reveal that
Fe/B
4
C gives maximum reflectivity at wavelengths above the boron K-edge. While MLs of
Fe with C with systematically varied periodicity are reported as super mirrors
75
, MLs of
Fe/B
4
C have not been reported earlier. Therefore, the study of interfaces in the Fe/B
4
C
system should be considerable practical interest and importance.
Beside soft x-ray applications, the Fe/B
4
C system and Fe/B
4
C MLS also have
excellent soft magnetic properties, useful in high-density magnetic recording equipment
76
. A
dependence soft magnetic properties on the layer thickness has been observed. Therefore, it
is important and interesting to investigate systematically the nature of surfaces and interfaces
in the Fe/B
4
C system, both for optical and magnetic applications. To our knowledge, this has
not been reported previously.
In the present work, an attempt has been made to understand the interface
characteristics in Fe/B
4
C. A systematic study of interface nature in Fe-on-B
4
C and B
4
C-on-
Fe bi-layers, B
4
C-Fe-B
4
C tri-layers and, finally, ML structures with varying Fe and B
4
C layer
thickness, has been carried out using glazing-incidence XRR and glazing angle x-ray
fluorescence techniques. We have observed an interesting feature of interfaces in the
evaporated Fe/B
4
C system. The interface properties fluctuate with layer thickness, as
discussed below.

5.4.1 Experimental

Bi-layers, tri-layers and MLs of Fe/B
4
C system were deposited on float glass
substrates using the UHV e-beam evaporation system. The base pressure of deposition
chamber was ~ 2 10
-9
mbar. The deposition rate of ~10 /min for both Fe and B
4
C was
maintained using a quartz crystal monitor (QCM). The purity of Fe and B
4
C was 99.999%
and 99.95%, respectively. B
4
C-on-Fe and Fe-on-B
4
C bilayers, and B
4
C-Fe-B
4
C trilayers, of
Chapter 5
147
varying thickness of both the Fe and B
4
C layers, were deposited. Finally, Fe/B
4
C MLs were
fabricated with varying the and N parameters. The periodicity of MLs was varied from 47
to 56 for polarization application in the wavelength range 67 to 80 . The
surfaces/interfaces were characterized by XRR and glancing incidence x-ray fluorescence
(GIXRF) techniques
77
with a Cu target (=1.54 ) (cf. Chapter 2). XRR and GIXRF data
were least-squares-fitted using Parratts formalism
36
. For the fitting, the thicknesses of the
individual layers were taken from the QCM and the optical constants of materials were taken
from Henke et al.
52
, as the initial guesses. The depth-graded scattering length density profiles
were calculated from the best-fit data and these parameters were extended to derive the layer
structure in the Fe/B
4
C system.

Bi-layers and tri-layers

Fig. 5.19 shows the measured and fitted XRR spectra of B
4
C-on-Fe bi-layers with
different layer thicknesses. To fit the experimental data, and for pure Fe and B
4
C taken
from Henke et al.
52
were
Fe
=22.510
-6
,
Fe
=
2910
-7
, and
B4C
=7.5910
-6
,
B4C
=0.07910
-7

respectively. The best-fit XRR results are
tabulated in Table 5.4. The best-fit results
indicate formation of low a density Fe layer
(interlayer) at the B
4
C-on-Fe interface. As the
bottom Fe layer thickness increases, this
interlayer thickness increases, and its density
decreases. For comparing samples S1 and S2,
the bottom Fe layer is kept constant, whereas
the top B
4
C layer thickness is varied. In this
case, the microstructural properties (thickness
and density) of interlayer do not change.

Fig. 5.19 XRR spectra of B
4
C on-Fe
bi-layers with different thicknesses
(at the Cu-K
wavelength).
0.0 0.1 0.2 0.3 0.4 0.5
10
-6
10
-3
10
0
10
3
10
6
10
9
Measured
Fitted

S6
S5
S4
S3
S1
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
S2
Chapter 5
148
Fe layer Interlayer B
4
C layer Sample
Thickness
()
Thickness
()

10
-6

10
-7

Thickness
()
S1 105 (7) 10 (6) 18.5 16.5 43 (6)
S2 106 (7) 10 (6) 18.4 13.3 25 (6)
S3 160 (8) 13 (6) 16.5 13.4 515 (9)
S4 285 (8) 17 (6) 15.1 11.3 145 (6)
S5 600 (9) 22 (6) 13.2 7.4 400 (8)
S6 730 (9) 28 (7) 11.3 4.3 400 (8)

Table 5.4 Best-fit XRR results of B
4
C-on-Fe bi-layers. Roughness is given in parantheses.

This indicates that the interlayer properties do not depend upon the thickness of the B
4
C
layer. Again, comparing samples S5 and S6, where the B
4
C layer thickness is kept constant
and the Fe layer thickness is varied, the change in interlayer properties is apparent. This
reveals that the microstructural properties of the interlayer at the B
4
C-on-Fe interface depend
only on the Fe layer thickness.
Fig.5.20 shows the GIXRF profile of all six B
4
C-on-Fe bi-layer samples, along with
the fitted profile. The microstructural parameters obtained from best-fit profile match
reasonably well with the XRR results. The fluorescence measurement confirms the presence
of a low density interlayer structure between the Fe and B
4
C layers. This low density
Fig.5.20 GIXRF profile of B
4
C-on-Fe
bi-layers with different thicknesses. The
experimental data are fitted with an
interlayer model
Fig. 5.21 GIXRF profile of the S3
sample fitted with bi-layer and tri-
layer models. The inset shows the
model structure.
0.0 0.3 0.6 0.9 1.2 1.5
0
2
4
6
8
10
12
Measured
Fitted
X 1
X 2
X 3
X 4
X 8
S
6

S
5

S
4

S
3

S
2

F
e
-
K

f
l
u
o
r
e
s
c
e
n
c
e

i
n
e
t
e
n
s
i
t
y

(
n
o
r
m
.
)
Incidence angle (degree)
S
1

X 10
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Experimenta
With interlayer
Without interlayer
Interlayer
Fe

S 3
F
e
-
K

f
l
u
o
r
e
s
c
e
n
c
e

I
n
t
e
n
s
i
t
y

(
n
o
r
m
.
Incidence Angle (degree)
B
4
C
Chapter 5
149
interlayer structure behaves as a non-continuous layer, which is mainly due to the substantial
roughness of the Fe layer. This effect is seen more clearly in Fig.5.21, wherein GIXRF
profiles have been plotted for sample S3 using both the bi-layer and tri-layer models. It is
clear that the fluorescence profile for the tri-layer model fits better than the bi-layer model.
The model structure of the B
4
C-on-Fe interlayer is shown in Fig. 5.21 as the inset.
Fig. 5.22 shows the measured and fitted XRR spectra of Fe-on-B
4
C bi-layers and
B
4
C-Fe-B
4
C tri-layers. The best-fit XRR results are shown in Table 5.5. In Fe-on-B
4
C bi-
layers, there is no indication of the formation of an interlayer at the interface. However, in
B
4
C-Fe-B
4
C tri-layers, an interlayer is only formed at the B
4
C-on-Fe interface.

At the B
4
C-on-Fe interface, the nature of interfacial structure indicates the formation of a
low-density interlayer, but not due to the general trend of inter-diffusion or interlayer
compound formation. If the formation of the interlayer is due to the aforementioned
mechanism, then, following its formation, the density and thickness would not change as
function of the thickness of the bottom layer. Once an interlayer is formed, it would act as a
barrier to stop further inter-diffusion.

Fig. 5.22 XRR spectra of Fe-on-B
4
C bi-layers and B
4
C-Fe-B
4
C tri-layer.
0.00 0.09 0.18 0.27 0.36 0.45
10
-5
10
-3
10
-1
10
1
10
3
Measured
Fitted
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
M1
M2
M3
Chapter 5
150

Table 5.5 Fe-on-B
4
C bi-layers and tri-layers. Roughness is given in parantheses.

Therefore, the type of interlayer formed at the B
4
C-on-Fe interface may be due to physical,
topographical roughness. As the bottom Fe layer thickness increases, the roughness of the Fe
layer increases. Therefore, more and more B
4
C atoms are embedded into the valley created
due to roughness. Hence, the low-density interlayer thickness increases, while its density
decreases. In the case of the Fe-on-B
4
C interface, B
4
C forms a smoother layer, and thus gives
a sharp interface without formation of interlayer.

Fe/B
4
C Multilayers

For the surfaces and interfaces study, we present two sets of ML structures. In one
set, we vary thickness of layers, keeping N constant. In the other set, N is varied, keeping
layer thickness constant. Fig. 5.23 shows measured and fitted data for the Fe/B
4
C ML with
B
4
C layer Fe layer Fe oxide layer Sample
Thickness
()
Thickness
()
Thickness
()

10
-6

10
-7
bi-layer
M1 200 (6) 310 (9) 18 (8) 16.8 16.6
M2 700 (9) 170 (8) 17 (7) 16.7 16.5
tri-layer
Inter layer B
4
C layer M3 30 (5) 96 (6)
9 (4) 18.8 16.7 25 (5)
Fig. 5.23 XRR spectra at Cu-K
wavelength for two different Fe/B

4
C MLs with
N=5 and varying layer thicknesses. The best-fit results are tabulated in Table 5.6.
0.0 0.1 0.2 0.3 0.4 0.5
10
-5
10
-3
10
-1
10
1
10
3
0 300 600 900
2.0x10
-5
4.0x10
-5
6.0x10
-5
ML 1
ML 2
S
L
D

(
r
h
o
/
2
)
Depth ()
Measured
Fitted
ML 1
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
10
2
ML 2
Chapter 5
151
N=5, for different layer thicknesses. The best-fit results are tabulated in Table 5.6. The ML
samples show Bragg peaks up to the seventh order, indicating high quality MLs in terms of
roughness and thickness control. In sample ML1, the best-fit results reveal no detectable
amount of interlayer at B
4
C-on-Fe interface. As we discussed in the bi-layer case, the
interlayer formation at B
4
C-on-Fe interface is due to physical roughness of the bottom Fe
layer. As the Fe layer thickness is small in the ML1 sample, the roughness is less and, hence,
no interlayer formation occurs at the B
4
C-on-Fe interface.

Table 5. 6 Best-fit XRR results of the data of Figure 5.23

In the ML2 sample, the bottom Fe layer thickness is more and forms a rougher surface than
in ML1. Therefore, B
4
C gets embedded into the hill and valley structure of the rougher Fe
layer, and forms the interlayer at the B
4
C-on-Fe interface. The small difference in the optical
constants of pure Fe and B
4
C layers w.r.t. Henke et al. is due to the small change in the
density of the thin film. The inset in Fig. 5.23 shows the scattering density profile as function
of depth inside the MLS calculated from the reflectivity profile. In essence, this is the
variation in electron density, if one divides scattering length density by the classical electron
radius.
In the sample ML2, the periodicity d=50 and =0.4 were kept constant and N was
varied. Fig. 5.24 shows XRR spectra for samples with N=5 and N=10. The best-fit results are
tabulated in Table 5.7. The inset in Fig. 5.24 shows the scattering length density profile
inside the MLS. The best-fit results reveal that there is no formation of a low density (w.r.t.
Fe) interlayer at the B
4
C-on-Fe interface. This is because the roughness of the Fe layer is
small, as discussed earlier.
Sample
No.
Number of
layer pair
Layer Thickness
()
Roughness
()

10
-6

10
-7

ML 1 5 B
4
C
Fe
108
42
7
6
7.21
21.7
0.079
28.7

ML 2
5 B
4
C
B
4
C -on-Fe
Fe
27.3
10
95
6
5
5
6.9
18.6
21.8
0.078
17.6
28.9
Chapter 5
152

Sample
No.
Number of
layer pair
Layer Thickness
()
Roughness
()

10
-6

10
-7

ML 3 5 B
4
C
Fe
30
20
5
5
7.2
21.6
0.079
28.8
ML 4 10 B
4
C
Fe
30
20
7
7
6.1
21.8
0.072
29.1
Fig. 5.24 XRR spectra at Cu-K
wavelength for two different Fe(20 )/B

4
C(30 )
MLs with N=5 (ML 3) and N=10 (ML4). The best-fit results are tabulated in
Table 5.7.
0.0 0.1 0.2 0.3 0.4 0.5
10
-6
10
-4
10
-2
10
0
10
2
10
4
0 200 400 600
2.0x10
-5
4.0x10
-5
6.0x10
-5
ML 3
ML 4
S
L
D

(
r
h
o
/
2
)
Depth ()
ML 4
Measured
Fitted
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
10
2
ML 3
Chapter 5
153
5.5 Mo/Y Bi-layers, Tri-layers and Multilayer Structures

The theoretical calculated reflectivity, shown in earlier (Fig. 5.3), suggests that the
Mo/Y MLs would be the best for wavelengths shorter than the Si L-absorption edge (=124
). At below the Si L-edge, Mo/Si MLs can no longer be applied. Moreover, Be- and Sr-
based MLs are generally not possible because of their health hazard. Linear polarizers based
on Mo/Y MLs have been reported
78
, for use near =80 . To date, not much research has
been carried out on the surfaces and interfaces in the Mo/Y system. Near-normal angle
incidence peak reflectances of only 21.8% and 34 % were measured at wavelengths of 80
and 115 , respectively
79
. These lower reflectivities than expected could be due to structural
imperfections such as interface roughness, thickness errors, or impurities present in the MLs.
Subsequently, the same group has studied the effect of impurities in the layers and of surface
contamination on the ML performance. In this case, the MLs were sputter-deposited, and
reflectivity measured by in situ soft x-ray reflectometry
80
. A near-normal incidence
reflectivity of 46.1% at =114.3 was measured. While somewhat higher than previous
values, this is still much lower than expected. Sae-Lao et al.
81
have reported a reflectance of
34.6 % at 92 , which is 77 % of the ideal calculated reflectivity. Recently,
Kjornrattanawanich et al.
82
studied the effect of oxygen content in the yttrium layer on the
reflectivity in sputter- deposited MLs.
Given that the quality of interface plays an important role in ML performance, there
have been no reported attempts to examine the nature of interfaces in the Mo/Y system. It is
thus important and interesting to do so. In this work, we have carried out a detailed
surfaces/interfaces analysis on the Mo-on-Y and Y-on-Mo bi-layers, Mo-Y-Mo and Y-Mo-Y
tri-layers, and Mo/Y MLs with varying periodicity and N.

5.5.1 Experimental

All the bi-layers, tri-layers and ML samples were fabricated on float glass substrate (rms
roughness ~6 ) using a UHV e-beam evaporation system. The base pressure of deposition
chamber was ~ 2 10
-9
mbar. The purity of Mo and Y were 99.999 and 99.95, respectively.
The deposition rate of ~5 /min for both Mo and Y was maintained using a QCM. Mo/Y
MLs were fabricated with varying periodicity from 55 to 86 , for N=5 to N=10, keeping
Chapter 5
154
= 0.26 constant. All samples were characterized by hard x-ray reflectivity with a Cu target
(=1.54 ).

5.5.2 Results and Discussions
Bi-layers and tri-layers

Before analyzing the surface and interface of Mo/Y MLs, we have done detailed
study on bi-layers and tri-layers by systematic varying Mo and Y thickness. Fig. 5.25 shows,
representative XRR spectra for bi-layers (Fig. 5.25 (a)) and tri-layers (Fig.5.25 (b)) measured
at Cu K
wavelength along with fitted curves. For the XRR data fitting, the optical constants
used for pure Mo and Y from Henke et al.
52
as
Mo
= 28.8 10
-6
,
Mo
= 1.88 10
-6
,
Y
= 12.6
10
-6
and
Y
= 0.68 10
-6
.The XRR spectra are well fitted up to whole q-range revealing
actual real bi-layer and tri-layer system. The best-fit results were tabulated in Table 5.8. Here
in sample name BL stands for bi-layer and TL stands for tri-layer. In samples BL1 and TL1,
top layer is Y, therefore, there are clear observation of two critical angles in the XRR figures
one for Y (q
c
=0.04) and other for Mo (q
c
=0.0619). Where as in samples BL2 and TL2, the Y
critical angle is not observable because Mo is top layers in these samples. The in-set of Fig.
5.25(a) and Fig. 5.25(b) show the in-depth scattering length density variation inside the films
which is calculated from XRR measured data. For samples BL1 and TL1, there is a low-
density (w.r.t. Y) layer of thickness ~20 at top of Y layer (Table 5.8). Since Y is the top
layer, therefore, this low-density layer may be because of formation of yttrium oxide when
the samples are exposed to air. Similarly, for samples BL2 and TL2, there is also a low-
density (w.r.t. Mo) layer of thickness ~13 at top of Mo layer (Table 5.8). Again this low
density is may be due to formation of molybdenum oxide when samples are exposed to air.
The thickness of low-density layer is more for Y than Mo, this may be due to more
absorption of Y to oxygen than Mo. The optical constants of pure Mo and Y slightly differs
from Henke et al. may be due to density variation.
Chapter 5
155

Table 5.8 Best fit results obtained from Fig. 5.25 for Mo/Y bi-layers and tri-layers.
The best-fit results of all these bi-layers and tri-layers samples reveal there is no any
interlayer at the interfaces. The roughness in these samples increases as the thickness of layer
increases.

Sample
No.
Layer Thickness
()
Roughness
()

10
-6

10
-6

BL 1
YO
x
Y
Mo
20
188
99
9
11
7
6.27
11.46
27.66
0.31
0.69
1.87

BL 2
MoO
x

Mo
Y
13
93
201
8
10
8
15.54
27.52
11.31
0.93
1.84
0.66

TL 1
YO
x
Y
Mo
Y
20
189
100
201
10
13
11
8
6.21
11.49
27.63
11.49
0.33
0.63
1.84
0.63

TL 2
MoO
x

Mo
Y
Mo
13
92
201
99
9
12
10
7
15.44
27.43
11.41
27.43
0.95
1.89
0.63
1.89
Fig. 5.25 Measured and Fitted XRR spectra at Cu K
wavelength (= 1.54).
(a) For bi-layer samples: BL1 stands for Y(200 ) on Mo(100 ) and BL2 stands
for Mo(100 ) on Y(200 ). (b) For tri-layer samples: TL1 stands for Y(200 )-
Mo(100 )-Y(100 ) and TL2 for Mo(100 )-Y(200 )-Mo(100 ). The inset
shows in-depth the scattering length density profile.
0.00 0.05 0.10 0.15 0.20 0.25
10
-4
10
-3
10
-2
10
-1
10
0
10
1
10
2
BL 2
0 200 400
0.00
2.50x10
-5
5.00x10
-5
7.50x10
-5
Y-on-Mo
Mo-on-Y
S
L
D

(
r
h
o
/
2
)
Depth ()
Fitted
Measured
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
110
2
BL 1
(a)
0.00 0.05 0.10 0.15 0.20 0.25
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
10
1
10
2
10
3
0 200 400 600
0.00
2.50x10
-5
5.00x10
-5
7.50x10
-5
Y-Mo-Y
Mo-Y-Mo
S
L
D

(
r
h
o
/
2
)
Depth ()
TL 1
Fitted
Measured
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
110
3
TL 2
(b)
Chapter 5
156
Mo/Y Multilayers

Here we present the results of only two-representative Mo/Y ML structures. The
measured and fitted XRR spectra of typical Mo (23 )/Y (63 ) MLs with number of layer
pairs N=5 and 10 are shown in Fig. 5.26. In all these MLs the top layer was Y. During data
fitting, a low-density (w.r.t. Y) layer of thickness 20 at top Y layer was included. The best-
fir results are tabulated in Table 5.9. Fig. 5.26 shows Bragg peaks up

to 3
rd
order. Also for N=5 sample, the distinct N-2 number of Kiessig oscillations are
observed. For N=10, the higher order measured Bragg peak is wider than the fitted one. This
may be due to thickness errors during deposition of ML structure. The inset in Fig. 5.26
shows, the in-depth scattering length density variation inside the ML structures, which is,
calculated from measured XRR data.

Sample
No.
Number of
layer pair
Layer Thickness
()
Roughness
()

10
-6

10
-7

ML 1 5 Y
Mo
63
23
9
7
11.32
27.34
0.67
1.84
ML 2 10 Y
Mo
63
23
12
9
11.33
27.32
0.65
1.85

Table 5.9 Best-fit XRR results Mo (23 )/Y (63 ) MLs.

Fig. 5.26 XRR spectra of two different Mo (23 )/Y (63 ) MLs with N=5 (ML1)
and N=10 (ML2). The best-fit results are tabulated in Table 5.9.
0.05 0.10 0.15 0.20 0.25 0.30
10
-4
10
-2
10
0
10
2
10
4
Fitted
Measured
0 300 600 900 1200
0.00
2.50x10
-5
5.00x10
-5
7.50x10
-5
ML1
ML2
S
L
D

(
r
h
o
/
2
)
Depth ()
ML2
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
ML1
Chapter 5
157
It is note that, the rms roughness of substrates used for fabrication of Mo/Y MLs were ~6.
This increases the roughness in the MLs (Table 5.9). Again as the number of layer pair
increases, the roughness increases because of presence of cumulative type roughness in
evaporated Mo/Y MLs. In the ML structures, the density of pure Mo and Y is slightly lower
than Henke et al.
52
values because of density variations.

5.6 Conclusions

Good quality Mo/Si MLs with varying number of layer pairs, periodicity and ratio of
high-Z thickness to the periodicity have been fabricated using electron beam evaporation
system. The detailed surface and interface studies are carried out using hard x-ray reflectivity
and depth graded x-ray photoelectron spectroscopy. The experimental results are co-related
with theoretical calculations. The actual performance of the Mo/Si MLs for polarization
applications is tested using Indus-1 synchrotron radiation. In another study, an attempt has
been made to understand the interface characteristics in Fe/B
4
C system. For that, a detailed
investigation on interface quality has been carried out in B
4
C on-Fe and Fe- B
4
C bi-layers,
B
4
C-Fe- B
4
C tri-layer structures with systematic varying Fe and B
4
C thicknesses. The
surface and interface studies are carried out using hard x-ray reflectivity and glancing angle
fluorescence techniques. Finally, surfaces interfaces in Fe/ B
4
C MLs with varying periodicity
and number of layer pairs are carried out for the polarization applications at wavelength
above the boron K-absorption edge. Mo/Y ML system is also evaluated for polarization
application at wavelength above the yttrium M-absorption edges. For that a systematic
surfaces and interfaces characterization have been carried out on bi-layers, tri-layers and
finally on MLs with varying periodicity and number of layer pairs. Following are the main
findings of the above-mentioned study.

Mo/Si Structure

A detailed surfaces and interfaces analysis of Mo/Si system reveals that an interlayer
(sub-nanometer scale) is present between pure Mo and Si layers in as deposited
samples.
The interlayer layer is asymmetric in thickness i.e. Mo-on-Si interlayer are thicker
than Si-on-Mo. These results are consistent with the results reported on Mo/Si ML
using transmission electron microscopy.
Chapter 5
158
The mechanism of asymmetry in interlayer thickness is explained on the basis of
thermally activated model.
XRR measured data are fitted using a four-layer model to take account of interlayer
formation at the interfaces.
The detailed simulation study indicates, the roughness parameter reduces only the
peak reflectivity, whereas the interlayer redistributes the reflectivity pattern.
The model is able to discern the asymmetry in interlayer thickness at two interfaces if
the interfacial roughness is small compared to in-depth interlayer thickness. The
limitation is that, the sensitivity decreases with increasing interfacial roughness.
Survey scan of XPS results indicates, the interfaces are not contaminated with
impurities, for example oxygen.
The analysis of depth profile narrow scan XPS spectra of Mo 3d and Si 2p reveals
formation of MoSi
2
composition at both the interfaces.
TDF model predicts that solid-state amourphizing reaction occurs at Mo/Si interfaces
due to large negative heat of formation resulting in amorphous phase.
Effective heat of formation model predicts MoSi
2
as the first phase formed at Mo/Si
interface that agrees well with our XPS results.
Finally, the MLs with the desired periods are fabricated for the polarizing elements to
operate in the wavelength region 125 to 160 . The reflectivity performance in the s
polarization geometry is tested using Indus-1 reflectometry beamline. Maximum
reflectivity of 45% is achieved at the Brewsters angle with incidence wavelength 130
from ML sample with number of layer pair 30.

Fe/ B
4
C Structure

In evaporated Fe/B
4
C system, a low density (w.r.t. Fe) layer is observed at B
4
C-on-Fe
interface, where as Fe-on- B
4
C interface exhibit no such structure.
At B
4
C-on-Fe interface, the interlayer properties (density and thickness) change as
function of bottom Fe layer thickness but independent of B
4
C layer thickness. The
density of interlayer decreases and its thickness increases as Fe layer thickness
increases.
Chapter 5
159
The observed results were discussed with a model that, the formation of interlayer in
Fe/B
4
C system may not due to general trend of either intermixing or compound
formation at interface but may due to topographical roughness.
Finally good quality Fe/B
4
C MLs (in term of roughness and thickness control) are
fabricated.

Mo/Y Structure

In Mo/Y system, the systematic studies on bi-layers, tri-layers and MLs reveal there
is no any interlayer formation at interfaces.
The slight variation of optical constants of pure Mo and Y from Henke et al. is may
be due to variation of density.
In Mo/Y system, if Y is top layer then a yttrium oxide of thickness ~20 is formed
and in case if Mo is top layer then a molybdenum oxide of thickness ~13 is formed
when these samples are exposed to air.
XRR results reveal presence of cumulative type roughness in evaporated Mo/Y MLs.
Chapter 5
160
5.7 References

1.
T. W. Barbee, proc. Soc. Photo-opt. Instrum. Engg. 563, 2 (1985).

2.
J. H. Underwood, T. W. Barbee, D. C. Keith, Proc. SPIE 184, 123 (1979).

3.
E. Spiller, Soft X-ray Optics, SPIE Optical Engineering Press, Washington, USA,
(1994).

4.
D. L. Windt, W. K. Waskiewicz, J. E. Griffith, Appl. Opt. 33, 2025 (1994).

5.
A. G. Michette, Optical Systems for Soft X-rays Plenum Press, New York, (1986).

6.
B. W. Batterman and H. Cole, Rev. Mod. Phys. 36, 681 (1964).

7.
G. Koppelmann, Ann., Physik 5, 388 (1960).

8.
D Attwood, Soft X-rays and Extreme Ultraviolet Radiation: Principles and
Applications, Cambridge University Press (2000).

9.
A. V. Vinogradov and B. Ya. Zeldovich, Appl. Opt. 16, 89 (1977).

10.
C. K. Carniglia and J. A. Apfel, J. Opt. Soc. Am. 70, 523 (1980).

11.
E. Spiller, Appl. Phys. Lett. 20, 365 (1972).

12.
A. F. Jankowski and D. M. Makowiecki, Proc. PSIE 984, 64 (1988).

13.
T. W. Barbee Jr., S. Mrowka, M. C. Hettrick, Appl. Opt. 24, 883 (1985).

14.
M. Toyoda, Y. Shitami, M. Yanagihara, T. Ejima, M. Yamamoto and M. Wanatabe,
Japanese J. Appl. Phys. part1, 39, 1926, (2000).

15.
C. W. Gwyn, R. Stulen, D. Sweeney and D. Attwood, J. vac. Sci. technol. B 16, 3142,
(1998).

16.
T. W. Barbee et al., Opt. Eng. 30, 1067, (1991).

17.
R. B. Hoover, D. L. Shealy, B. R. Brinkley, P. C. Backer, T. W. Barbee and A. B. C.
Walker, Opt. Eng. 30, 1086, (1991).

18.
W. Hu, T. Hatano, M. Yamamoto and M. Watanabe, J. Syncrotron Rad. 5, 732
(1998).

19.
M. Yamamoto and M. Furudate, Thin Solid Films 313-314, 751 (1998).

20.
M. Yamamoto, H. Nomura, M. Yanagihara, M. Furudate and M. Watanabe, J.
Electron Spectroscopy and Related Phenomena 101-103, 869 (1999).

21.
N. Biswas, J. Gurganus and V. Misra, Appl. Phys. Lett. 86, 02210 (2005).

22.
J. Im, S. Yalisove, B. Adams, Y. Zhu, R. Chen, Mater. Res. Soc. Symp. Proc. Vol.
376, (1997).

23.
H. Nakajima, M. Ikebe, Y. Muto and H. Fujimori JAP, 65(4), 1637 (1989).

Chapter 5
161
24.
G. S. Elliott, A. D. Gromko, F. V. Veegaete, C. D. Johnson and D. C. Johnson, PRB
58, 8805, (1998).

25.
S. P. Murarka, Silicides for VLSI Applications, published by Academic, Press, Inc.,
London (1983).

26.
V. Demuth, H. P. Strunk, D. Worle, C. Kumpf, E. Brkel and M. Schulz, Appl. Phys.
A 68, 451 (1999).

27.
K. Holloway and R. Sinclair, J. Appl. Phys. 61, 1359 (1987).

28.
W. H. Wang and W. K. Wang, J. Appl. Phys. 76, 1578 (1994).

29.
A. K. Petford-long, M. B. Stearns, C-H. Chang, S. R. Nutt, D. G. Stearns, N. M.
Ceglio and A. M. Hawryluk, J. Appl. Phys. 61, 1422 (1987).

30.
S. Bajt, D. G. Stearns and P. A. Kearney J. Appl. Phys. 90, 1017 (2001).

31.
M. B. Stearns, C. H. Chang and D. G. Stearns, J. Appl. Phys. 71, 187 (1992).

32.
J. M. Slaughter, D. W. Schulze, C. R. Hills, A. Mirone, R. Stalio, R. N. Watts, C.
Tarrio, T. B. Lucatorto, M. Krumrey, P. Mueller and C. M. Falco, J. Appl. Phys. 76,
2144 (1994).

33.
A. E. Yakshin, E. Louis, P. C. Gorts, E. L. G. Maas and F. Bijkerk, Physica B 283,
143 (2002).

34.
M. H. Modi, G. S. Lodha, M. Nayak, A. K. Sinha and R. V. Nandedkar, physica B,
325, 272 (2003).

35.
M. J. H. Kessels, F. Bijkerk, F. D. Ticheelaar and J. Verhoeven, J. Appl. Phys. 97,
093513 (2005).

36.
L. G. Parratt, Phys. Rev. 95, 359 (1954).

37.
L. Nevot and P. Croce, Rev. Phys. Appl.15, 761 (1980).

38.
A. Ulyanenkov, R. Matsuo, K. Omote, K. Inaba and J. Harada, J. Appl. Phys. 87,
7255 (2000).

39.
Dong-Eon Kim, Soo-Mi Lee, In-Joon Jeon and M. Yanagihara, Appl. Surf. Sci., 127-
129, 531 (1988).

40.
T. Feigl, H. Lauth, S. Yulin and N. Kaiser, microelectronic Eng. 57-58, 3 (2001).

41.
M. Nayak, G. S. Lodha, A. K. Sinha, R. V. Nandedkar and S. A. Shivashankar, Appl.
Phys. Lett. 89, 181920 (2006).

42.
J. C. Bravman and R. Sinclair, J. Electron Microsc. Tech. 1, 53 (1984).

43.
J. F. Ziegler, SRIM 2000, IBM NY.

Chapter 5
162
44.

45.
H. Nohira, T. Shiraishi, K. Takahashi, T. Hattori, I. Kashiwagi, C. Ohshima, S. Ohmi,
H. Iwai, S. Joumori, K. Nakajima, M. Suzuki and K. Kimura, Appl. Suf. Sci. 234, 493
(2004).

46.
I. S. Choi, J. C. Park, J. T. Choi, H. J. Kim and S. Y. Lee, Appl. Phys. Lett. 80, 4339
(2002).

47.
G. Larrieu, E. Dubois, X. Wallart, X. Baie and J. Katcki, J. Appl. Phys. 94, 7801
(2003).

48.
Nguyen et. al Surface Sci. 162, 651 (1985).

49.
S. Rai, M. K. Tiwari, G. S. Lodha, M. H. Modi, M. K. Chattopadhyay, S. Majumdar,
S. Gardelis, Z. Viskadourakis, J. Giapintzakis, R. V. Nandedkar, S. B. Roy and P.
Chaddah, Phys. Rev. B 73, 035417 (2006).

50.
S. M. Chaudhari, D. M. Phase, A. D. Wadikar, G. S. Ramesh, M. S. Hegde and B. A.
Dasannacharya, Current Science 82, 305 (2002).

51.
G. S. Lodha, M. H. Modi, V. K. Raghuvanshi, K. J. S. Sawhney and R. V.
Nandedkar, Synchrotron Radiation News 17, 33 (2004).

52.
B. L. Henke, E. M. Gullikson and J. C. Davis, Atomic data and nuclear data table 54,
181 (1993).

53.
D. G. Stearns, M. B. Stearns, Y. Chang, J. H. Stith and N. M. Ceglio, J. Appl. Phys.
67 2415 (1990).

54.
Y. I. Jdiyaou, M. Azizan, E. L. Ameziane, M. Brunel and T. A. Nguyen, Appl.
Surface science 55, 165 (1992).

55.
W. Liwen, S. Wei, B. Wang and W. Liu , J. phys.: Condensed matter 9, 3521 (1997).

56.
W. L. Morgan and D. B. Boercker, Appl. Phys. Lett. 59, 1176 (1991).

57.
C. Kittel, Introduction to Solid State Physics, Wiley, New york (1976).

58.
E. A. Brandes and G. B. Brook (Eds.), Smithells metals reference book,
Butterworth-Heinemann, Jordan Hill, Oxford, London (1992).

59.
S. Yulin, T. Feigl, T. Kuhlmann, N. Kaiser, A. I. Fedorenko, V. V. Kondratenko, O.
Poltseva, V. A. Sevryukova, A. Yu. Zolotaryov and E. N. Zubarev, J. Appl. Phys. 92,
1216 (2002).

60.
J. M. Slaughter, A. Shapiro, P. A. Kearney and C. M. Falco, Phys. Rev. B 44, 3854
(1991).

Chapter 5
163
61.
R. B. Schwarz and W. L. Johnson, Phys. Rev. Lett. 51(5), 415 (1983).

62.
S. F. Gong and H. T. G. Hentzell, J. Appl. Phys. 68(9), 4542 (1990).

63.
R. Pretorius, A. M. Vredenberg and F. W. Saris, J. Appl. Phys. 70, 3636 (1991).

64.
A. R. Miedema, J. less-common metals 46, 67 (1976).

65.
A. R. Miedema, R. Boom and F. R. De boer, J. less-common metals 41, 283 (1975).

66.
S. F. Gong, A. Robertsson, H. T. G. Hentzell and X-H. Li, J. Appl. Phys. 68, 4535
(1990).

67.
J. Y. Shim, J. S. Kwak, E. J. Chi, H. K. Baik and S. M. Lee, Thin Solid Films 269,
102 (1995).

68.
J. B. Kortright and J. H. Underwood, Nucle. Instru. And Methods in Physics Research
A . 291, 272 (1990).

69.
H. Grimmer, O. Zaharko H. Ch. Mertins and F. Schafers, Nuclear Instrument and
Method A 467-468, 354 (2001).

70.
T. Kawamura, J. J. Delaunay, H. Takenaka, T. Hayashi and Y. Watanabe, J.
Synchrotron Rad. 5 (1998) 735.

71.
R. V. Nandedkar, K. J. S. Sawhney, G S Lodha, A. Verma, V. K. Raghuvanshi, A. K.
Sinha, M. H. Modi and M Nayak, Current science 82, 298 (2002).

72.
T. Bottger, D. C. Meyer, P. Paufler, S. Braun, M. Moss, H. Mai and E. Beyer, TSF
444, 165 (2003).

73.
E. Ziegler, O. Hignette, Ch. Morawe and R. Tucoulou, Nuclear Instrument and
Methods A 467-468, part2, 954 (2001).

74.
M. Yanagihara, K. Mayama, Y. Goto and I. Kusunoki, Nuclear Instruments and
Methods A 334, 638 (1993).

75.
I. V. Kozhevnikov, I. N. Bukreeva and E. Ziegler, Nnclear Instrument and Methods A
460, 424 (2001).

76.
T. Kobayashi JAP 73, 858 (1993).

77.
M. K. Tiwari, K. J. S. Sawhney, B. Gowri Sankar, V. K. Raghuvanshi and R. V.
Nandedkar, Spectrochemica Acta Part B 59, 1141 (2004).

78.
B. Kjornrattanawanich, R. Soufli S. Bajt, D. L. Windt and J. F. Seely, Proc. SPIE.
5538, 17 (2004).

79.
C. Montcalm, B. T. Sullivan, M. Ranger, J. M. Slaughter, P. A. Kearney, C. M. Falco
and M. Chaker, Opt. Lett. 19 1173 (1993).

Chapter 5
164
80.
C. Montcalm, B. T. Sullivan, S. Duguay, M. Ranger, W. Steffens, H. Pepin and M.
Chaker, Opt. Lett. 20, 1450 (1995).

81.
B. Sae-Lao, S. Bajt, C. Motcalm and J. F. Seely, Appl. Opt. 41, 2394 (2002).

82.
B. Kjornrattanawanich and S. Bajt, Appl. Opt. 43, 5955, (2004).

Chapter 6
165

CHAPTER-6

Soft X-ray Resonant Reflectivity for Low-Z Thin Films and Multilayers

This Chapter describes the application of soft x-ray resonant reflectivity (SXRRR) for
the characterization of low-Z thin films and the nature of their interfaces in multilayer
structures, using the fine structure features of energy-dependent atomic scattering factor
near the atomic absorption edges. Element specificity is achieved by tuning the energy of x-
rays to the absorption edge of the element being investigated. SXRRR measurements were
carried out using the energy tunability of the synchrotron radiation source Indus-1.
Two different studies have been carried out using SXRRR. In the first, we have shown
that SXRRR would open up a non-destructive technique for the determination of interlayer
composition at the interfaces with sub-nanometer scale sensitivity. Near the absorption edge,
atomic scattering factor is sensitive to composition, and can be exploited for compositional
analysis of buried interfaces. We demonstrate this for a well-characterized Mo-Si multilayer
system, using simulations, and by SXRRR measurements, tuning the incident photon energy
to near the Si L-edge. This is in general applicable to any low-Z bi-layer, tri-layer, and ML
structure.
In the second investigation, we provide direct experimental evidence of elemental
specificity of SXRRR through the effect of anomalous optical constants on the reflectivity
profile. At the anomalous region, the real part of the atomic scattering factor undergoes sign
reversal. Therefore, by tuning the photon energy in such a way that the refractive index of the
material would be very close to the refractive index of vacuum, the material becomes
virtually invisible to that energy. The interfaces are selectively probe due to high and tunable
optical contrast using resonant reflectivity, which is not possible using conventional XRR.
We have demonstrated this through the characterization of both B
4
C thin films and Fe/B
4
C
bi-layers by tuning the incident photon energy to the anomalous region near the boron K-
edge.
Chapter 6
166
6.1 Introduction

As discussed earlier (Chapters 2 and 5), x-ray reflectivity (XRR) is an important
nondestructive research tool to probe surfaces and interfaces
1-4
in thin film and multilayer
structures. The large dynamic range and high momentum scattering vector range (q
z
-range)
allow the use of Fourier transform methods to study the micro-structural parameters of
condensed matter films, such as thickness, surface and interface roughness, and interface
diffusion profile. It is important to recognize that electrons are responsible for the interaction
of x-ray with materials and x-rays actually probe the effective electron density, which is
directly related to the index of refraction
2
.
X-ray reflectivity shows a resonant behavior near the atomic absorption edges
5, 6

because of the strong variation of the atomic scattering factor near the edges. This is because
anomalous terms of the optical constant become significant near the atomic absorption
edges
7
. Stronger anomalous dispersion occurs near the absorption edges of the soft x-ray
region than in the hard x-ray region
8
. Elemental specificity of resonant XRR would be
achieved by tuning the energy of x-rays to the absorption edge of the element being
investigated, and can indeed be suited for the investigation of buried interfacial layers. Thus,
the structural, compositional, and magnetic structure of MLs can be depth-profiled by
varying the angle of incidence and the energy in this nondestructive technique. The use of x-
ray techniques based on resonant scattering phenomena has been rapidly increasing ever
since continuously tunable monochromatic x-rays became available at synchrotron sources.
While the imaginary part of the scattering factor has been used widely in the x-ray absorption
fine structure spectroscopy and related techniques, more attention has recently been paid to
the use of its real part. Examples are, the diffraction anomalous fine structure technique for
site-sensitive local structure measurements
9
, and resonance scattering on crystal truncation
rods from dissimilar buried interfaces
10
. Resonant XRR has also been used for the
characterization of magnetic materials
11
, ion distribution at biomimetic membranes
12
, oxide-
water interfaces
13
, polymer thin films
14
, organic thin films
15
, passive oxide on stainless
steel
16
, and bromine-labeled langmuir monolayers
17
. Although the basic principle of resonant
effect at near atomic absorption edges had been recognized previously as noted above, the
sensitivity of resonant XRR to determine the composition of an interlayer formed at buried
interfaces, with sub-nanometer resolution, has not yet been addressed. It is interesting and
important experimentally to observe the elemental specificity of SXRRR.
Chapter 6
167
In this study, we extend the resonance principle to determine the composition of a
phase formed at buried interfaces through the use of fine structure features of the atomic
scattering factor at the absorption edge of constituent elements. Although we demonstrate it
for the Mo-Si ML system, it is in general applicable to any bilayer/multilayer system near the
resonant absorption edges. In a companion study, we have determined the anomalous optical
constant of B
4
C materials experimentally near boron K-edge. In the anomalous region, the
refractive index of B
4
C is tuned such that it is very close to the refractive index of vacuum,
so that B
4
C becomes virtually invisible. The selected buried interfaces have been studied. We
demonstrate this through the characterization of B
4
C thin films and Fe/B
4
C bi-layers at the
boron K-absorption edge.
This Chapter is organized in the following manner: First, the energy-dependent
atomic scattering factor and hence optical constants are derived, and the effects near the
atomic absorption edges are discussed in a theoretical approach. This is followed by a brief
overview of synchrotron radiation, with particular emphasis on Indus-1 synchrotron source
parameters. Then, the possibility of determining interlayer composition using SXRRR is
discussed. Finally, direct experimental evidence of the elemental specificity of soft x-ray
resonant reflectivity is presented through the observation of the effect of anomalous optical
constants on the reflectivity pattern.

6.2 Theoretical Background

The optical properties of materials can be derived from the response of electrons to an
oscillatory electric field E. An atom is represented by a massive nucleus surrounded by
electrons held at discrete binding energies. In the semi-classical model, a massive nucleus
does not respond dynamically to the high frequency incident fields, but the electrons are
caused to oscillate at the same frequency as the incident electromagnetic wave, giving rise
to radiation scattering in all directions. Electrons bound to the nucleus by different forces
respond differently to the incident field. The response depends on the resonant frequencies
s

of the bound electrons; more specifically, on the closeness of the incident wave frequency to
the resonance frequency. In this semi classical model, a bound multi-electron atom is a
collection of harmonic oscillators, each with its own set of resonances. Bound electrons are
forced by the incident electric field to execute simple harmonic motion, in the presence of the
restoring central field of the positive nucleus. When the incident wave frequency is very
Chapter 6
168
close to the resonance frequency
s
, the absorption becomes significant. This absorption is
accounted for by including a damping factor in the equation for forced oscillation. Then, the
equation of motion for each of these electrons can be written as
18

i s s
s s
eE x m
dt
dx
m
dt
x d
m = + +
2
2
2
(6.1)

where x
s
is the displacement of bound electrons from mean position, is the damping
constant, m is the mass of electron, E
i
is the incident electric vector. The periodic electric
field of the incident wave can be written as

( )
s i
r K wt i
i i
e E E

=
.
0
, (6.2)

where E
i0
is the amplitude of incident electric vector, K
i
is the incident wave vector and r
s
is
the vector displacement of each electron from the nucleus. The solution of eq
n
(6.1) gives the
harmonic displacement of each bound electron as a function of time in the presence of
incident the electric field, and can be written as

m
eE
i
t x
i
s
s

=
2 2
1
) ( (6.3(a))

The acceleration of each of these bound electrons, which is the double derivative of
displacement, can be written as

m
eE
i
t a
i
s
s

=
2 2
2
) ( (6.3(b))

This displacement of each of the bound electrons gives rise to a dipole moment, and, in turn,
to a polarization
i e
E P = , where is the polarisability and
e
is the free electron density.
Then dielectric constant is
i
E
P
4 1+ = . The refractive index n, which is the square root
of dielectric constant, is given by
18

( )

=

=
Z
s
s
s a e
i
g N r
n
1
2 2
2 2
2
1

(6.3(c ))

Chapter 6
169
where incident wavelength, r
e
=2.8210
-13
cm is the classical electron radius, N
a
is the
number density of atoms in the material. Z is the atomic number, g
s
is the oscillator strength
which, in the semi-classical model, is defined as the number of electrons with a resonance
frequency
s
such that the sum of oscillator strengths is equal to the total number of electrons
i.e., Z g
s
s
=
. Thus, the refractive index has a strong frequency dependence.

Each of these oscillatory electrons radiates electromagnetic waves of the same
frequency. The electric field scattered by a multi-electron atom to a distant observer can be
written as
18

( )
( )
( )
4 4 4 4 3 4 4 4 4 2 1
4 4 4 3 4 4 4 2 1
electron free a by field electric Scattered
c r t i
i
e
q f
Z
s
s
r iq
s
e Sin E
r
r
i
e g
t r E
s
(
+
=

=

/
,
1
2 2
. 2
) , (

(6.4)

where r is the position vector, is the angle between the scattered radiation w.r.t. the
polarization direction of the incident electric field, and q is the change in momentum transfer.
Thus, the radiation scattered by a multi-electron atom is f(q,) times the radiation scattered
by a free electron. The term f(q,) is known as the atomic scattering factor. Total scattering
amplitude from a group of atoms is calculated as vector sum of the scattering amplitude from
the individual atoms.

6.2.1 Atomic Scattering Factor

The eq
n
(6.4) gives clear definition of the atomic scattering factor, which is defined as
the ratio of the scattered electric field from an atom to the scattered electric field from a free
electron. In other word, atomic scattering factor is the effective number of electrons per atom,
and can be written as

( )

=

+
=
Z
s
s
r iq
s
i
e g
q f
s
1
2 2
. 2
,

(6.5)

Thus, atomic scattering factor is a complex quantity. It depends on energy as well as the
momentum vector. It must be noted that the expression
s
r q . gives the phase variation of the
scattered fields, due to differing electron positions, as seen by the observer. Now, the phase
term is bounded by
Chapter 6
170

o s
a r q
sin 4
. (6.6)

where a
o
is the Bohr radius. The phase expression will be simplified in two special cases:

1 0 . <<

o
s
a
for r q (Long wavelength limit) (6.7 a)

1 0 . << for r q
s
(Forward scattering) (6.7 b)

In each of theses two cases, the atomic scattering factor ( ) , q f reduces to

( )

=
+
=
Z
s
s
s
i
g
f
1
2 2
2
0

(6.8)

This means that, in case of longer wavelength and/ or smaller scattering angle, the atom is
considered as a point to give scattered radiation in phase.
Therefore, the atomic scattering factor is a complex quantity and can be written as
7

" '
0
f i f f f + = (6.9)

where f
0
is the scattering factor for Thomson scattering (number of free electrons associated
with a given atom that are active in the scattering process), and is a real parameter;
'
f (also
real) is the correction due to dispersion, and
"
f i is the correction due to absorption.
'
f and
"
f typically become significant compared to f
0
only in the vicinity of absorption edges, and
are referred to as resonant (or anomalous) terms of the scattering factor. The anomalous
terms originate mostly from tightly bound inner electrons. If the incident photon frequency
is in the vicinity of an atomic resonance, the response of the corresponding core electrons
is influenced by their binding to the nucleus. The photon generates real and virtual transitions
in which it is absorbed and reemitted. This is the origin of resonance terms in the expression
for scattering rates.

Physical Picture of the Anomalous Scattering Factor:

Considerable insight may be gained from the approximate classical considerations of
photon scattering from a system of bound electrons with binding energies less than the
incident photon energy, and from those with binding energies greater than the photon energy.
Chapter 6
171
The former behaves as though they were free and, to a good approximation, scatter
according to the Thomson free electron model
18
. The phase of the polarization attributable to
these electrons lags the applied electric field by at frequencies far above resonance. The
second group of electrons, those with binding energies significantly greater than the photon
energy, behaves as though they were bound electrons. They contribute an in-phase
component to the polarization. But, except in the region of anomalous dispersion near edges,
this polarization is negligible relative to that of the free electrons. But at the absorption
edge, this cannot be neglected because of the quantum-mechanical exchange of oscillator
strength between various core levels.
Exceptions to this simple picture occur at absorption edges where the approximation
fails badly, and below the plasma frequency (
2
0
4 c r N
e
= , where N
e
is the density of free
electrons) where valence or conduction electrons give a strong, broad absorption. In this
region Rayleigh, i.e., resonant, scattering
19
, must be taken into account. In particular, at x-ray
edges, the polarization involving energy levels responsible for the edge becomes very large
and resonant effects dominate. At energies below an absorption edge, transitions to the states
responsible for the edge are virtual and the polarization is in phase with the incident
radiation. Well above the edge, the transitions are real and the polarization is 180
0
out of
phase with the incident field. Thus, the existence of a sign reversal in the real part of the
scattering factor (
'
0
f f + ) below an absorption edge depends on whether or not the in-
phase resonant polarization arising from near-edge transitions is sufficient to overcome the
out-of-phase polarization from transitions at lower energies.
Benfatto et al.
20
have given a detailed theory of the resonant atomic scattering factor
based on a multiple-scattering approach. Henke et al.
21
have given a comprehensive
tabulation of values for the atomic scattering factor (calculated and experimental, for some
elements) for all the elements Z=1-92 in the energy range 50 eV to 30 keV. They measured f
2

from transmission measurements and determined f
1
using the Kramers-Kronig relation. At
the absorption edges, there may be uncertainties in their tabulated values due to the missing
oscillator strength in the absorption co-efficient data used. The eq
n
(6.9) is more generally
written as

2 1
if f f = (6.10)

Chapter 6
172
where
'
0 1
f f f + = and
"
2
f f = . The real part f
1
gives the effective number of free
electrons involved in the scattering, and the imaginary part f
2
is proportional to the
absorption coefficient . The real part of the atomic scattering factor f
1
can be calculated
from f
2
by the Kramers-Kronig integral transformation given by
22

( )
+
(
(
=
0
2
1
2
2
1 1
2
) ( dE
E E
E Ef
E f
(6.11)

Similarly, f
2
can be calculated from f
1
using inverse the Kramers-Kronig relation.

Candidate Systems:

Conditions favorable for a sign change in the x-ray scattering factor include a strong
absorption edge with a sharp onset to provide an in-phase bound-electron polarization
comparable to the out-of-phase free-electron contribution. These conditions do not favor
edges originating from levels with only a few electrons, especially s-like levels with just two
electrons, such as K, L
I
, M
I
, etc., edges. (A possible exception may be K-edges in the second-
period elements, which lack a well-developed L shell). More likely are elements with L
II, III
,
M
IV, V
, etc. edges, which involve a large number of electrons. Secondly, the edge in question
must be sufficiently isolated that the high-energy absorption tails of transitions at lower
energies are small at the edge. Otherwise, the jump ratio at the edge will not be large, and the
edge not sufficiently abrupt to give strong anomalous dispersion. Further, absorptions at
lower energies must have sufficiently low total oscillator strength that their out-of-phase
polarization does not overwhelm the in-phase component of the bound-electron transitions.
These requirements do not favor the 3d transition metals in which a strong, broad M-shell
absorption extends throughout much of the vacuum ultra-violet. Together, these
considerations suggest that the L
II, III
edges of the third-period elements such as Al, Si, P etc.,
are among the most favorable candidates for sign reversal of (
'
0
f f + ). Other possibilities
include isolated M
IV, V
edges in the fourth-period elements and K edges in the second period.
Chapter 6
173
Fig. 6.1 shows the variation of f
1
and f
2

as function of photon energy for Si using the
Henke et al.
21
forward atomic scattering data
base. The real part f
1
approaches a value equal
to the atomic number Z (number of electrons
per atom) for the higher photon energies
except at the absorption edges. This is because
at higher energies, all the electrons behave as
free electrons. A perfectly free electron is
loss-free and all the energy received by the
electron is reradiated. Formally, losses are
introduced in the imaginary part f
2
.
.
The imaginary part f
2
decreases in general, but increases
at the absorption edge. Close to the absorption edge, f
1
and f
2
vary rapidly with energy
(anomalous dispersion). There is a sharp jump in f
2
from below the edge (in energy), while f
1

dips to a large negative value through the edge. Since the atomic-like assumption is clearly a
poor approximation in the vicinity of absorption edges because of chemical environment, the
resolution of the calculations of Henke et al. may not sufficient to establish the details of the
dispersion at the edges, but a sign change in f
1
would have several intriguing consequences.
For example, in such anomalous regions the intensity of the forward scattering, which is
proportional to
2
2
2
1
f f + , shows very deep minima. The atomic scattering factor of a
compound can be calculated using the scattering factor of individual atoms weighted with
density
23
. At the absorption edges, fine structure features of the atomic scattering factors
depend on the chemical environment of an atom. Again, at the absorption edges, f
1
and f
2
can
vary drastically with a small change in the composition of the materials and, hence, would
make possible an analysis of composition of buried interfaces through contrast in the optical
constants.

Refractive Index:

Comparing eq
n
(6.3 (c)), eq
n
(6.8), and eq
n
(6.10), refractive index of materials in the
x-ray region where the incident photon frequency is larger than plasma frequency can be
written as
18

Fig.6.1 Real (f
1
) and imaginary (f
2
)
part of the atomic scattering factor of
Si as a function of photon energy.
10 100 1000 10000
0.1
1
10
5
10
15
K-edge

f2
Energy (eV)
L-edge
f1
A
t
o
m
i
c

s
c
a
t
t
e
r
i
n
g

f
a
c
t
o
r
Chapter 6
174
( )
2 1
2
2
1 1 if f
N r
i n
a e
= + =
(6.12)

where 1- is the real part of refractive index. and are known as optical constants and are
related to the real and imaginary part of the atomic scattering factor f
1
and f
2
by the relations

1
2
2
f N
r
a
e
= (6.13 a)

2
2
2
f N
r
a
e
= (6.13 b)

N
a
(number of atoms per unit volume) can also be written as,
AW
N
a

= , where of mass
density, AW is the mass of one atom (atomic weight/Avogrados number). For all materials,
typically, ~10
-2
to 10
-5
and ~10
-2
to 10
-7
, in the x-ray region (except at atomic absorption
edges where anomalous effect occurs), with decreasing trend towards higher energies.
The real part of the complex energy-dependent refractive index is less than unity for
energies above the valence-electron plasma frequency (commonly 10-20 eV), i.e., for VUV
and x-rays
18
. The important consequence of this is the phenomenon of total external
reflection (in the limit of 0) at the material/air or vacuum interface, below the critical angle
2
c
. Speculation is that the real part of refractive index (1-) exceeds unity near some
x-ray absorption edge because of f
1
becoming negative (because of the sign reversal effect).
In such an anomalous region, total external reflection can not occur at any angle.
Away from absorption, varies as
2
and varies as
3
. Therefore, absorption
decreases faster than the refractive index for increasing photon energy, and materials become
absorption-free dielectrics at short wavelength. The variation of and as a function of
wavelength of the various materials commonly used in x-ray multilayers is plotted in Fig. 1.1
(Chapter 1). In the hard x-ray to vacuum ultraviolet region, the optical constants are generally
measured using angle-dependent reflectivity measurements
24-27
and transmission
measurements
21
. In the transmission measurements, is measured and is obtained using the
Kramers-Kronig relation whereas, in reflectivity measurements, both and are obtained
from a fitting of the Fresnel equations to the reflectivity data. The other techniques which can
also be used for optical constant measurements at longer wavelengths are ellipsometry
28
,
electron energy loss spectrometry
29
and interferometry
30
.
Chapter 6
175

6.3 Synchrotron Radiation

When a charged particle moving at relativistic speeds accelerates, such as one traveling in a
curved trajectory, it emits radiation in a narrow cone tangential to the path of the particle
31
.
This radiation is known as synchrotron radiation (SR). The SR spectrum extends from IR to
hard x-rays, depending on the energy of the charged particle. Generally, in synchrotrons, the
SR is extracted by accelerating (change of direction) charged particles in a magnetic field.
There are three types of magnetic structures commonly used to produce SR, i.e., bending
magnets, undulators, and wigglers
32
. The last two are commonly known as insertion devices.
Undulators are periodic magnetic structures of relatively weak magnetic field so that
electrons experience harmonic oscillation. The resultant radiation cone is narrow and of high
brilliance. Wigglers are a strong magnetic version of undulators to give higher photon flux
and a shift of radiation towards shorter wavelengths. Synchrotron sources are described in
terms of a characteristic parameter known as the critical wavelength or frequency or critical
energy, which divide the power spectrum into two equal parts. If the magnetic field is B
(Tesla), electron energy is E (GeV), and R the radius of the electron orbit (meter), then, the
critical wavelength (in ) can be written as

3 2
59 . 5
64 . 18
E
R
BE
= = (6.14)

In the general process of SR emission, low energy (~50 MeV) electrons or positrons
are injected from a linear accelerator or a microtron into the synchrotron, also known as
booster. In the booster, electrons are accelerated to the required energy by a RF power supply
and made to revolve in a fixed orbit by increasing magnetic fields. When the appropriate
energy is attained, the electrons are transferred to the storage ring. In the storage ring,
electrons revolve in bunches and radiate SR when they accelerate through the magnets. To
minimize electron-gas atom collisions, the synchrotron source operates in ultra-high vacuum
(10
-10
Torr). The SR is brought to the experimental station through a beam line with optical
components appropriate for the desired experiments. The unique characteristic features of SR
are as follows:
Energy tunability from IR to the x-ray region.
High flux, several orders of magnitude higher than laboratory sources.
High brilliance and high degree of collimation (~a few mrad ).
Chapter 6
176
Well-defined state of polarization, linearly
in the plane of electron orbit and elliptically
polarized above and below the orbit.
Time structure, pulse duration of ~pico
seconds.

6.3.1 Indus-1 Synchrotron Source

Indus-1 is a 450 MeV electron storage ring
having a circumference of 18.9664 m to provide
synchrotron radiation continuously from IR to soft
x-rays
33
. The SR with critical wavelength of 61
(202 eV) is emitted from four 1.5 T bending
magnets. The photon flux from Indus-1 machine
34
is
shown in Fig. 6.2. The energy loss of electrons due to emission of SR is given by a RF cavity
of 31.619 MHz. There are two electron bunches of bunch length 2 ns and bunch separation
32 ns. The parameters of the Indus-1 synchrotron source are summarized in Table 6.1.

Electron energy 450 MeV
Beam current ~125 mA
Beam life time 2.5 hours
Critical wavelength 61
Bending magnetic field 1.5 T
Circumference 18.9664 m
Photon flux @
C
7.2 10
11
photons/ sec/ mrad horiz./ 0.1% BW
Brightness 3.110
12
photons/ sec/ mm
2
/ mrad
2
horiz./ 0.1% BW

In this Chapter we report on using the energy tunability of the SR to measure soft x-
ray resonant reflectivity by tuning the photon energy near the atomic absorption edges. As
already noted, reflectivity decreases inversely as the fourth power of the momentum transfer
Table 6.1 Parameters of the Indus-1 Synchrotron Source
Fig. 6.2 Photon flux as a function
of photon energy of the Indus-1
synchrotron source.
10
-2
10
-1
10
0
10
1
10
2
10
3
10
4
10
8
10
9
10
10
10
11
10
12
10
13
3T Wiggler
P
h
o
t
o
n
s
/
s
/
m
r
a
d

H
o
r
z
.

0
.
1
%
B
W
Photon Energy (eV)
Dipole
Chapter 6
177
vector (q
z
=

sin 4
). Therefore, for reflectivity measurements, a large dynamic range (five
to seven) is required to scan large q
z
. Hence, a high photon flux beam is required. The
polarization of SR is useful for the polarization measurements, as discussed in Chapter 5. The
following sections present results of the SXRRR measurements carried out using the Indus-1
synchrotron source.
Chapter 6
178
6.4 Soft X-ray Resonant Reflectivity as a Non-destructive Technique for Interlayer
Composition Analysis

Generally, the composition of interlayer of sub-nanometer thickness at buried
interfaces can be determined using high resolution cross-sectional transmission electron
microscopy
35
and/or depth-graded XPS
36
. However, these are destructive methods. In this
work, we present a non-destructive method for identifying the composition of interlayers at
buried interfaces with sub-nanometer sensitivity, using SXRRR. Normally, in conventional
XRR measurements, the x-ray photon energy is far from the absorption edges of materials of
interest, and the poor contrast in optical constants is not sufficient to identify the composition
of buried interfaces on a sub-nanometer scale. However, the energy- dependent atomic
scattering factor is sensitive to composition at the atomic absorption edge of the constituent
element. Therefore, it would be possible to identify the composition from the reflectivity
pattern using the fine structure feature of the atomic scattering factor. This can be done by
tuning the photon energy near the absorption edge to get sufficient contrast in optical
constant among different phases of slightly varying composition. But, this approach has not
been taken so far. We demonstrate this method theoretically, and then experimentally, using
the Indus-1 SR tuned to the Si L-edge, for the characterization of the Mo-Si multilayer (ML)
system. The technological importance of the Mo/Si system has been detailed in Chapter 5.
The present method would in general be applicable to any bi-layer/multilayer system near the
resonant absorption edges.

6.4.1 Simulations

The optical constants depend on the
electron distribution in the material, through
the atomic scattering factor, as given in eq
n

(6. 13 a) and eq
n
(6.13 b). Figure 6.3 shows
the optical constants computed in the energy
range from 94 eV to 113 eV for Mo, Si, and
different silicide compositions, using the
Henke et al.
21
atomic scattering database.
For the silicide phases, we took densities of
FIG. 6.3 Energy dependence of the
optical constants ( and ) for Mo, Si,
MoSi
2
, Mo
5
Si
3
and Mo
3
Si, calculated
near the Si L-absorption edge.
95 100 105 110 115
-0.02
0.00
0.02
0.04
0.06
0.08
0.000
0.003
0.006
0.009
0.012
0.015
0.018
0.021
0.024
Mo
Si
MoSi
2
Mo
5
Si
3
Mo
3
Si

o
p
t
i
c
a
l

c
o
n
s
t
a
n
t

(
)
(eV)

o
p
t
i
c
a
l

c
o
n
s
t
a
n
t

(
)
Chapter 6
179
MoSi
2
, Mo
5
Si
3
and Mo
3
Si to be 6.24, 8.24 and 8.968 gm/cm
3
, respectively
37
, and the optical
constants were calculated using weighted averages of atomic scattering factors. In the
vicinity of Si L-edges (L
II
=100.47 eV and L
III
=99.9 eV), the optical constants of Si and its
different silicide compositions change rapidly as a function of photon energy. In general, the
fine structures of the optical constants near absorption edges depend on the composition of
the phases. It is also clear from Fig. 6.3 that the three different silicide phases (MoSi
2
,
Mo
5
Si
3
, and Mo
3
Si) of varying Si content display markedly different optical constants at the
edge and that the optical contrast is sufficiently large to enable observation of changes in
reflectivity patterns from buried interfaces.
The effect of energy-dependent optical
constants on reflectivity R(q
z
) near the Si L-edge
is illustrated in Fig. 6.4 through calculations for
an ideal (without roughness) Mo/Si ML structure
with varying interlayer thicknesses (t
i
=0.5, 1, and
1.5 nm), and with changing interlayer
compositions. For the Mo/Si ML system, we have
taken the period d=9 nm, the number of layer
pairs, N= 5, and the ratio of Mo thickness to the
period, =0.3, and simulated the XRR using
Parratts formalism
2
. The same formalism was
also used for experimental data fitting based on a
non-linear least squares-fitting algorithm. The
reflectivity pattern clearly distinguishes three
different compositions, even for interlayer
thickness on the sub-nanometer scale. Refraction
changes the angle of propagation of the radiation
entering the ML and therefore changes the angular position of the Bragg peak for different
compositions. Sensitivity of resonant reflectivity to phase composition at the interfaces of bi-
layers and tri-layers was checked through simulations.

Fig. 6.4 Calculated soft x-ray
reflectivity of [Mo/Si]
5
ML structure
with period d=9 nm and = 0.3 for
selected photon energy at Si L-edge.
0.00 0.02 0.04 0.06 0.08 0.10
10
-4
10
-2
10
0
10
2

t
i
=1.5 nm
t
i
=1 nm
t
i
=0.5 nm
=100.8 eV
q
z
(
-1
)
10
-2
10
0
10
2
MoSi
2
Mo
5
Si
3
Mo
3
Si
=100 eV
t
i
=1.5 nm
t
i
=1 nm
t
i
=0.5 nm
R
e
f
l
e
c
t
i
v
i
t
y

Chapter 6
180
6.4.2 Experimental

To test this idea experimentally, Mo/Si MLs were deposited using a UHV e- beam
evaporation system (base pressure ~210
-9
mbar). A deposition rate of ~6 /min for both Mo
and Si was maintained using a quartz crystal microbalance. The Mo/Si MLs were fabricated
on float glass substrate with a periodicity of 9 nm, =0.3, and N = 5. Hard x-ray reflectivity
measurements were carried out on an x-ray reflectometer with a Cu target (cf. Chapter 2).
Angle-dependent soft x-ray reflectivity measurements were carried out using the EUV/soft x-
ray reflectometry beamline on the Indus-1 SR source
38
. The reflectometry beamline
employed a toroidal grating monochromator for high flux and moderate spectral resolution.
Three gratings with different groove densities were used to deliver monochromatic radiation
in the energy range 10 eV-300 eV with resolution EE= 200-300. Various absorption edge
filters are mounted in the beamline to suppress higher order contamination. The beamline
operating in a UHV environment was connected to a high-vacuum reflectometer through a
differential pumping system. The goniometer assembly of the experimental station comprises
two rotary and one linear stage, operating in a high vacuum environment. The linear stage
was employed to bring the sample in and out of the SR beam, for direct beam monitoring.
Measurement of the photocurrent, with the incident beam passing through a 1 mm gold
pinhole, continuously monitors the incident intensity variation. The SR beam current is also
continuously monitored. The angle-dependent
reflectivity was measured with a XUV silicon
photodiode. The angular resolution of the
reflectometer is 0.005 degree. The present
measurements were performed in the s-
polarized geometry.


Figure 6.5 shows the hard XRR of the
Mo/Si ML fabricated for the present study.
The measured reflectivity, R(

sin 4
=
z
q ),
was fitted to model the reflectivity. The
successive higher order Bragg peaks (up to the
FIG.6.5 XRR spectrum of [Mo/Si]
5
ML with period d=9 nm and = 0.3
at Cu K energy (E=8047 eV).
0.0 0.1 0.2 0.3 0.4 0.5 0.6
10
-8
10
-6
10
-4
10
-2
10
0
Measured
Fitted (MoSi
2
)
Fitted (Mo
5
Si
3
)
Fitted (Mo
3
Si)
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
Chapter 6
181
7
th
order), and the distinct N-2 Kiessig oscillations between peaks in measured data, reveal
the high quality of the ML structure. The microstructural parameters (thickness t and
roughness ) obtained from best-fit are: t
Si
= 4.70.1 nm with = 0.5 nm; Si-on-Mo
interlayer, t
So-on-Mo
= 0.80.1 with = 0.5 nm; t
Mo
= 2.40.1 with = 0.6 nm; Mo-on-Si
interlayer, t
Mo-on-Si
= 1.00.1 with = 0.45 nm. The interlayer optical constant =19.110
-6

and = 11.610
-7
. The best-fit result also reveals the formation of a native oxide of 1.2 nm
thickness with a roughness of 0.5 nm on the Si layer at the top of the ML structure. The
theoretical fitted data for three possible (different) interlayer phase compositions
39
show
clearly that conventional XRR is not sensitive to the composition of phases formed at buried
interface. This is due to the small difference in optical constants among these compositions.
Figure 6.6 shows the SXRRR sprectra measured near the Si L-edge energy along with
the fitted curves. The experimental data are fitted for three different compositions i.e., MoSi
2
,
Mo
5
Si
3
, and Mo
3
Si. For the fitting, the
microstructural parameters obtained from
hard XRR (given above) were used, and the
optical constants ( and ) were taken from
Fig. 6.3 as the initial guess. The measured
reflectivity profile shows an overall trend
similar to that in the simulations (Fig. 6.4).
For two different photon energies, good
agreement between measured and best-fit
curves is obtained for the MoSi
2

composition, not only near the Bragg peak,
but also in the Kiessig oscillations and at
grazing angles of incidence around the
critical angle of total reflection. The
amplitude of the intensity modulation is
smaller at E=100.8 eV than at E=100 eV,
which is due to the increase in the effective electron density which, in turn, depends on the
real part of the atomic scattering factor f
1
. The best-fit results are tabulated in Table 6.2. At
E=100 eV and 100.8 eV, both and values for Si and MoSi
2
differ significantly from those
of Henke et al.
21
, given in Table 2 within square brackets. At the absorption edge, these
Fig. 6.6 Measured soft x-ray
reflectivity of [Mo/Si]
5
ML at the
same photon energies as calculated
in Fig. 6.4. The best fit is obtained
for MoSi
2
composition and the
results are given in Table 6.2.
0.00 0.02 0.04 0.06 0.08 0.10
10
-3
10
-2
10
-1
10
0
10
1
E=100 eV
E=100.8 eV

Measured
Fitted (MoSi
2
)
Fitted (Mo
5
Si
3
)
Fitted (Mo
3
Si)
R
e
f
l
e
c
t
i
v
i
t
y

q
z
(
-1
)
210
Chapter 6
182
discrepancies in optical constants may be due to their rapid variation and/or extreme
sensitivity to any slight error in energy calibration or resolution. For Mo, both and values
agree well with those of Henke et al.
21
(within experimental error) for the two different
energies. The fitted curves for Mo
5
Si
3
and Mo
3
Si deviate significantly from the measured
data. Therefore, we conclude that the interlayer formed at the Mo/Si interface is of the MoSi
2

composition. The observed results obtained from SXRRR, too, agree well with depth-graded
XPS measurements discussed in Chapter 5.

Table 6.2 Best-fit results of soft x-ray resonant reflectivity measurements on Mo/Si ML
with MoSi
2
composition at the interfaces, obtained by tuning photon energy to the
Si L- edge. In square brackets are the tabulated values of and from Henke et al.
21
Energy
(eV)
Si
Si
Mo
Mo
MoSi2
MoSi2
R(%)
100.8 -0.012 1
[-0.014]
0.015 9
[0.0167]
0.058 5
[0.0589]
0.004 8
[0.005]
0.007 9
[0.0089]
0.017 6
[0.0184]
8.6
100 -0.015 5
[-0.0163]
0.007 4
[0.0066]
0.060 5
[0.0603]
0.005 1
[0.005]
0.008
[0.0071]
0.0092
[0.0085]
9.3
Chapter 6
183
6.5 Elemental specificity of Soft X-ray Resonant Reflectivity

Resonant x-ray reflectivity exploits the energy dependence of atomic scattering factor
to locate resonant atoms within the electron density distribution of the films
7
. Resonant
reflectivity provides selective sensitivity to specific chemical composition near the
absorption edges of constituent elements. Elemental specificity is achieved by tuning the
energy of the x-rays to the resonant (absorption) edge of the element being investigated. In
the interference pattern, the -2 specular resonant reflectivity at a fixed energy contains
structural information of the films. Specular reflectivity provides the effective electron
density profile distribution of the material projected onto the normal to the sample surface
2
.
The real part of refractive index (1-) is normally less than unity for photon energies
above the valence-electron plasma frequency
p
(commonly 10-20 eV), i.e., from the
vacuum ultra-violet to x-rays. Henke et al.
21
calculated the complex atomic scattering factor
for most elements and suggest the real part of the atomic scattering factor f
1
may become
negative for some atomic absorption edges and, hence, the real part of refractive index may
become positive. The sign reversal of f
1
is due to the dispersion effect at the absorption edges
of certain materials. It may be noted that f
1
is negative over a substantial frequency range to
satisfy the inertial sum rule
40
. Strong anomalous dispersion occurs near the absorption edges
in the soft x-ray region rather than the hard x-ray region.
8, 41

In this work, we have experimentally determined the anomalous optical constant of
B
4
C at the boron K-edge (E=187.7 eV) using angle-dependent specular SXRRR
measurements. We provide direct experimental evidence of elemental specificity of SXRRR
by showing how the anomalous variations of the optical constants affect the SXRRR
spectrum. We demonstrate the virtual invisibility of B
4
C by tuning the photon energy such
that refractive index of B
4
C (0 and 0) is very close to the refractive index of vacuum
(=0 and =0). The interfaces have been studied selectively due to high and tunable optical
contrast using SXRRR. Specifically, we demonstrate this through the characterization of B
4
C
thin films and Fe/B
4
C bi-layers using SR.

6.5.1 Experimental

B
4
C thin films and B
4
C-on-Fe bi-layers were deposited on float glass substrates using
the UHV e-beam evaporation system at the base pressure of ~210
-9
mbar. The deposition
Chapter 6
184
rate of ~10 /min for both Fe and B
4
C was maintained using a quartz crystal microbalance
(QCM). The purity of Fe and B
4
C was 99.999% and 99.95%, respectively. Microstructural
parameters such as film thickness, density, and roughness were determined from hard XRR
measurements with a Cu target (=1.54 ), before angle-dependent SXRRR measurements
using the Indus-1 synchrotron source. The experimental data were fitted to the Fresnel
equations in Parratts formalism
2
.


Hard X-ray Reflectivity:

Prior to the SXRRR measurements, the microstructural parameters of B
4
C single and
B
4
C-on-Fe bi-layers were determined using hard XRR at the Cu-K
energy (E=8047 eV).

The measured and fitted XRR spectrum of the B
4
C film and the B
4
C-on-Fe bi-layer are
shown in Fig. 6.7 and Fig. 6.8, respectively. The inset in these figures shows the depth-
graded scattering length density profile of the films, which is calculated from the best-fit
XRR spectrum. The best-fit results of the B
4
C film and the B
4
C-on-Fe bi-layer are shown in
Table 6.3 and Table 6.4, respectively. In Table 6.3, the thickness obtained from the fitting is
slightly larger than that recorded by the QCM because of a multiplication factor, which arises
due to the relative position of the sample with respect to quartz crystal. The best-fit results
also reveal that the density of B
4
C in the film is 96% of its bulk value.
0.00 0.05 0.10 0.15 0.20 0.25
10
-6
10
-4
10
-2
10
0
Measured
Fitted
0 400 800
0.000000
0.000005
0.000010
0.000015
0.000020
S
L
D

(
r
h
o
/
2
)
Depth ()

R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
Fig. 6.7 XRR spectrum of 800
B
4
C single film at Cu-K
energy
(E=8047 eV). The inset shows the
depth-graded SLD profile.
0.00 0.09 0.18 0.27 0.36
10
-5
10
-3
10
-1
-300 0 300 600 900
0.000000
0.000015
0.000030
0.000045
0.000060
S
L
D

(
r
h
o
/
2
)
Depth ()
Measured
Fitted
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
Fig. 6.8 XRR spectrum of B
4
C(600)-
on-Fe (180) bi-layer film at Cu-K
energy (E=8047 eV). The inset shows

the depth-graded SLD profile.
Chapter 6
185

In the B
4
C-on-Fe bi-layer sample, the best fit results (Table 6.4) reveal that there is a
low density Fe layer at the interface between the pure Fe and B
4
C layers. The density of this
layer is 65% of the pure Fe layer density and its thickness is 13 . The formation of this low-
density layer may be due to porosity at the top of the Fe layer, which arises because of
physical roughness. The density of the pure Fe layer is 96 % of its bulk value.

Soft X-ray Resonant Reflectivity:

Before discussing the SXRRR results, let us consider the variation of optical
constants with energy, which ultimately depends on
the energy-dependent forward atomic scattering
factor through the eq
n
. (13 (a)) and eq
n
. (13 (b)).
Figure 6.9 shows the optical constants computed in
the energy range from 165 eV-205 eV for B
4
C and
Fe using the Henke et al.
21
atomic scattering
database. The density of B
4
C and Fe obtained from
the hard XRR measurements was used for the
calculation of the optical constants. In the vicinity
of the boron K-absorption edge (E=187.7 eV), the
optical constants ( and ) of B
4
C change rapidly as a function of photon energy. The
dispersion term of the optical constant () even becomes negative, because of sign reversal of
the real part of the forward atomic scattering factor (f
1
) at the absorption edge stemming from
the anomalous effect. The reflected intensity as a function of scattering vector (q
z
) of a real
Fitted Parameters Film
()
Layer t
()

()
f
B
4
C 805 12 96%
of
bulk
B
4
C
(800)

Float
glass
5

Table 6.3 Best-fit results of XRR
measurement of 800 B
4
C film.
Fitted Parameters Film
()
Layer t
()

()
f
B
4
C
Low density Fe
Fe
596
13
172
8
6
6
96 % of bulk
65 % of Fe
97 % of bulk
B
4
C/Fe
(600/180)

Float glass 4

Table 6.4 Best-fit results of XRR measurement
of B
4
C(600)-on-Fe (180) bi-layer film.
160 170 180 190 200 210
-0.008
0.000
0.008
0.016
0.024
0.032
Fe ()
Fe ()
B
4
C ()
B
4
C ()
O
p
t
i
c
a
l

c
o
n
s
t
a
n
t

(

a
n
d

)
Photon energy (eV)
FIG. 6.9 Calculated optical
constant for B
4
C and Fe near
the boron K-absorption edge.
Chapter 6
186
structure, and the extent of the interference oscillations, depend on the relative Fresnel
reflection co-efficient at each interface as well as on the absorption in the material.
Therefore, the reflectivity from a particular interface can be selectively tuned using the fine
structure features of energy dependent optical constants at the absorption edge of constituent
element. This has been shown as follows.

B
4
C thin film:

Fig. 6.10 shows soft x-ray reflectivity spectra as a function of scattering vector (q
z
) of
a 805 thin B
4
C film at different photon energies near the boron K-absorption edge. For the
fitting, the microstructural parameters (such as thickness, density, and roughness) obtained
from hard XRR (Table 6.3) were used, and the optical constants ( and ) were taken from
Fig. 6.9 as the initial guess. The best-fit results are tabulated in Table 6.5. The amplitude of
the intensity modulation is larger for 173 eV due to the relatively well-matched Fresnel
reflection co-efficient of the vacuum/B
4
C and B
4
C/substrate interfaces. As the photon energy
increases, for 180 eV, the amplitude modulation decreases. For 185.3 eV, there is no
observable amplitude of the intensity modulation. At this energy, =0 and 0 for B
4
C
(Table 6.5), which is close to the refractive index for vacuum (=0 and =0). Therefore, at
this energy, B
4
C becomes virtually invisible and, the only reflection takes place from the
substrate/virtually invisible B
4
C interface. For 187.5 eV, again, the amplitude of intensity
modulation re-appears due to reflection from the vacuum/B
4
C and B
4
C/substrate interfaces.
As the energy increases further, the absorption in the layer increases significantly (for 188.5
eV, 192 eV, and 200 eV) and, hence, the amplitude of intensity modulation is only observed
at higher q
z
values. The nearly step fall of reflected intensity at the critical angle is clearly
observed for 173 eV and 180 eV, due to total external reflection (which is characteristic of x-
rays) and the low absorption in the layer at these energies. In the anomalous range where real
part of refractive index (1-) is positive (for 187.5 eV and 188.5 eV), total external reflection
could not occur at any angle and, hence, no critical angle exists. For the energy 192 eV and
200 eV, the absorption in the layer becomes significant and, hence, the reflected intensity at
the critical angle falls gradually.
Chapter 6
187
The optical constants obtained from the best-fit angle-dependent soft x-ray
reflectivity measurements are shown in Table 6.5, along with tabulated values from Henke et
al.
21
Far from the absorption edge (for 173 eV, 180 eV and 200eV), our values of optical
constants agree well with those of Henke et al., within experimental error.

At the absorption edge, our values of optical constants are significantly different from those
of Henke et al. At the absorption edge, these discrepancies in optical constants may be due to
rapid variation, and/or extreme sensitivity to slight error in energy calibration or resolution,
and/or uncertainty in the tabulated values of Henke et al. near the absorption edge. Henke et
Energy (eV)
B4C
(expt.)
B4C
(Henke)
B4C
(expt.)
B4C
(Henke)
173 0.0059 0.00611 0.0009 0.00082
180 0.0039 0.00414 0.0005 0.00057
185.3 0.00001 0.00085 0.0003 0.00052
187.5 -0.0082 -0.00675 0.0007 0.00051
188.5 -0.0059 -0.00381 0.0082 0.00921
192 0.0014 0.00226 0.0095 0.00869
200 0.0046 0.00505 0.0081 0.00766
Table 6.5 The best-fit results of soft x-ray reflectivity measurements of a B
4
C film.
Fig. 6.10 Soft x-ray reflectivity spectra of a B
4
C film with photon energy at the boron
K-absorption edge. The best-fit results are given in Table 6.5.
0.00 0.03 0.06 0.09 0.12 0.15 0.18
10
-5
10
-2
10
1
10
4
10
7
10
10
10
13
B
4
C (805) thin film
Measured
Fitted
200 eV
192 eV
188.5 eV
187.5 eV
185.3 eV
180 eV
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
173 eV
Chapter 6
188
al. measured using transmission measurements and calculated from through Kramers-
Kronig integral relations. Therefore, at the absorption edge, the tabulated values of optical
constant by Henke et al. may be inaccurate due to the missing oscillator strength in the
absorption co-efficient data used.

B
4
C-on-Fe bi-layer film:

The reproducibility of the results obtained from SXRRR measurements on a B
4
C film
has been checked through the characterization of a B
4
C-on-Fe bi-layer structure as follows.
Fig. 6.11 shows angle-dependent soft x-ray reflectivity spectra of a B
4
C (600)- on-Fe
(180) bi-layer film at different photon energies, which are the same as in Fig. 6.9 for the
B
4
C film. Again, for the fitting, the microstructural parameters (such as thickness, density
and roughness) obtained from hard XRR (Table 6.4) were used, and the optical constants (
and ) of B
4
C were taken from our experimentally determined values (Table 6.5), and those
of Fe were taken from Henke et al. (Fig. 6.9), as the initial guess. The best-fit results of
optical constants are tabulated in Table 6.6. In Fig. 6.11, the bilayer effects in interference
pattern are clearly visible for 173 eV, 180 eV and 187.5 eV, due to high optical contrast,
indicating strong sensitivity to B
4
C/Fe interface. At 173 eV and 187.5 eV, the amplitude of
modulation is larger compared to 180 eV, suggesting relatively greater sensitivity to B
4
C/Fe
interface. Due to high sensitivity to B
4
C/Fe interface at theses energies, the best-fit results
clearly suggest that an intermediate layer is formed at the B
4
C/Fe interface with density of ~
65 % of pure Fe layer. This low-density layer (thickness of 1.3 nm) may arise due to porosity
at the top of Fe layer. The high frequency oscillation gets modulated over the low frequency
oscillation marked by vertical dotted line. The high frequency amplitude oscillation with
0096 . 0
4
=
C B
z
q
-1
corresponds to the thicker B
4
C layer, where as the low frequency
amplitude oscillation with 031 . 0 =
Fe
z
q
-1
corresponds to the thinner Fe layer. At 185.3
eV, only low frequency oscillation is observed which corresponds to Fe layer. At this energy,
optical constants 0 and 0 for B
4
C (Table 6.6). This corresponds to optical constant for
vacuum (=0 and =0). Therefore, B
4
C layer is virtually invisible. Hence amplitude
modulation arises due to reflected beam from vacuum/Fe and Fe/substrate interface. At 188.5
eV, 192 eV and 200 eV bilayer effect is not observed due to significant absorption of
radiation in B
4
C layer (Fig.6.9). As the energy increases, amplitude of intensity modulation is
Chapter 6
189
observed at higher q
z
values due to increased penetration depth. The amplitude of oscillation
is more at 188.5 eV compared to 192 eV and 200 eV, due to relatively high reflected
intensity of vacuum/ B
4
C and B
4
C/Fe interface. This is due to higher optical contrast at this
energy (Fig. 6.9). The interface tuned selectively by adjusting the incident photon energy and
reveals greater sensitivity to particular interface due to high tunable optical contrast which is
not possible using conventional hard x-ray reflectivity.

From Table 6.6, it may be seen that our results for the optical constants match well with
those of Henke et al. for Fe, for all the energy, within experimental error. For B
4
C, our
optical constants are at variance with respect to Henke et al. for different photon energies, in
a manner to the case of the B
4
C film discussed above.

Fig. 6.11 Soft x-ray reflectivity spectra of a B
4
C-on-Fe bi-layer film, with photon
energy tuned to the boron K- edge. The best-fit results are given in Table 6.6.
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21
10
-5
10
-2
10
1
10
4
10
7
10
10
10
13
Measured
Fitted
200 eV
192 eV
188.5 eV
187.5 eV
185.3 eV
180 eV
R
e
f
l
e
c
t
i
v
i
t
y
q
z
(
-1
)
173 eV
B
4
C(600 ) on Fe (180 )
Chapter 6
190

Fig. 6.12 shows in-depth profile of dispersion part of refractive index () as obtained
from best-fit of the reflectivity measured data. For Fe, is positive and decreases
systematically as the energy increases. Whereas for B
4
C, shows anomalous effect near
boron K-absorption edge (E=188 eV). At energy 173 eV and 180 eV, is positive and hence
total external reflection occurs which is observed from the step fall of reflected intensity at
Energy
(eV)
B4C
(expt.)
B4C
(expt.)
Fe
(expt.)
Fe
(Henke)
Fe
(expt.)
Fe
(Henke)
173 0.0058 0.0008 0.0256 0.02582 0.0152 0.01482
180 0.0038 0.0006 0.0242 0.02439 0.0139 0.01382
185.3 0.00001 0.0003 0.0233 0.02349 0.0131 0.01292
187.5 -0.0084 0.0007 0.0227 0.02312 0.0122 0.01256
188.5 -0.0058 0.0084 0.0224 0.02295 0.0121 0.01238
192 0.0015 0.0097 0.0219 0.02234 0.0116 0.01198
200 0.0048 0.0082 0.0213 0.02092 0.0112 0.01058
Table 6.6 The best-fit results of soft x-ray reflectivity measurements of B
4
C-on-Fe bi-
layer film.
Fig. 6.12 In-depth variation of dispersion part of refractive index () obtained after
fitting of the measured soft XRR profile of Fig. 6.11.
0 225 450 675 900
-0.010
-0.005
0.000
0.005
0.010
0.015
0.020
0.025
Substrate B
4
C

173 eV
180 eV
185.3 eV
187.5 eV
188.5 eV
192 eV
200 eV
Depth in
Fe
Chapter 6
191
the critical angle 2
c
(Fig. 6.11). At particular energy of 185.3 eV, 0 correspond to
for vacuum, and no external reflection occurs. At energies of 187.5 eV and 188.5 eV, is
negative (anomalous region). Thus total external reflection does not occur at any angle, and
no critical angle is observed (Fig. 6.11). As energy increases, at energies 192 eV and 200 eV,
becomes positive. At these energies absorption becomes significant (Fig. 6.9) and hence the
reflected intensity at the critical angle shows a gradual fall (Fig.6.11). In Table 6.6, our
results of the optical constants for Fe matches with Henke et al. for all the energies within
experimental error of ~2%. In Fe/B
4
C bilayer, the optical constants of B
4
C vary within
~2.1% with respect to pure B
4
C thin film.
Chapter 6
192
6.6 Conclusions

In this work, we have presented the case that the SXRRR technique can be exploited for
the characterization of low-Z thin films and multilayers. In one study, it is demonstrated that
SXRRR opens a non-destructive technique to the analysis of interlayer composition with sub-
nanometer scale sensitivity. The methodology has been examined for sensitivity to
composition at buried interfaces through the characterization of Mo/Si multilayers. This is
done by tuning the photon energies near the Si L-edge, both in simulations and in
experiments using the Indus-1 synchrotron source. It is shown clearly that SXRRR has
greater sensitivity to composition of interlayers than normal hard x-ray measurements. Our
best-fit soft x-ray results reveal that the MoSi
2
composition is formed at both the interfaces,
in good agreement with results obtained using depth-graded XPS measurements.
In the second study, we have provided experimental evidence of elemental specificity
property of SXRRR by observing the effect of anomalous optical constants on the angle-
dependent specular SXRRR spectrum. The optical constants of B
4
C obtained from angle-
dependent specular reflectivity measurements, and fitted to Fresnel equations, show
anomalous behavior near the boron K-edge. In the anomalous region, the real part of the
atomic scattering factors undergoes sign reversal (from positive to negative values).
Therefore, a material becomes virtually invisible if the photon energy is tuned such that the
refractive index of the material is very close to the refractive index of vacuum. The buried
interfaces are tuned selectively due to high and tunable optical contrast using resonant
reflectivity, which is a not possible using conventional XXR measurement. We demonstrate
this elemental specificity of SXRRR through the characterization of the B
4
C material in B
4
C
thin films and in Fe/B
4
C bi-layers, through experiments using the Indus-1 synchrotron
radiation source.
Chapter 6
193
6.6 References

1.
H. Kiessig, Ann. Phys. (Leipzig) 10, 715 and 769 (1931).

2.

3.
L. Nevot and P. Croce, Rev. Phys. Appl.15, 761 (1980).

4.
S.K. Sinha, E.B. Sirota, S. Garoff, H. B. Stanley, Phys. Rev. B 38, 2297 (1988).

5.
O.G. Shpyrko, A.Yu. Grigoriev, R. Streitel, D. Pontoni, P S. Pershan, M. Deutsch, B.
Ocko, M. Meron, B. Lin, Phys. Rev. Lett. 95, 106103 (2005).

6.
C.-C. Kao, C.T. Chen, E.D. Johnson, J.B. Hastings, H.J. Lin, G.H. Ho, G. Meigs, J.-
M. Brot, S.L. Hulbert, Y.U. Idzerda and C. Vettier, Phys. Rev. B 50, 9599 (1994).

7.
Resonant Anomalous X-ray Scattering: Theory and Applications, edited by G.
Materlik, C.J. Sparks, K. Fischer (North-Holland, Amsterdam, 1994).

8.
Anomalous scattering, edited by S. Ramaseshan and S.C. Abrahams (Munksgaard,
Copenhagen, 1975).

9.
H. Stragier et al. Phys. Rev. Lett. 69, 3064 (1992).

10.
E. D. Specht and F.J. Walker, Phys. Rev. B 47, 13743 (1993).

11.
S. Roy, M.R. Fitzsimmons, S. Park, M. Dorn, O. Petracic, I.V. Roshchin, Z-P. Li, X.
Batlle, R. Morales, A. Misra, X. Zhang, K. Chesnel, J.B. Kortright, S.K. Sinha, I.K.
Schuller, Phys. Rev. Lett. 95, 047201 (2005).

12.
D. Vaknin, P. Kruger, M. Losche, Phys. Rev. Lett. 90, 178102 (2003).

13.
C. Park, P.A. Fenter, N.C. Sturchio, J.R. Regalbuto, Phys. Rev. Lett. 94, 076104
(2005).

14.
C. Wang, T. Araki, H. Abe, Appl. Phys. Lett. 87, 214109 (2005).

15.
S.K. Sinha, M.K. Sanyal, B.L. Carvalho, M. Rafailovich, J. Sokolov, X. Zhao, and
W. Zhao, Resonant Anomalous X-ray Scattering: Theory and Applications, (Elsevier
Science, Amsterdam. 1994) p. 421.

16.
D.H. Kim, H.H. Lee, S.S. Kim, H.C. Kang, D. Y. Noh, H. Kim, S. K. Sinha, Appl.
Phys. Lett. 85, 6427 (2004).

17.
J. Strzalka, E. Dimasi, I. Kuzmenko, T. Gog and J.K. Blasie, Phys. Rev. E 70, 051603
(2004).

18.
D. Attwood, Soft x-rays and extreme ultraviolet radiation: Principles and Application
(Cambridge University Press, 1999).

Chapter 6
194
19.
R.W. James, The Optical Principles of the Diffraction of X-rays (Bell and Sons,
London, 1948), pp.74-75.

20.
M. Benfatto and R. Felici, Phys. Rev. B. 64, 115410 (2001).

21.
B.L. Henke, E.M. Gullikson, J. C. Davis, Atomic Data and Nuclear Data Tables 54,
181, (1993); and at http://www-cxro.lbl.gov/optical_constants/.

22.
J. Tnnerre L. Seve, D. Raoux, G. Soullie, B. Rodmacq, P. Wolfers, Phys. Rev. Lett.
75. 740 (1995).

23.
E. Spiller Soft X-ray Optics (SPIE Optical Engineering Press, Washington,D.C.
1994).

24.
E. Filatova, V. Lukyanov, R. Barchewitz, J.M. Andre, M. Idir, Ph. Stemmler, J.
Phys.: Condensed Matter 11, 3335 (1999).

25.
T.W. Barbee, Jr., W.K. Warburton, J.H. Underwood, J. Opt. Soc. Am. B1, 691
(1984).

26.
R. Soufli and E.H. Gullikson, Appl. Opt. 36, 5499 (1997).

27.
P. Tripathi, G.S. Lodha, M.H. Modi, A.K. Sinha, K.J.S. Sawhney and R.V.
Nandedkar, Optics Communications 211, 215 (2002).

28.
D. E. Aspnes, Handbook of Optical Constants of Solids, Edited by E.D. Palik
(Academic Press, London UK, 1998) pp.89-112.

29.
J. Pfluger and J. Fink, in Handbook of Optical Constants of Solids, Edited by E.D.
Palik (Academic Press, London, UK, 1998) pp.293-311.

30.
D.P. Siddons, in Proceedings of the Topical Conference for Soft X-ray Diagnostics,
edited by D.T. Attwood and B.L. Menke (AIP, New York, 1981).

31.
J.D. Jackson Classical Electrodynamics (Wiley, New York, 1998).

32.
H. Winick (editor) Synchrotron Radiation Source (World Scientific, Singapore,
1994).

33.
S.S. Ramamurthi and G. Singh, Nuclear Instrumentation and Methods, A359, 15
(1995).

34.
G. Singh et al. Indus-1 Activity Report (2003).

35.
J.C. Chen, G.H. Shen and L.J. Chen, J. Appl. Phys. 83, 7653 (1998).

36.

37.
S.P. Murarka, Silicides for VLSI Applications (Academic Press, London, 1983).

Chapter 6
195
38.
R.V. Nandedkar, K.J.S. Sawhney, G.S. Lodha, A. Verma, V.K. Raghuvanshi, A.K.
Sinha, M.H. Modi, M Nayak, Current science 82, 298 (2002).

39.
Smithells Metals Reference Book edited by E.A. Brandes and G.B. Brook
(Butterworth-Heinemann, Jordan Hill, Oxford, UK, 1992) p.11-376.

40.
M. Altarelli, D.L. Dexter, H.M. Nussenzveig, D.Y. Smith, Phys. Rev. B 6, 4502
(1972).

41.
Y. Weseda, Novel Applications of Anomalous (Resonance) X-ray Scattering for
Structure Characterization of Disordered Materials (Springer-Verlag, Berlin,
1984)).

Chapter 7
196

CHAPTER-7

Summary and Conclusions

This chapter summarizes the main findings of the research carried out for the present
thesis. An outline is provided for the work that might continue this investigation in the future.
Chapter 7
197
7.1 Summary and Conclusions

The fabrication of multilayer (ML) mirrors for the soft x-ray/extreme ultraviolet part
of the spectrum is a challenging task from practical (optical device) point of view. These ML
coatings require a periodicity d in the range of a few nanometers with tens of layer pairs in
them. The thickness error and boundary roughness should be in the range d/10 to yield high
normal-angle reflectivity with moderate spectral width at Braggs angle in the soft x-
ray/EUV spectral range. Therefore, the fabrication of high quality soft x-ray ML mirrors
require precise control (angstrom level) over layer thickness and uniformity over a large
substrate area, minimum contamination and stress in the layers, and atomically sharp, well-
defined interface boundaries. Therefore, a very tight control of the deposition process is
essential. We have developed indigenously an ultra-high vacuum electron beam evaporation
system for the fabrication of these ML mirrors.
The quality of surfaces and interfaces plays an important and decisive role in
achieving optimum performance of a ML structure. As part of the present thesis work, a
detailed study was carried out to investigate the nature of interfaces in soft x-ray MLs
developed for the polarization applications on the Indus-1 synchrotron source. We have also
carried out a detailed study of the growth of thin Mo films so as to achieve optimum
performance in Mo-based MLs. In addition, we have presented the possibility of determining
interlayer composition at buried interfaces, using soft x-ray resonant reflectivity produced in
the Indus-1 synchrotron source. We have demonstrated the elemental sensitivity of soft x-ray
resonant reflectivity, as well.

Indigenously Developed UHV Electron Beam Evaporation System

The main motivation behind developing this specially designed electron beam system
has been the fabrication of high quality multilayer thin film structures, the study the basic
physics of MLs and, finally, the development of x-rays optical elements for the Indus-1 and
Indus-2 synchrotron radiation sources. All the required hardware, including the deposition
chamber, load-lock chamber, shutter arrangements, and substrate latching/detaching system,
were designed and fabricated in house. Provision was made for a large number of extra
vacuum ports to accommodate different accessories for different thin film-related
experiments. The system has a 6 kW multi-packet (3-crucible) electron gun. The system has
facilities of sample rotation through a rotary feedthrough. Automatic control of shutter and
Chapter 7
198
the thickness monitor is implemented through a PC. Films (layers) of the desired thickness,
with thickness errors <0.2%, can be deposited in the custom-made e-beam system. The
system is able to coat substrates as large as 20 cm 10 cm, with thickness non-uniformity of
less than 1% at a source-to-substrate distance ~75 cm. Provision has been made for the
installation in the future of an in situ reflectivity measurement apparatus. All the thin films
and other ML structures used in the present thesis research were fabricated using this e-beam
system. (This system is now being used regularly as the thin film/multilayer fabrication
facility for users of the Indus-1/Indus-2 synchrotron sources.)

Study of Molybdenum UltraThin Film Growth

To be able to achieve the optimum (small) thickness of Mo in Mo-based ML
structure, we have studied in detail the growth of ultra-thin Mo films using in situ sheet
resistance measurements, which has apparently not been employed previously. The nature of
thin film growth and a detailed microscopic picture of the different stages of growth are
derived from modeling the sheet resistance data obtained in situ. The various conduction
mechanisms operating at different stages of film growth have been identified. In the island
growth regime, the conductance (G) has two exponential dependencies on thickness ( ) t :
t B G
1
log and t B G
2
log , explained as anisotropic and isotropic growth of
islands, respectively. In the insulator-to-metal transition regime, the conductance of the films
follows the scaling law ( )
q
c
t t G , where q and t
c
are critical exponent and critical
thickness, respectively. The value of q agrees well with the theoretically predicted values for
critical exponent of conductivity in two-dimensional percolating systems. The microstructure
of films in the growth stage, derived from in situ sheet resistance measurements is in good
agreement with that predicted by percolation theory. The minimum thickness required for
continuous film formation is 1.8 nm and 2.2 nm for Mo films deposited at a substrate
temperature of 300 K and 100 K, respectively. We provide also experimental evidence for
the universality of the conductivity critical exponent in 2D percolating systems by showing
the identity of the conductivity critical exponent in the two geometries studied. An
amorphous-to-crystalline transition is also observed.

Chapter 7
199
Study of Multilayer Structures for soft X-rays

In the present thesis work, three multilayer systems viz., Mo/Si, Fe/B
4
C, and Mo/Y
have been studied in detail for their application as polarizing elements on the Indus-1
synchrotron source. Greater emphasis has been placed on the Mo/Si ML system. Mo/Si MLs
were fabricated with varying periodicity (d), number of layer pairs (N), and varying ratio ()
of the thickness of the high-Z layer to the period of the ML. In all the Mo/Si MLs fabricated,
interlayers are formed due to inter-diffusion of the two materials, Mo and Si, at the
interfaces. The thickness of the interlayer is asymmetric in nature, i.e., Mo-on-Si thickness is
different from the Si-on-Mo thickness. This is dealt with by employing a four-layer model of
the x-ray reflectivity data. We have presented a systematic simulation of the effect of
interfacial roughness and depth-graded interlayers on the reflectivity profile, and studied the
different interlayers possible. A possible mechanism for the asymmetry of the interlayer is
proposed. The composition of the interlayers has been determined using depth-graded XPS,
which reveals the formation of the MoSi
2
composition on both the interfaces. The
experimental results agree well with theoretical calculations based on thermodynamics.
Finally, Mo/Si ML-based soft x-ray reflection polarizing elements were fabricated with the
desired period for operation in the wavelength range 125 -160 (above the Si L-edge). The
reflectivity performance in the s-polarization geometry was tested on the Indus-1
reflectometry beam line. At 130 , a maximum reflectivity of 45% at the Brewsters angle
was measured from a ML with 30 layer pairs (N).
In the Fe/B
4
C system, an attempt was made to understand the interface
characteristics. A study of the surfaces and interfaces in B
4
C-on-Fe and Fe-on-B
4
C bi-layers,
and in Fe-B
4
C-Fe tri-layers was carried out. It was observed that a low density (w.r.t. Fe)
layer forms at the B
4
C-on-Fe interface, whereas no such layer is formed at the interface in the
Fe-on-B
4
C. As the bottom Fe layer thickness increases in the B
4
C-on-Fe bi-layer, the
thickness of this interlayer thickness increases, but its density decreases. The nature of the
interlayer is independent of the top B
4
C layer thickness. The results obtained have been
interpreted on the basis that the formation of the interlayer in Fe/B
4
C stems from
topographical roughness. Finally, high-quality Fe/B
4
C MLs have been fabricated with
varying layer thicknesses and number of layer pairs. A study of surfaces and interfaces
studies in these structures has been are carried out in the context of their application in
polarizing elements for wavelengths above the boron K-edge.
Chapter 7
200
In the Mo/Y system, too, a systematic study of surfaces and interfaces was carried out
on bi-layers, tri-layers and MLs, revealing that no interlayer is formed at the various
interfaces. A layer of yttrium oxide of thickness ~20 is formed if Y is the top layer, and a
molybdenum oxide layer of thickness ~13 is formed if Mo is top layer in these structures,
both due to exposure to air. XRR results reveal the presence of a cumulative- type roughness
in evaporated Mo/Y MLs.

Soft X-ray Resonant Studies

In the present thesis, the fine structure features of energy-dependent atomic scattering
factor near the atomic absorption edges have been explored for the characterization of low-Z
thin films and multilayer structures. In one study, we have demonstrated for the first time that
soft x-ray resonant reflectivity can be employed as a non-destructive technique for
determining the interlayer composition at buried interfaces on a sub-nanometer scale. Near
absorption edges, the atomic scattering factors are sensitive to composition, and can be
exploited for compositional analysis at buried interfaces. We have demonstrated this for a
well-characterized Mo-Si multilayer system, using simulations and soft x-ray resonant
reflectivity measurement where the incident photon energy is tuned to near the Si L-edge (on
the Indus-1 synchrotron source). This can in general be applied to any low-Z bi-layer, tri-
layer and ML.
In a second study, the anomalous behavior of the optical constant of B
4
C at the boron
K-edge has been observed using angle-dependent specular soft x-ray reflectivity
measurements. We have obtained experimental evidence of the elemental specificity of soft
x-ray resonant reflectivity by showing how the anomalous variations of the optical constants
affect the specular soft x-ray resonant reflection spectrum. A layer of B
4
C becomes virtually
invisible when the photon energy is tuned such that the refractive index of B
4
C is very close
to the refractive index of vacuum. We have demonstrated this invisibility through the
characterization of B
4
4
C bi-layers using synchrotron radiation on the
Indus-1 synchrotron source. We have also demonstrated the possibility of tuning selected
buried interfaces due to high and tunable optical contrast using resonant reflectivity, which is
not possible using conventional x-ray reflectivity.

Chapter 7
201
7.2 Scope for Investigation in the Future

The soft x-ray/ extreme ultra violate region of the electromagnetic spectrum offers
great opportunities in both science and technology. The shorter wavelength allows one to see
smaller features in microscopy and write finer features in lithography. Development of x-ray
ML mirrors is playing an important role in the exploitation of the soft x-ray/extreme ultra
violet (XUV) region of the electromagnetic spectrum. High reflectivity with moderate
spectral bandwidth at normal/near-normal incidence can be achieved by using these ML
mirrors. The depth-graded multilayers (known as super mirrors) are used as wide-band
reflectors, whereas laterally-graded MLs are employed as multi-wavelength reflectors. The
great advantage of MLs is that they can be fabricated on figured (contoured) surfaces for
imaging applications. The Indus-1 is a 450 MeV synchrotron source, which generates
radiation up to the soft x-ray range, and is operational. Indus-2 is a 2.5 GeV machine, which
has been commissioned recently to give radiation upto hard x-ray regime (E>25 keV). The
utilization of synchrotron radiation is becoming wide spread not only in physics, but in all
areas of science for basic research, and in technological applications. The development of x-
ray optical elements is a front-line research area for the realization of above applications.
Mo/Si, Fe/B
4
C, and Mo/Y ML structures have been designed for optimum reflectivity
for polarization applications in the wavelength range 67 -160 on the Indus-1 synchrotron
source. The performance of Mo/Si MLs at soft x-ray wavelengths has been tested for
polarization applications, but a polarimeter has to be designed using these MLs. In the Mo/Si
ML structure, the interlayer formation at interface reduces the reflectivity. The formation of
this interlayer can be prevented by placing a layer of carbon of optimum thickness between
Mo and Si. In the present study, Mo/Si MLs have been designed to reflect at Brewsters
angle. For x-ray lithography and Schwachild microscopy applications, a detailed study of
Mo/Si MLs is required to design them for normal-incidence reflectivity at wavelengths above
the Si L-edge. Theoretical work on such designs shows that Fe/B
4
C and Mo/Y ML structures
have high reflectivity at wavelengths above the boron K-edge and the Y K-edge,
respectively, at Brewsters angle. The surfaces and interfaces of these systems have been
investigated in detail in the present thesis. It would be interesting to evaluate actual
performance of theses MLs in the desired soft x-ray wavelength region at Brewsters angle.
A combination of Indus-1 and Indus-2 will provide radiation in the broad energy
spectrum from IR to hard x-rays. Thus, there is a very rewarding opportunity to study MLs
Chapter 7
202
with different material pairs, and with different parameters, such as periodicity, optimum
thickness of individual layer pairs and . For the Indus-2 applications, it would be interesting
to study depth-graded and laterally-graded MLs. It would also be interesting to study the
thermal stability of MLs for the high heat-load applications in the Indus-2 source. The
fabrication and characterization of ultra-short period MLs (d=15-20) needs to be investigated
for microscopy applications in the water window region.
In this present work, we have established a method to determine the minimum
thickness for which a film becomes continuous, applied it to Mo films, to find the
discontinuous-to-continuous transition and to study film growth. These are important to
optimizing the thickness of Mo layers in Mo-based MLs. In a similar manner, the growth of
thin films of other high-Z materials needs to be investigated for the optimization of high-Z
thickness in other material pair combinations.
Finally, we have demonstrated the possibility of characterization of low-Z thin films
and their interfaces in the multilayer structures, using soft x-ray resonant reflectivity.
Specifically, we have shown that soft x-ray resonant reflectivity can be a non-destructive
technique for the determination of interlayer composition. We have also established the
elemental specificity of soft x-ray resonant reflectivity through the effect of anomalous
optical constant on reflectivity patterns. Based on this work, the depth profiling of
composition at the interfaces using soft x-ray resonant reflectivity needs to be investigated
further. It is also appropriate to investigate the sensitivity of soft x-ray resonant reflectivity
with known multi-phase composition interlayers. It would also be interesting to identify the
position of low-Z marker layers, especially between other low-Z layers, using soft x-ray
resonant reflectivity. (Hard x-ray reflectivity would not in general be able to do so because of
poor optical contrast among the low-Z materials.)
List of Publications
203

LIST OF PUBLICATIONS
A. In Refereed Journals
Thesis work:
1. Study of low atomic number containing hard condensed matter thin films and their
interface nature using soft x-ray resonant reflectivity
M. Nayak and G. S. Lodha, J. Appl. Phys. (submitted).
2. Study of optical response near the absorption edge using vacuum ultra violet/ soft x-
ray reflectivity beamline on Indus-1 synchrotron
G S Lodha, M Nayak and R V Nandedkar, J. Phys. C: Condensed Matter (Accepted).
3. Design, development and performance of an UHV electron beam evaporation system
for thin films and x-ray multilayer mirrors
M. Nayak, G. S. Lodha and B. Gowrishankar, Asian J. Phys. 16, 395 (2007).
4. Determination of X-ray compression efficiency of a thin film X-ray waveguide
structure using marker layer fluorescence
M. K. Tiwari, M. Nayak, G. S. Lodha and R. V. Nandedkar, Spectochimica Acta
part B 62, 137 (2007).
5. Phase composition determination at buried interfaces using soft x-ray resonant
reflectivity
M. Nayak, G.S. Lodha, A.K. Sinha, R.V. Nandedkar, S.A. Shivashankar Appl.
Phys. Lett. 89, 181920 (2006).
6. Nucleation, growth, percolation and crystallization transitions of ultra thin
molybdenum films
M. Nayak, G.S. Lodha, R.V. Nandedkar, J. Appl. Phys. 100, 113709 (2006).
7. X-ray reflectivity investigation of interlayer at the interfaces of x-ray multilayer
structures: application to Mo/Si multilayers
M. Nayak, G.S. Lodha, R.V. Nandedkar, Bulletin of Materials Science, 29, 1
(2006).
8. Interlayer composition in Mo/Si multilayers using x-ray photoelectron spectroscopy
M. Nayak, G.S Lodha, A.K Sinha, R.V Nandedkar J. Electron Spectroscopy and
Related Phenomena, 152, 115 (2006).
Others:
9. Characterization of molybdenum/silicon x-ray multilayers
204
M Nayak, M.H. Modi, G.S. Lodha, A.K. Srivastava, P. Tripathi, A.K. Sinha,
K.J.S. Sawhney, R.V. Nandedkar, Nucl. Instr. Methods B 199, 128 (2003).
10. Determination of layer structure in Mo/Si multilayers using soft x-ray reflectivity
M.H. Modi, G.S. Lodha, M. Nayak, A.K.Sinha, R.V. Nandedkar, Physica B 325, 272,
(2003).
11. Formation of Mo
5
Si
3
phase in Mo/Si multilayer
A.K. Srivastava, P Tripathi, M. Nayak, G.S. Lodha, R.V. Nandedkar, J. Appl.
Phys. 92, 5119, (2002).
12. First results on the reflectometry beam line on Indus-1
R.V. Nandedkar, K.J.S. Sawhney, G.S. Lodha, A. Verma, V.K. Raghuvanshi,
A.K. Sinha, M.H. Modi, M Nayak, Current Science 82, (2002).
13. High-temperature studies on Mo-Si multilayers using transmission electron
microscopy
A.K. Srivastava, P. Tripathi, M. Nayak, G.S. Lodha, R.V. Nandedkar, Current
Science 83, 997, (2002).
14. Characterization of laser and laser/thermal annealed semiconducting iron silicide
thin films
A. Datta, S. Kal, S. Basu, M. Nayak and A. K. Nath, J. of materials science:
Materials in electronics, 10, 627, (1999).

B. In Conference/ Symposium Proceedings
Thesis Work:

1. Soft x-ray resonant reflectivity as non-destructive technique for phase composition at
sub-nanometer scale buried interfaces
M. Nayak, G.S. Lodha, A.K. Sinha, R.V. Nandedkar, S.A. Shivashankar,
Asian/Oceanic Forum for Synchrotron Radiation Research, Nov. 2006, Tsukuba,
Japan (presented).

2. Beam lines on Indus I and II for x-ray multilayer optics and micro fabrication research
G.S. Lodha, V.P. Dhamgaye, M.H. Modi, M. Nayak, A.K. Sinha, R.V. Nandedkar, AIP
conference Proc. 879, 1474 (2007).
205
3. Indigenously developed electron beam evaporation system for fabrication of thin films and x-
ray multilayer optics
M. Nayak, G.S. Lodha, B. Gowrishankar, R.V. Nandedkar, Proceedings of the DAE
Solid State Phys. Symposium 51, 493 (2006).
4. Mo/Si multilayer based soft x-ray reflection polarizing elements for Indus-I
M. Nayak, A.K. Sinha, G.S. Lodha, R.V. Nandedkar: Proceedings of the DAE Solid
State Phys. Symposium 50, 391 (2005).
Others:
5. Effect of substrate temperature on the percolation threshold and rms roughness of
molybdenum thin films
M. Nayak, G.S. Lodha, R.V. Nandedkar, Proceedings of the DAE Solid State Phys.
Symposium 46, 449 (2003).
6. Transmission electron microscopy studies on Mo/Si Multilayer
A.K Srivastava, P. Tripathi, M Nayak, D.V. Shridhar Rao, K Muraleedharan, G.S.
Lodha, R.V. Nandedkar, Presented at the National Symposium on Science & Technology
of Vacuum and Thin Films IVSNS September 2001, Bangalore, India.
7. Percolation threshold of Molybdenum in Mo/Si multilayers
M. Nayak, G.S. Lodha, M.H. Modi, A. Gupta, Presented at the National Symposium on
Science & Technology of Vacuum and Thin Films IVSNS September 2001, Bangalore, India.
8. Characterization of Soft x-ray/Extreme Ultra-Violet Multilayers using INDUS-1 M.H.Modi,
G.S.Lodha, M.Nayak, A.K.Sinha, K.J.S.Sawhney, R.V.Nandedkar, Presented at the National
Symposium on Science & Technology of Vacuum and Thin Films IVSNS September 2001,
Bangalore, India.
9. Soft X-ray Reflectivity Measurements of Amorphous Carbon Thin Films Using
Indus-1
M.H. Modi, G.S. Lodha, M. Nayak, A.K.Sinha, R.V. Nandedkar, DAE Solid State Physics
Symposium 2001, Mumbai, India.

206
CURRICULUM VITAE

Present Affiliation: Scientific Officer
X-ray Optics Section
Synchrotron Utilization and Materials Research Division
Raja Ramanna Centre For Advanced Technology (RRCAT)
Department of Atomic Energy, Indore- 452013, M. P., INDIA
With RRCAT since 1999
E-mail: mnayak@cat.ernet.in
Persional Details:
Date of Birth: 4
th
April 1975
Educational Qualification:
1997: M. Sc. (Physics) from Utkal University, Orissa, INDIA
1999: M. Tech. (Materials Science and Engg.) from Indian Institute of Technology,
Kharagpur, India
Membership of Professional Bodies: Indian Physics Association
Experiences
1. Fabrication of thin films and multilayer optics for x-rays.
2. Design and development of ultra-high vacuum electron beam evaporation system. Experience
in depositions using sputtering systems.
3. Thin film technology related to the study of ultra-thin films, to understand the basic physics of
film growth, and to optimise the performance of multilayer structures for use as x-ray optical
elements.
4. Instrumentation for ultra-high vacuum environment.
5. Experience in setting up a soft x-ray reflectometer on Indus-1 synchrotron source and in
performing experiments using synchrotron radiation.
6. Surface and interface studies of thin film multilayers using hard and soft x-rays, specifically
using x-ray reflectivity and resonant x-ray reflectivity.
7. Investigation of interfacial behaviour of multilayers with different material combinations, e.g.,
Mo /Si, Mo/Y, Fe/B
4
C etc. x-ray for polarizer applications.
8. Fundamental research using synchrotron radiation, particularly soft x-ray resonant reflectivity,
for the study low-Z thin films and multilayers.
9. Numerical analysis of experimental data.
Employment Record:

1999 -- Continuing: Scientific Officer at Raja Ramanna Centre For Advanced Technology,
Government of India, Indore, India.

PhDthesis Nayak PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PhDthesis Nayak PDF

Uploaded by

Copyright:

Available Formats

Soft X-ray Multilayers as Polarizing Elements:

Fabrication, and Studies of Surfaces and

, respectively, make equal angles with the surface boundary, i.e.,

|=2. The change in momentum transfer

-k. The magnitude of momentum

, with incidence angle and exit angle

-k, and can be resolved into z- and x-components, taking into

value, and the sample is

propagate in vacuum, they are real and equal in magnitude

and turning angle

, in the medium, have real and imaginary components.

vectors are perpendicular to each other. In

, each radiating a sin

= . In modern XRR evaluation

radiation (=1.54 ) for single layer thin films of Mo (density =10.2

radiation is employed because of basically divergence of beam limit.

= . In the multilayer structure, the structure factor or the form factor

= . These Kiessig oscillations arise due to

wavelength (=1.542 ). Roughness being zero means

wavelength. The effect of different combinations of

wavelength (=1.542 ) for a

radiation. A relative error

, whereas an ordinate intercept yields the specular parameter. The values of

). The concentric hemispherical analyzer with has a

source. Fig.5.12 shows the survey scans of Mo/Si

wavelength for two different Fe/B

wavelength for two different Fe(20 )/B

. Thus, the refractive index has a strong frequency dependence.

energy (E=8047 eV).

energy (E=8047 eV). The inset shows

You might also like