TEL AVIV UNIVERSITY THE IBY AND ALADAR FLEISHMAN FACULTY OF ENGINEERING

Developing of Advanced Sputtering Methods

By

Alexander Axelevitch

THESIS SUBMITTED DOR THE DEGREE OF "DOCTOR OF PHYLOSOPHY"

SUBMITTED TO THE SENATE OF TEL-AVIV UNIVERSITY

December 2001

TEL AVIV UNIVERSITY THE IBY AND ALADAR FLEISHMAN FACULTY OF ENGINEERING

Developing of Advanced Sputtering Methods

By

Alexander Axelevitch

THESIS SUBMITTED DOR THE DEGREE OF "DOCTOR OF PHYLOSOPHY"

SUBMITTED TO THE SENATE OF TEL-AVIV UNIVERSITY

This Research Work was Carried Out at Tel-Aviv University in The Faculty Of Engineering

Under The Supervision of Prof. Yossi Rosenwaks and Dr. Gady Golan

December 2001

ii

This work was carried out under the supervision of

Prof. Yossi Rosenwaks

and

Dr. Gady Golan

Department of Physical Electronics

Faculty of Engineering

Tel-Aviv University

iii

ACKNOWLEDGMENTS

Firstly, I would like to give my deep gratitude to my supervisors, Prof. Yossi

Rosenwaks and Dr. Gady Golan for their highly professional guidance, attention,

patience and helpful participation.

I am grateful especially to Dr. Gady Golan for his everyday cooperation and

support of my work in the Thin Film and Microelectronics Laboratory of Holon

Academic Institute of Technology.

I would like to thank to my colleague in the Thin Film and Microelectronics

Laboratory Dr. Emma Rabinovitch and engineer Gennady Makrinich for their help

during my work in the experimental elaboration and fruitful discussions.

I would like to thank everyone who has helped me during my course of my

Ph.D. studies. I have enjoyed with very productive discussions with Prof. Nathan

Croitoru, Prof. Ady Seidman, Prof. Menachem Nathan, Prof. Reuven L. Boxman,

Prof. Gil Rosenman, Prof. Uzi Efron, Prof. Aharon Peled, Dr. Aviezer Israeli. My

special gratitude is given to the leader of the Electrical and Electronic Engineering

Faculty of the Holon Academic Institute of Technology Prof. Arie Shencman for his

adequate help.

Deep appreciation to my family and especially to my wife Ella, without their

patience, love, faith, and encouragement, this work would not have been possible.

iv

Abstract

Ultra-thin semiconductors and metal films have gained high technological

importance in recent years. Sputtering is considered to be the preferable way for

industrial thin semiconductor systems preparation. However, the magnetron

sputtering, usually used for this purpose, does not enable independent control of the

various physical factors, affecting the growth process. Each independent physical

deposition parameter, such as sputtering voltage, substrate temperature, gas pressure,

etc., can be taken as a degree of freedom in the deposition process design. The

controllability of the required coating properties increases with the number of degrees

of freedom.

The main goal of this work was to develop a sputtering based method with an

increased number of degrees of freedom, to produce ultra-thin semiconductor films

and various metal contact films suitable for electronic and opto-electronic

applications. The basic idea behind the developed method was to create a plane of gas

discharge (plasma) placed between the sputtering target and the growing film in order

to enable the sputtered atoms to reach the substrate without collisions. Thus, the shape

of the created plasma is viewed as a thin wall. The thickness of the plasma wall

should be small enough and the distance between the target and the substrate should

be less than the mean free path of the gas atoms. The latter requirement limits the

pressure of the used gases, while the plane plasma shape requires efficient

confinement. All these requirements are best satisfied with the artificial discharge,

supplied by electrons emitted from an independent heated cathode. The external

magnetic field confines the plasma to the shape of the cathode. If the cathode is made

of a tungsten coil, we obtain a wall-shaped plasma. This system enables a fine control

of the plasma parameters, independent of each other.

v

The first part of the thesis is dedicated to the modeling and practical

implementation of the novel sputtering method. First, a simple qualitative model of

the diode and triode electrode systems in vacuum is reviewed. Then a basic

quantitative description of the plane gas discharge is provided. The mean free path of

the gas molecules in the vacuum chamber is chosen as the critical parameter that

defines the type of the sputtered particle transport and the range of the gas pressure

used in the vacuum chamber. The properties and behavior of the plane plasma are

considered under the conditions of ballistics (collisionless) and boundary transfer of

the sputtered atoms (taking into account the diffusion part as well).

A laboratory vacuum setup equipped with a diffusion high vacuum pump and

a mechanical fore-vacuum pump was used as the basis for the novel sputtering

implementation. A heated tungsten coil was used as the cathode. A tantalum plate

supplied with a positive DC voltage with respect to the cathode served as the anode.

The electron flux emitted from the cathode and directed to the anode was confined

using a collimating aperture set up near the cathode and the external magnetic field of

about 100 Oe. The basic properties of the plane plasma were studied with the

Langmuir probe introduced in plasma. The electron temperature and ion

concentrations were estimated from the Langmuir probe I-V characteristics. The

evaluation was done using the Bohm approximation for collisionless conditions,

which were created in the designed system. The measurements were taken in a gas

pressure range from 0.2 mTorr up to 5 mTorr in various points of the vacuum

chamber: along the plasma axis and in its vicinity. It was found that the electrons with

the maximum temperature of about 7-7.5 eV occur along the plasma axis. Positive

argon ion concentration was found to vary in the range of 3.5×1011 cm-3 to

6×1011 cm-3 on the plasma axis, depending on the gas pressure.

vi

The sputtering target was placed on a water-cooled target holder. It was

arranged in parallel to the plasma wall at a distance of about 5 cm from the plasma

axis. The substrate holder was placed at the same distance from the plasma, opposite

to the target holder, in order that the electrons emitted by the cathode will not irradiate

the substrate. The substrate in the novel system is completely protected from the

plasma and the electron irradiation. The substrate temperature does not exceed 60 °C

during 1 hour of operation. A negative high voltage was applied to the target, relative

to the supporting tungsten or tantalum ring electrode (a novel element in the system)

located in the plasma body. This element allowed to obtain high deposition rates with

low applied power. Titanium and silicon were tested as sputtering targets. The

deposition rates for these materials were 60 A°/min and 100 A°/min, respectively,

with the applied power not exceeding 50 W. All of the electrical supplies were

isolated from ground, to protect against parasitic discharges in the vacuum chamber.

This novel sputtering method enables one to independently vary the sputtering

voltage and the ion current while sputtering. This is of course impossible in a

conventional sputtering system such as a magnetron sputtering system. The sputtering

rate is defined by the ion current. In this system, the ion current can be varied in

accordance with the distance between the plasma and the sputtering target. A

magnetic rectangular screen creating a non-homogeneous magnetic field, moves the

plasma wall by rotation around the target holder. This magnetic screen serves as the

fourth electrode in the system. The following table shows the position of the proposed

sputtering method in comparison with the other known methods:

Process Type Typical Particle Energy Vacuum (Pa) Evaporation < 1 eV 0.001

Triode (tetrode) Sputtering 500 … 3000 0.05 Magnetron Sputtering 10 … 100 0.5

Diode Sputtering 1000 10

vii

The second part of the thesis is devoted to the method allowing a major

reduction in the number of trial and error needed to develop and optimize the

deposition process. The novel deposition system described in the first part of the

thesis, used no less than 7 independent parameters. In this part we describe an

approach using the "random section method" and the "response surface

methodology". The main idea behind this approach is the presentation of a real multi-

parameter process as a multi-variable equation (model) and approximate numerical

solution of it, based on the carried experiments. This method was applied for the

deposition and optimization of the transparent conductive In2O3 films and resistive

tantalum films.

The third part of the thesis describes a novel "post-deposition treatment

process", called "the Vacuum Photothermal Processing" (VPP), which enables the

improvement of the obtained properties of the sputtered thin films. This process

consists of simultaneous irradiation of the treated samples with non-coherent light and

an electron beam. This method is considered as an additional level of freedom in the

multiparameter space. The VPP enables stabilization of the deposited coatings,

improving their homogeneity, and curing the irreversible electrical breakdown in p-n

junctions. In addition, the VPP decreases the number of surface states and minimizes

the roughness between the surfaces and interfaces in the semiconductor and the

conducting film systems. Controlling the electron flux during VPP makes it possible

to change the value of the potential barrier between the semiconductor (silicon) and

the conductor (metal contact). The potential barriers in Ti-Si and Ni-Si systems were

varied using the VPP method from 0.605 eV to 0.785 eV and from 0.571 eV to 0.672

eV, respectively.

viii

CONTENTS

List of symbols x

List of figures xii

List of tables xvi

1. INTRODUCTION 1

1.1 General 1

1.2 Sputtering Deposition Methods 2

2. PLANE GAS DISCHARGE APPLICATION FOR SPUTTERING 5

2.1 Introduction 5

2.2 Sputtering discharge modeling (qualitative description) 8

2.2.1 Vacuum diode 8

2.2.2 Gas in the vacuum chamber 9

2.2.3 Triode sputtering system 11

2.3 Plane plasma modeling (quantitative description) 14

2.3.1 Continuity equation 14

2.3.2 Quasi-neutrality and screening length 16

2.3.3 Electrostatic plasma oscillation 18

2.3.4 Ambipolar diffusion 20

2.3.5 Charged carriers generation 21

2.3.6 Charged carriers recombination 24

2.3.7 External magnetic field influence 24

2.3.8 Sputtering in the plasma discharge 27

2.4 Triode sputtering application 31

2.4.1 Triode sputtering system implementation 31

2.4.2 External electromagnet 32

2.4.3 Internal parameters of the plasma 35

2.4.4 Practical Langmuir probe measurement 38

2.4.5 Thin film deposition by the plane discharge 46

2.5 Tetrode sputtering application 49

3. OPTIMIZATION AND IMPROVEMENT OF THE DEPOSITED FILMS 57

3.1 Mathematical modeling of the sputtering process with random

section method application 57

ix

3.1.1 Deposition of In2O3 conductive transparent coatings by DC

sputtering 61

3.1.1.1 Introduction 61

3.1.1.2 Experimental details 62

3.1.1.3 Practical model implementation 64

3.1.1.4 Experimental results 67

3.1.2 Deposition Ta resistive coatings with defined conductivity 74

3.1.2.1 Introduction 74

3.1.2.2 Experimental details 76

3.1.2.3 Practical model implementation 78

3.1.2.4 Experimental results 79

3.2 Vacuum Photothermal Processing (VPP) 85

3.2.1 Introduction 85

3.2.2 Development and investigation of VPP treatment 86

3.2.2.1 Basic Principles of Rapid Thermal Processing 86

3.2.2.2 Experimental details 88

3.2.2.3 Sample preparation 90

3.2.2.4 Control experiments 92

3.2.2.5 Experimental results 94

3.2.2.6 The mechanical properties of the films 98

3.2.2.7 The electrical properties of the films 104

3.2.3 VPP influence on the Schottky barrier height 108

4. CONCLUSIONS AND FUTURE DIRECTIONS 114

4.1 Conclusions 114

4.2 Future research direction 116

5. REFERENCES 117

x

List of symbols. A – total area N – north direction of a permanent

Ar - Richardson constant magnet

B – magnetic field N – turns number

B – recombination coefficient P0 – partial oxygen pressure

C – constant Pres – residual pressure

Da – ambipolar diffusion coefficient PAr – argon pressure

De – diffusion coefficient of electrons Q - charge

Di – diffusion coefficient of ions R – current loop radius

E – electrical field vector Re – recombination rate of electrons

Eg – band gap Ri – recombination rate of ions

F – Fermi level R# - sheet resistance

F – force vector Rs – sputtering rate

F – response function S – south direction of a permanent

F - transparency magnet

Ge –generation rate of electrons T – absolute temperature

Gi – generation rate of ions T, Ts – substrate temperature

H – magnetic field tensile vector Te – electron temperature

I – current Ti – ion temperature

Ic – cathode current V – voltage

Ih – magnetic coil current Va – anode voltage

Ia – anode current VB – probe voltage

It – target current Vf – floating potential

J – current density Vp – plasma potential

Je – electron current density Vt – target voltage

Ji – ion current density W0 – electron energy in vacuum

K – Kelvin degree X1,X2,Xi, Xm – variable parameters

L – geometrical dimension Xi* - normalized variable

L – length of a plasma wall Zi – ionization coefficient

a – acceleration m - number

a – distance n – atom concentration

c – light velocity n0 – particles concentration in the point

xi

c1,c2,ci,ck,cm – model coefficient with potential ϕ = 0

d – thickness of a plasma wall ne – electron concentration

d – film thickness ni – ion concentration

e – charge of electron ng – argon atoms concentration

f(v) – energy distribution function p – pressure

h – Planck constant q – charge density

h – width of a plasma wall qi – elementary ion charge

i – current r – space coordinate

i - index rc – gyration radius

js – saturation current rp – specific probe radius

jis - probe current saturation density t – time

k – Boltzmann constant v – velocity

k - number vB - Bohm velocity

lD – Debye length x, y, z – coordinates

m – mass of a particle

Φ - work function of a metal ε0 - permittivity of free space

ΦB – potential barrier ν - frequency

Φf – floating potential µ0 – permeability of free space

Φp – plasma potential µe – electron mobility

∆ - absolute error µi – ion mobility

∆E – activation energy γt - sputtered atoms yield

ℜ - irradiation intensity ρt - target atomic density

α - absorption edge σ - conductivity

φ0 – arbitrary phase σi – ionization cross section

ϕ - potential τe – lifetime of electron

λ - mean free path τi – lifetime of ion

λ - wavelength ω0 – Langmuir frequency

λD – Debye length, Debye radius ωc – cyclotron frequency

δ - relative error

xii

List of Figures.

Fig. 1. Momentum exchange processes at the target.

Fig. 2. Simple scheme of a vacuum diode.

Fig. 3. A front view of the triode sputtering system.

Fig. 4. Principle scheme for the plasma parameters estimation.

Fig. 5. Schematic view of the potential distribution in front of a solid surface.

Fig. 6. External view of the triode sputtering set up.

Fig. 7. Internal arrangement schematic of the triode sputtering system.

Fig. 8. Schematic presentation of Helmholtz coil pair arrangement.

Fig. 9. Theoretical distribution of magnetic field in the Helmholtz pair.

Fig. 10. Measured magnetic field in the triode sputtering set up.

Fig. 11. Definition of voltage and current for a Langmuir probe (a) and a typical I-V

characteristic for a Langmuir probe (b).

Fig. 12. A side view of the applied Langmuir probe.

Fig. 13. Principal electrical scheme of the Langmuir probe measurement.

Fig. 14. Dimensions of the internal arrangement.

Fig. 15. Anode current dependence on the thermo-emission cathode current.

Fig. 16. Langmuir probe current vs. an applied bias.

Fig. 17. External magnetic field influence on the Langmuir probe current.

Fig. 18. Electron temperature influence on the external magnetic field.

Fig. 19. Langmuir probe characteristics for a = -10 mm.

Fig.20. Langmuir probe characteristics for a = 0.

Fig. 21. Langmuir probe characteristics for a = 10 mm.

Fig. 22. Langmuir probe characteristics for a = 20 mm.

Fig. 23. Electron temperature of plasma in the inter-electrodes space.

xiii

Fig. 24. Plasma density influence on the argon pressure.

Fig. 25. Applied triode sputtering system.

Fig. 26. Discharge current as a function of the applied voltage.

Fig. 27. Discharge current as a function of argon pressure.

Fig. 28. Tetrode sputtering system.

Fig. 29. External view of the magnetic screen.

Fig. 30. Target current versus magnetic screen position.

Fig. 31. Target current influence on the argon pressure.

Fig. 32. Target current as a function of the sputtering voltage.

Fig. 33. Sheet resistance of Ti films measured by four-point probe method.

Fig. 34. Sheet resistance of Si films measured by four-point probe method.

Fig. 35. EDS spectrum of the silicon sample N1.

Fig. 36. Electron micro-photography of the Si sample on the glass

Fig. 37. Experimental magnetron vacuum station.

Fig. 38. A principle scheme of the magnetron sputtering source.

Fig. 39. Example of a bi-parameter experiment space.

Fig. 40. Illustration of the random partial sections of the multi-parameter’s space.

Fig. 41. A diagram of the evaporated ITO films composition.

Fig. 42. Transparency characteristics of In2O3 on the borosilicate glass.

Fig. 43. Typical measurement of the surface resistance diagram.

Fig. 44. A type of substrate influence on the resistivity of In2O3 film.

Fig. 45. A growth rate influence on the In2O3 film properties.

Fig. 46. The transparency of In2O3 film in the IR wavelength range.

Fig. 47. Thickness contact less measurement scheme.

Fig. 48. A typical sample prepared for thickness measurement.

xiv

Fig. 49. Tantalum thin films photography (X 160): a–Ar pressure of 90 mTorr, b–Ar

pressure of 110 mTorr, c–Ar pressure of 130 mTorr, d–Ar pressure of 300 mTorr.

Fig. 50. Conductivity of tantalum films as a function of two variables: argon pressure

and sputtering voltage.

Fig. 51. Tantalum film conductivity at the calculated point.

Fig. 52. A 3-D presentation of the tantalum films conductivity as a function of the

technological parameters.

Fig. 53. Typical Rapid Thermal Annealing (RTA).

Fig. 54. Principle schematic of Vacuum Photothermal Processing set-up.

Fig. 55. Principle scheme of the I-V measurement.

Fig. 56. Typical MDSM structure for I-V characterization.

Fig. 57. Schematic upper and front view of the treated MSM samples.

Fig. 58. Typical Vacuum Photothermal Processing.

Fig. 59. Principle scheme of the VPP treatment set-up including a control electrode.

Fig. 60. Tungsten steams pressure.

Fig. 61. Tungsten coil temperature during VPP.

Fig. 62. Tungsten resistivity dependence on the temperature.

Fig. 63. Spectral efficiency of tungsten for different temperatures.

Fig. 64. Thermo-emission current density of Tungsten.

Fig. 65. Microscopic view of the Al-Al2O3-Ge-Au system (× 80).

Fig. 66. A 3D presentation of the Ge thin film surface (AFM microscope).

Fig. 67. External view of the processed MDSM samples.

Fig. 68. Experimental photographs of Ni electrodes on the silicon for as deposited and

VPP treated samples, in two magnifications: (a) as deposited film, *150; (b) as

deposited film, *3000; (c) VPP treated film, *150; (d) VPP treated film, *3000.

xv

Fig. 69. Experimental photographs of Ti electrodes on the silicon for as-deposited and

VPP treated samples for two different magnifications: (a) as deposited film, *150; (b)

as deposited film, *3000; (c) VPP treated film, *150; (d) VPP treated film, *3000.

Fig. 70. Samples temperatures vs. processing time for various thermal treatments.

Fig. 71. AFM 3D photographs of as deposited and VPP treated Ti films: (a) as

deposited Au-Si-Ti system; (b) VPP treated Au-Si-Ti system.

Fig. 72. AFM section analysis of the roughness for as-deposited and VPP treated Au-

Si-Ti systems: as deposited Au-Si-Ti system; (b) VPP treated Au-Si-Ti system.

Fig. 73. Electrical property measurements before and after a traditional furnace

annealing treatment of the glass-Al-Al2O3-Ge-Au system.

Fig. 74. VPP with glass filter (sample 1b).

Fig. 75. VPP processing of a glass-Al-Al2O3-Ge-Au system.

Fig. 76. Sheet resistance distribution of the glass-In2O3 system.

Fig. 77. Sheet resistance distribution of VPP processed glass-In2O3 system.

Fig. 78. I-V characterization of as deposited and VPP treated Au-Si-Au systems.

Fig. 79. I-V characterization of as deposited and VPP treated Au-Si-Ti systems with a

bias voltage in the control electrode.

Fig. 80. I-V characterization of as deposited and VPP treated Au-Si-Ni systems with a

bias voltage in the control electrode.

Fig. 81. The influence of a biased VPP treatment on the potential barrier magnitude in

Au-Si-Metal systems.

xvi

List of Tables.

Table 1. Typical sputtering parameters of the triode deposition system.

Table 2. Sputtering deposition rates using Argon (A°/min).

Table 3. Sputtering deposition parameters used for the Si films deposition.

Table 4. Different deposition process and its basic parameters.

Table 5. Standard form for the model coefficient estimation: F is the response

function and Xi* are the normalized process parameters.

Table 6. Normalized active technological parameters.

Table 7. Experimental data on the In2O3 film deposition.

Table 8. Optimized technological parameters.

Table 9. Parameters of the films after optimization.

Table 10.Variable deposition parameters of Ta films.

Table 11. Tantalum conductivity as a function of two variables.

Table 12. Parameters of the experimental trials.

1

1. INTRODUCTION

1.1 General

Over the past decades, thin-film science and technology have developed into a

major industry. This growth is a result of a fundamental understanding and

development of the processes and tools that are used in thin-film deposition. The

importance of coatings for the industry and the progress in synthesis of new materials

have produced an explosion of processing technologies1. There are many different

methods for thin-film deposition: chemical vapor deposition (CVD), plasma enhanced

chemical vapor deposition (PECVD), liquid reaction, condensation, and physical

vacuum deposition (PVD), including evaporation and sputtering which are

particularly important in thin-film deposition.

The main idea of these methods is separation of atoms from a substance and

deposition of these newly obtained atoms in the form of the required coating on a

substrate. We can regard the PVD process as consisting of three parts: 1) material

decomposition; 2) transportation of the separated material atoms to the substrate;

3) growing a thin film on the substrate. Each of these parts may be considered

separately. Each one is influenced by numerous technological parameters.

1. The first part of the process is defined by the energy introduced in the

decomposition process (for example, by heating the material in the vacuum

evaporation method). Such parameters as the residual vacuum level, presence of

reactive gas and pressure, or thermal properties of the material are only a few of

the many parameters that affect the first part of the process.

2. The space between the atom source and the substrate location, presence of plasma

in this space, and the level of vacuum are parameters that define the second part of

the process.

2

3. In the third part of the process, the focus is on the substrate and the processes

occurring in the vicinity of it in the vacuum chamber. Processes in the third part

are defined by such parameters as the energy of deposited atoms, the

contamination present on the surface of the substrate, and the substrate

temperature.

In this work, I aim to improve each part of the film growth process.

1.2 Sputtering Deposition Methods

Sputtering is one of the most effective methods of thin film deposition. It is

widely used in the present-day semiconductor2, photovoltaic3, and microelectronics4

industries. Materials with high melting points such as ceramics and refractory metals,

which are hard to deposit by evaporation, are easily deposited using sputtering.

Different sputtering methods, from a simple DC glow discharge sputtering to the

multi-electrode DC and RF sputtering, enable deposition of any type of material and

synthesis of new materials with pre-defined properties. For instance, sputtering can be

used to create a new metallic material from an intermetallic compound, this new

material having the same chemical composition as an intermetallic semiconductor

with a zinc-blend or wurtzit-type crystalline structure and being stable at the room

temperature and atmospheric pressure5. At the same time, each sputtering method has

its particular limitation (for example, DC sputtering requires a conductive target1, etc.)

Sputtering is a process in which highly energetic ions strike the surface of a

solid target, causing the emission of neutral atoms or molecules from the target

surface as a result of momentum transfer6. Assuming a hard-sphere elastic-collision

model7, the exchange of momentum among atoms in and around a collision cascade is

3

a statistical process similar to that encountered in billiards. This process is illustrated

in fig.1.

Fig. 1. Momentum exchange processes at the target.

Upon the bombardment of the target with ions, the following processes may occur:

1. The incident ion is reflected back and gets neutralized in the process.

2. Secondary electrons are emitted due to ion impact.

3. The incident ion causes ejection of the atom/atoms of the target.

4. The ion itself is buried in the target with/without simultaneous target atom

ejection.

Consideration of the momentum transfer between the incident and the target particles

reveals this transfer to be at its maximum when the particle masses are equal. The

maximum momentum transfer from the incident ions correlates to the maximum

target atom sputter rate. The average energy of the atoms sputtered from the surface

lie in the range of 10 to 40 eV. However, the energy of the sputtered particles can

reach that of the sputtering ions 8. Typical sputter yields for metals lie in the range of

0.5 to 2 atoms/ion6.

All sputtering methods can be divided into two main categories. In the first

category, the plasma occupies the whole volume of the vacuum chamber and comes in

4

contact with the grown layer and with the walls of a vacuum chamber. In the second

category, the plasma is confined to a restricted volume by means of electromagnetic

field or other method. In this case, independent control of different physical

parameters affecting the deposition process becomes especially important.

Each independent physical deposition parameter such as sputtering voltage,

substrate temperature or gas pressure, can be considered as a degree of freedom in the

deposition process. Controllability of the required coating properties increases with

the number of degrees of freedom. All known sputtering methods are based on a

number of variable input parameters that are independent and controllable. Up to now,

however, the distance between the plasma and the sputtering target has never been

used as an independent and controllable parameter. This novel parameter permits to

modify the film growth rate independently of all other deposition parameters.

The first part of this Thesis describes the mathematical modeling of plasma,

construction of a novel sputtering deposition system with several independent control

parameters and studying the processes in this system.

In order to develop a novel sputtering deposition process for a specific type of

film in the system with several independent parameters, a large number of

experiments are required. This number can be substantially reduced when using the

mathematical planning methods. A methodology for reducing the trial-and-errors

number is described in the second part of the Thesis.

A post-deposition treatment is also a very useful process for modifying and stabilizing

the film properties. Our novel post-deposition treatment method leads to an additional

level of freedom and to enhanced controllability. The innovative method for post-

deposition treatment called Vacuum Photo-thermal Processing (VPP) is described in

the third part of the Thesis.

5

2. PLANE GAS DISCHARGE APPLICATION FOR SPUTTERING

2.1 Introduction

Sputtering technology, which includes a large number of PVD methods, is

widely used for deposition of thin-film layers on semiconductor wafers. These layers

are used as diffusion barriers, primary conductors, antireflection coatings, etc. With

the progression toward finer topographical dimensions on wafers and increasing

aspect ratios, the broad angular distribution of sputtered atoms leads to poor coating

quality in deep grooves or other singularities on substrates9. This is mainly

conditioned by high energy of sputtered particles.

Sputter deposition is usually carried out in diode plasma systems known as

magnetrons. In these systems, the cathode is sputtered by ion bombardment and emits

atoms, which are then deposited on the wafer in the form of a thin film.

Originally, physical sputter deposition utilized DC diodes, which simply

represented parallel plates powered by a power supply of several kilovolts. This

process was conducted under the working pressure of several tens to hundreds of

mTorr. The negative plate, also known as the cathode or the sputtering target, was

bombarded with ions from the plasma volume set up between these two plates. The

sputtered cathode atoms could then deposit on various surfaces inside the vacuum

chamber, forming films. The DC diodes were characterized by slow deposition rates,

high voltage and low currents. This method was not suitable for deposition of

dielectric films.

Replacing the DC supply with an RF supply, usually operated at a frequency of

13.56 MHz, made deposition of dielectric films possible. However, the deposition

rates were no higher. A further improvement in sputtering techniques was achieved by

confining plasma electrons by means of magnetic field. This technique is known as

6

magnetron sputtering. Higher plasma density reduces the discharge impedance and

results in much higher current and lower voltage discharge. Another known sputtering

technique is the vacuum arc deposition process, which enables very high deposition

rates10.

The electron confinement in a magnetron is due to the presence of orthogonal

electrical (E) and magnetic (B) fields at the cathode surface. These fields result in a

classical E××××B drift for electrons (the Hall effect), which gives rise to a sequence of

cycloidal hopping steps parallel to the cathode face. As a result, the secondary

electrons, which are emitted from the cathode due to ion bombardment, are confined

to the nearest vicinity of the cathode. In a magnetron, the electric field is always

oriented normally to the cathode surface. The transverse magnetic field is configured

in such a way that the E××××B drift paths form closed loops, which trap the drifting

electrons and force them to circulate many times around the cathode face. As a

consequence, the ions sputter only a limited part of the cathode11 and create

heterogeneous etching ring zones on its surface.

In all of the above-mentioned sputtering methods, plasma occupies the entire

volume of the vacuum chamber and comes in contact with the internal metal parts and

the grown layer. Also, the self-sustained gas discharge in the diode system is mainly

supported by the auto-electron and secondary emission7. The voltage necessary to

initiate a discharge is defined by known framework of the Paschen law7, 110. After

initiating, the discharge current depend on the gas pressure and nearly independent of

the applied voltage (the normal glow discharge region111, 112). These conditions

negatively act on the quality of deposited coatings. A substrate charges negatively and

undergoes damage from plasma ions113. Also, this negative charge uncontrollably

changes conditions of the film growth.

7

Artificially supported sputtering methods can be divided into two basic

categories: in the first category, unconfined plasma contacts the substrate (so-called

ion-plating methods); in the second category, the plasma is confined by electro-

magnetic field, as seen in example [12]. If the plasma is in contact with the substrate

surface, its influence on the growing film gets out of control. This influence may

cause contamination in the deposited films, while uncontrolled heating of the growing

film may change its physical properties. An independent source of electrons, such as a

hot filament, is an important part of the artificial discharge system. It promotes a

significant decrease in the working pressure during sputtering and independently

controls the plasma current. In this case, the sputtering voltage (a high negative

voltage) is applied to the third electrode: the target to be sputtered. Therefore, this

system includes three electrodes with three independent supply sources: the cathode

heating supply, the plasma maintain supply (voltage applied to the anode-cathode

pair), and the high voltage sputtering supply (voltage applied to the sputtering target).

This system is called the triode sputtering system. Such a system was used for

deposition the signal electrodes for vidicons sensitive in the near UV spectrum13, 14.

As described in ref. [13, 14], the plasma was confined in the narrow flat band form by

a combination of collimating shield placed near the cathode and of an external

magnetic field. This magnetic field was produced by two identical electro-magnetic

coils placed at the distance equal to the coil radius (Helmholtz coils). This approach

enabled the formation of reproducible deposits of bi-layer thin-film systems of

Pd+Al2O3 with a transparency of 72% on the wavelength of 250 nm and resistivity no

more than 2 kOhm/sq. In these experiments, the Pd layers of the deposited films were

about 20 Å thick, and the alumina coating layers about 50 Å thick. The film

thickness was measured by a laser ellipsometer.

8

Since the considered discharge structure is highly non-uniform and very

complicated, a complete quantitative model of the discharge has not been developed.

It is of interest to estimate the parameters of plasma in the triode sputtering system.

These parameters are the plasma density, the electron temperature, and the floating

potential as a function of position of a point in a discharge gap.

2.2 Sputtering discharge modeling (qualitative description)

2.2.1 Vacuum diode.

A vacuum diode is a closed glass flask with two electrodes introduced into it.

There are two types of vacuum diode lamps: those with vacuum inside the lamp and

those with low-pressure gas. Figure 2 shows a simple diagram of a vacuum diode.

Fig. 2. Simple diagram of a vacuum diode.

The current in the circuit shown in figure 2 appears only when the cathode is heated.

The Boguslavsky-Langmuir law15 determines the anode current in this circuit as

where i is the current, V is the voltage applied to the diode, and C is a constant

defined by the shape and dimensions of electrodes.

As known, the saturation current of a metal heated in vacuum is defined by the

Richardson formula16:

where js is the saturation current density, Φ = W0 - F the work function of the metal,

W0 the electron energy in vacuum, F the Fermi level for this metal, k the

Boltzmann constant, T the metal temperature in Kelvin degrees, and Ar the

(1) CVi 2/3=

(2) )kT/exp(TAj 2rs Φ−=

mA Vacuum diode

Thermoemission cathode

9

Richardson constant equal to 120 A/cm2K2. This expression describes the current

density as a function of a heated metal temperature measured in Kelvin degrees. The

corresponding number of electrons emitted from the cathode surface can be simply

estimated from this current density.

A homogeneous magnetic field applied co-axially to the electrical field

(applied in the anode-cathode direction) confines the electrons emitted from thermo-

cathode. Thus, the greater part of the produced electrons will be collected on the

anode, while only a small part of them will recombine on the vacuum chamber walls

or return to the cathode. In this case, the current through the anode-cathode system

(diode) should increase. The maximum diode current is limited by the Richardson

expression. Note that the described case was collisionless.

2.2.2 Gas in the vacuum chamber.

Introducing gas in a vacuum chamber changes the picture due to additional

processes that may occur to gas atoms, such as collisions of various types, ionization,

etc. First of all, it is necessary to define the model which is best to apply here, i.e.,

vacuum, diffusion or intermediate model. As is known, gas particles moving inside a

bell jar collide. The average path between collisions, called the mean free path (λ), is

the critical parameter defining the type of model. If λ exceeds the geometrical

dimensions L of the vacuum chamber (λ>> L), one can use the vacuum (collisionless)

model, neglecting the diffusion processes in plasma. Practically, this case is described

by the relation λ > 4L. In the case when λ ≈ L, both vacuum and diffusion transfer

types should be considered. In the third case (λ < L), the transport properties of

plasma should be described using a diffusion model similar to the continuity equation

for semiconductors, although the case of gas requires a more complicated description.

A description using such a model (Schottky diffusion model for medium pressure)

10

appeared in a review by N. Braithwaite7. Complications here arise due to the presence

of non-homogeneous plasma and high rate of collisions between electrons, gas

molecules and ions.

Collisions may be of several types. For example, a collision of an electron

with an argon atom may excite the atom to a higher energy level without ionization:

e + Ar ⇒ Ar* + e (excitation), corresponding to the change of state 3p6 → 4p54s1 for

the valence electrons4. This metastable atom can then emit a photon and return to a

lower energy state e + Ar ⇒ Ar* + e ⇒ Ar + e + hν (relaxation process). The

radiation is usually in the visible or ultraviolet spectrum. Each gas is known to

possess an individual color at the plasma state. This color is explained by both

radiation processes: relaxation and recombination. Other processes are ionization acts

of the following types: e + Ar ⇒ Ar− or e + Ar ⇒ 2e + Ar+. The lifetime of these ions

is defined by the gas pressure, the applied voltage, and the applied magnetic field.

Each act of recombination is accompanied by an act of radiation.

The type of sputtering model to be applied can be determined by the critical

parameter, λ. Typical internal dimensions of vacuum stations in operation are ~ 30-50

cm. The working pressure p (Torr) appropriate for these dimensions (λ is measured in

cm) can be calculated from the known relation2:

p ≈ 5×10-3/λ (3)

If the gas pressure in the vacuum chamber is equal to ~1×10-3 Torr, we have the

boundary (intermediate) model. The diffusion processes can be neglected for the

pressure values of p ≤ 0.5×10-3 Torr (the ballistic transport model17). For the pressure

values higher than 1×10-3 Torr, only the diffusion transport model7, 17 can be used to

describe the processes of sputtering and deposition. It is important to distinguish

between two types of sputtered atom collisions: those with ions and those with neutral

gas molecules. Collisions between a sputtered atom and a gas ion may cause

11

ionization of the atoms and promote chemical reactions between the atoms and the

gas ions. Collisions between neutral gas atoms and a sputtered atom are called

thermalization. This term describes a cool-down of the sputtered atom energy to the

point where it matches the gas temperature (typically a few hundred degrees).

Obviously, the type and the amount of collisions define the film quality.

2.2.3 Triode sputtering system.

A triode sputtering discharge system is based using of a special electron

source. This source supports the sputtering process by emitting electrons throughout

its duration. Figure 3 shows a diagram of a triode sputtering system13. This system

consists of a thermo-emissive cathode (the first electrode), a round anode, placed in

front of the cathode (the second electrode), and a water-cooled multiple target holder

connected to a high voltage supply (the third electrode). A metallic shield with a

collimating aperture in its center serves as the cathode screening. Externally mounted

electromagnet coils (Helmholtzs coils) generate a homogeneous magnetic field on

the electron beam axis. The substrate is located opposite the target, normally to the

anode-cathode axis; a rounded turning shield protects the substrate.

Fig. 3. A front view of the triode sputtering system.

12

The installation works as follows. The plasma-forming gas argon fills up a

bell jar after it has been evacuated up to residual vacuum. The argon pressure is set to

a required level defined by the sputtering conditions. The cathode current is switched

on to heat the tungsten coil. A constant electrical field is applied to the cathode-anode

gap to fire a forced discharge. The collimating aperture collimates this discharge to

the form of a band. The homogeneous magnetic field produced by the Helmholtz coils

confines the collimated discharge to a thin plasma band, to serve as the ion source in

the system. Confining of the plasma discharge increases the plasma current, therefore,

the ion concentration in the plasma band significantly increases. The plasma band

thickness is essentially less than the vacuum chamber dimensions. Therefore, we can

believe that the sputtered atoms overcome the targetsubstrate distance without

collisions with ions. Therefore, high ion concentration in the plasma band makes it

possible to decrease the gas pressure in the bell jar and thus transform the type of

sputtered atom transport from diffusion to intermediate or ballistic.

Let us now consider a model that can represent the triode sputtering process

described above and its physical parameters. Figure 3 represents a simplified scheme

of a triode sputtering system. The plasma in our case is a thin flat rectangle glow

volume. This volume consists of rarefied gas (argon), electrons and ions (mainly

positive single-ionized argon atoms). This volume occupies the space between the

cathode and the anode, as shown in figure 4, and is confined in the bell jar.

13

Fig. 4. Schematic diagram for plasma parameters estimation.

First of all it should be noted that the plasma in use represents a weakly

ionized glow discharge, which is on average electrically neutral. The plasma has the

following features:

• it can be driven by applied electrical field,

• charged particle collisions with neutral gas molecules are important,

• surface losses are important at plasma boundaries,

• ionization of neutrals maintains the plasma in a steady state,

• electrons are not in thermal equilibrium with ions.

In other words, we are dealing with weakly-ionized cold plasma with the

following typical parameters4: the electron temperature is Te ≈ 1-10 eV, ion

temperature is less than electron temperature Ti << Te, and the plasma density is

ne ≈ ni ≈ n ≈ 108 1013 cm-3.

x

y

z

Cathode

Anode

Plasma

L

h

d

BB

B

Homogeneousmagnetic field

Sputteringtarget

Gas inlet

Pumping

14

2.3 Plane plasma modeling (quantitative description)

2.3.1 Continuity equation.

The usual macroscopic form of Maxwells equations is30

where E and H are the electric and magnetic field vectors and µ0 = 4π⋅10-7 H/m and ε0

≈ 8.85⋅10-12 F/m are, respectively, the free space permeability and permittivity. The

sources of the electrical and the magnetic fields are, respectively, the charge density q

and the current density J. The charge density change in a point can be obtained by

taking a partial time derivative of Equation (6) and substituting the expression for the

magnetic field source from Equation (5).

According to the known formula of the vector analysis,

∇∇∇∇(∇∇∇∇××××H) = 0 (10)

Substituting Equations (8) and (10) into (9), we get the charge continuity equation in

the form of the charge conservation law:

====⋅⋅⋅⋅∇∇∇∇

====⋅⋅⋅⋅∇∇∇∇

++++====××××∇∇∇∇

−−−−====××××∇∇∇∇

) (H

) (E

)(JEH

) (HE

7 0

6 q

5 t

4 t

0

0

0

ε

∂∂

ε

∂∂

µ

(((( )))) )8(tq

t 0 E ∂∂

ε∂∂

====∇∇∇∇

(((( )))) (((( )))) (((( )))) (9) tt 00 JHJHEE ∇∇∇∇−−−−××××∇∇∇∇∇∇∇∇====−−−−××××∇∇∇∇∇∇∇∇====

∇∇∇∇====∇∇∇∇

∂∂

εε∂∂

)11(0tq +J ====∇∇∇∇

∂∂

15

The charge density q may be appeared as a sum q = niqi - nee of ion and electron

components. Taking into account this assumption in the weakly-ionized plasma, the

equation (11) can be written independently for each type of charged carriers.

Therefore, we can write these equations in a one-dimensional form as follows:

for electrons and

for ions. Here A⋅dy = d⋅h⋅dy is an elementary volume of a plasma slice with cross-

section area A and thickness dy; ni and ne concentrations of the ions and electrons in

the slice; e and qi are elementary charges of electron and ion; Ji and Je ion and

electron current densities in the corresponding part of the slice; Gi and Ge generation

rates of ions and electrons in the slice; and Ri and Re decreasing rates of the ion and

electron amounts in the slice.

If we assume that the ion current in the Y-direction changes continuously (at least in

the positive column region), we can write

Now equation (13) transforms to the following equation:

(14) dyyJ

)yy(J)y(J iii

∂∂

−=δ+−

(15) RGyJ

q1

tn

iii

i

i −+∂∂

−=∂

∂

( ) ( )[ ] (12) dyRAdyGAyyJyJeAdyA

tn

eeeee ⋅−⋅+δ+−−=⋅

∂∂

( ) ( )[ ] (13) dyRAdyGAyyJyJqAdyA

tn

iiiii

i ⋅−⋅+δ+−=⋅∂

∂

16

In the single-ionized gas plasma the ion charge is equal to the elementary electron

charge qi = e = 1.6⋅10-19 C. The ion current in the glow discharge is defined by both

diffusion and electrical drifts:

Substituting (16) into (15), we arrive to the complete continuity equation:

Here µi is the ion mobility in the discharge, and Di is the ion diffusivity. This equation

describes the ion concentration behavior in the gas discharge plasma shown in

figure 4. The equation describing the electron behavior looks as follows:

Equations (17) and (18) are similar to the continuity equation for semiconductors18

and can be analyzed in a similar way.

2.3.2 Quasi-neutrality and screening length.

Equation (6) claims that the electric field is created by the charge carriers. One

can say that each charged particle has the electric potential ϕ dependent on the space

coordinates. Thus, for single-ionized particles we have:

ϕ = Q/r (19)

where r is a coordinate in space and Q = e is the charge of the particle. Each particle

in plasma is driven with thermal velocity in random directions, with distribution

(16) dydn

eDEneJJJ iiiiiEiDi −µ=+=

(17) RGyn

Dyn

EyEn

tn

ii2i

2

ii

iiii −+

∂∂

+∂∂

µ−∂∂

µ−=∂

∂

(18) RGyn

Dyn

EyEn

tn

ee2e

2

ee

eeee −+

∂∂

+∂∂

µ+∂∂

µ=∂

∂

17

subject to Boltzmann rule. The electrical field E is equal to the potential gradient and

acts in the opposite direction:

E = -∇∇∇∇ ϕ (20)

Substituting (20) into (6) yields the well-known Poisson equation:

where q is the charge density dependent on the space coordinates. Taking into

account the Boltzmann distribution

where T is the temperature, n is the concentration of charged particles in the point

with the potential of ϕ, and n0 is the concentration of charged particles in the point

with ϕ = 0, i.e., the average concentration of particles over the entire space. The

Poisson equation can be written in the form:

To solve this equation, we simplify it, assuming the random deviations of the

momentary concentrations and potential fluctuation. Based on this assumption, we

can develop the exponential function in a Taylor series and stop after the second

term20, thus linearizing the right side in Equation (24):

(21) q∆000 ====−−−−====−−−−∇∇∇∇∇∇∇∇====∇∇∇∇ ϕεϕεε )(E

(22) q

0ε−=ϕ∆

(23) enn kTe

0

ϕ−

=

(24) ene kTe

00

ϕ−

ε−=ϕ∆

(25) kTen

0

20 ϕ

ε−=ϕ∆

18

For a spherical symmetrical distribution of potential near the charged particle

(charged point), we obtain the solution in the form of:

where C is an integration constant that approximates to e (see Prediction (19)), and lD

is the screening length known as the Debye length:

The Debye length is a typical length such that a small deviation in the charge

density from the equilibrium within this length is relaxed or screened (in other words,

is no longer felt31). A characteristic feature of plasma is its ability to screen out an

electrical charge. A potential disturbance in plasma will attract particles of the

opposite charge. This cloud of charge provides screening from the rest of the plasma.

This phenomenon is referred to as Debye screening28. In physical units, it can be

described by the following expression4:

2.3.3 Electrostatic plasma oscillation.

Electrons and ions move with thermal velocities just like neutral molecules do.

An electron mass is much smaller than that of an ion. Therefore, staying within the

cold plasma approximation, one can suppose that the ions rest while the electrons

move among them. Spontaneous motion results in a small deviation of the space

(26) erC

Dlr

−

=ϕ

(27) enkT

l 20

0D

ε=

(28) [cm] nT

743le

eDD ≈λ≡

19

charges and creates charge divisions in the plasma volume. Therefore, accordingly to

the charge conservation law (equation (11)) and assuming that only electrons are

responsible for the electric current, one can write:

where v is the electron velocity. Charge division creates an electrical field E that

affects the electron motion:

Substituting equation (30) into equation (29), and differentiating, we get:

Considering equation (6) and assuming that the full derivative is equal to the partial

one, we obtain the following equation to describe the casual motion of plasma charge

density:

This equation describes the natural sinusoidal oscillation of the electron cloud with

respect to the ion cloud with the natural frequency ω0, called the Langmuir

frequency20. This frequency is the self-frequency of plasma; it is defined by the

plasma concentration only:

)29 (ne)ne(tq v vJ ⋅⋅⋅⋅∇∇∇∇====−−−−−−−−∇∇∇∇====−−−−∇∇∇∇====

∂∂∂∂∂∂∂∂

)30(edtdm Ev

−−−−====

)31 (m

netq 2

2

2

E∇∇∇∇−−−−====∂∂∂∂∂∂∂∂

(32) qqm

nedt

qd 20

0

2

2

2

ω−=ε

−=

(33) m

ne

0

2

0 ε=ω

20

2.3.4 Ambipolar diffusion.

In the previous section, we assumed the ions to be motionless with respect to

electrons. However, they do move and, moreover, the two species, electrons and ions,

do not move independently. The motion of electrons creates a charge imbalance, and

the resulting electric field drags the ions along with electrons32. In other words, an

electric field spontaneously arises in plasma, with the field intensity just required for

the resulting forced diffusion flows to preserve local charge neutrality33. This effect is

referred to as the ambipolar diffusion. Similar effects were found in semiconductors

such as silicon34 and gallium arsenide35.

The total drift of charged particles such as electrons and ions is described by

equation 16 and is composed of a drift component and a diffusion component. If we

assume the imbalances in fluxes and densities to be very small, we can consider the

ion and electron currents in the quasi-neutral plasma to be equal (ne = ni = n). Thus we

can write the following relation:

Now we can express the electric field as follows:

Substituting this expression into equation (16), we obtain the flux (current) in terms of

density:

(34) JndydeDEnen

dydeDEneJ eeeeeiiiii =−µ−=−µ=

(((( ))))(((( )))) )35 (

n

ndyd

DDE

ei

ei µµ ++++

−−−−====

(((( ))))(((( )))) )36 (n

dydDD

endydeDn

dydDD

eJei

ieeii

ei

eii

µµµµ

µµµ

++++++++

−−−−====−−−−++++−−−−

====

21

Now we can rewrite the expression (36), with the term Da as an ambipolar diffusion

coefficient:

Here the ambipolar diffusion coefficient is equal to

Since the electron mobility is much greater than that of ions, we can simplify the form

of Da as:

Taking in account the known Einstein relation36:

we get:

From (41) we see that the ambipolar diffusion is controlled by the slower particles (in

our case the ions), but increases in direct proportion to the temperature ratio. Thus, if

Te >> Ti, which is the usual case in weakly ionized plasma, the diffusion rates of both

ions and electrons are much higher than that of free ions.

2.3.5 Charged carriers generation.

Ionization of neutral particles (argon atoms) in the case of not self-sustained

discharge is normally achieved by electron impact. The electrons are produced by

(37) dyneDJ a

∂=

(38) DD

Dei

ieeia µ+µ

µ+µ=

(39) DDD ee

iia µ

µ+=

(41) TT

1DDi

eia

+=

(40) e

kTD µ=

22

thermo-electron emission from a hot filament (tungsten) and accelerated to about 100

eV. This energy is sufficient to ionize argon atoms, whose ionization potential is

equal to 15.76 eV23. The energy of thermally produced electrons is not

monochromatic, but rather has a Boltzmann distribution37. The amount of collisions

between the electrons and other particles in the plasma is defined by the argon

pressure. A detailed consideration of all collision types: ionization, excitation, and

recombination processes in the plasma is very complicated and practically

impossible19. Moreover, all processes in plasma are probabilistic, so that a precise

solution of a system describing the behavior of each particle is useless20.

A solid surface in contact with plasma will charge up negatively with respect

to the plasma. The reason for this is the mass difference between the electrons and the

ions. The main part of the potential difference between the solid and the plasma will

be confined to a narrow sheath restricted to several Debye lengths in thickness28. This

potential distribution is schematically shown in figure 5.

Fig. 5. Schematic view of the potential distribution in front of a solid surface.

Electrons emitted by the hot cathode run through the sheath distance and begin

to move through the plasma volume. Initially there are no electrons in the volume, and

23

we can write the first condition for the electrons movement as a function of two

variables, the distance y (see fig. 4) and the time t:

n(y,t) = n0(y,0) = 0 (42)

In the course of time, an area with high electron concentration arises at the border of

the plasma volume. This surface concentration is constant and equal to the amount of

emitted electrons (see formula (2)). Electrons move through the plasma volume in

accordance with the Fick law. On their way, however, the electrons collide with the

neutral argon atoms, and the number of electrons decreases. One can say that this

diffusion process never takes electrons through the entire depth of the plasma volume.

In other words, if we assume that the plasma volume is infinite, we can write:

nt(∞,t) = 0 (43)

Taking into account conditions (42) and (43), we obtain that the electron movement in

the case of the forced (non-self-sustained) discharge can be described as diffusion

from a constant source, i.e., in terms of the error function38.

This type of electron behavior lasts for a limited time period, until a sufficient amount

of positive ions is generated.

A typical degree of ionization in the weakly ionized plasma does not exceed 1 atom in

104 items; therefore most of argon atoms remain neutral21. These neutral atoms have

the thermal energy corresponding to the ambient temperature and thus play a minor

role in the process. The ion generation rate can be presented in the form22:

Gi = Zine (45)

where Zi is the ionization coefficient and ne is the number of emitted electrons. Here

the electron concentration ne is defined by the Richardson formula16 and is equal to

(44) Dt2

yerfcn)t,y(n 0

=

24

8.2⋅1015 electrons · cm-2s-1 for the cathode temperature 2000 K 7. The ionization

coefficient Zi is defined by the pressure in the vacuum chamber, the electron energy

distribution, the electron velocity, and the cross-section of the argon atoms7:

where f(v) is the energy distribution function for electrons, ng the argon

concentration, σi the ionization cross-section, and f(v)dv the weighting. The

integration in the numerator does the adding, while the denominator completes the

weighting averaging.

2.3.6 Recombination of charged carriers.

Parameters Ri and Re in the continuity equations (17) and (18) describe the

recombination rate of the charged carriers. The processes of recombination are caused

by collisions between electrons and ions7. The amount of collisions is determined by

the argon concentration in the vacuum chamber and the argon cross-section, that is,

by the time between atom collisions, or lifetime:

where τi and τe are lifetimes of ions and electrons and B is a recombination

coefficient, which depends on the collision cross-section and the energy distribution

of the interacting particles.

2.3.7 External magnetic field influence.

Every charge is a source of the electric field, through which it exerts force on

any other charge in inverse square proportion to the distance between them. An

(46) dv)v(f

dv)v(vf)v(nZ

0

0 ig

i

∫∫

∞

∞σ

=

(47) n

Bn/1n

Bn/1nn

RRe

e

g

e

g

i

i

iie τ

===τ

≈≈

25

electric field imposed externally will apply forces to any charge that enters it. Charges

that move with respect to magnetic fields also experience forces. The Lorentz force

conveniently combines the electric and magnetic effects on a charge7:

F = e(E + v××××B) (48)

Here e is the elementary charge in Coulombs; E is the electric field in Volts per

meter; v is the velocity in meter per second; ×××× implies the vector product; and B is the

magnetic flux density in Teslas. In the case of constant electric field E = E0 without

magnetic field (B ≡ 0), a particle moves with a constant acceleration along E0 (see

figure 4):

y(t) = y0 + v0t + 0.5a0t2 (49)

where y0 and v0 are the particle position and velocity at t = 0 and a0 = eE0/m.

For the negative particles and a constant magnetic field B = B0, which we take

to lie along y, with E ≡ 0, the components of (48) are:

A motion of charge in the direction y is coupled from the x and z motions.

Differentiating (51) and eliminating vx with the help of (50), we obtain

This equation describes the circular motion of a charged particle with gyration

(cyclotron) frequency equal to:

=

=

−=

(52) 0dt

dvm

(51) dt

dvm

(50)

y

0z

0

Bev

Bevdt

dvm

x

zx

(53) vm

)eB(dt

vdz2

20

2z

2

−=

26

Solving Equation (53) and using (54) to obtain vx, we have:

where v⊥0 is the speed component perpendicular to B0 (y-axis), and φ0 is an arbitrary

phase. Integrating yields the particle position

where

is the gyration radius, which can be expressed through the particle energy15:

Equations (58) - (60) show that the particle moves in a circular orbit perpendicular to

B, with frequency ωc and radius rc, around a guiding center with coordinates

(x0,z0,y0+vy0t), which moves uniformly along y. Positive ions gyrate around the

(54) m

eB0c =ω

=

φ+ω−=

φ+ω=

⊥

⊥

(57) vv

(56) )t(Sinvv

(55) )t(Cosvv

0yy

0c0x

0c0z

+=

φ−+φ+ω=

φ−+φ+ω=

(60) tvyy

(59) )Cosrx()(Cosr x

(58) )Sinrz()(Sinrz

0y0

0c00cc

0c00cc

(61) v

rc

0c ω

= ⊥

(62) B

)kT(em2r

0

5.05.0

=

27

magnetic field lines according to the left-hand rule, and electrons gyrate according to

the right-hand rule.

The non-self-sustained discharge applied in the system on figure 4 uses both

constant electric and constant magnetic fields. The directions of both fields are

parallel to each other. Thus, the field components can be written as follows:

Bx = Bz = 0; By = B0; Ex = Ez = 0; Ey = E0 (63)

Expressing v as a vector sum v = vy(t) + v⊥(t) in the Lorentz force equation (48), we

obtain the uniform acceleration along y and an equation for the transverse motion:

Solving this equation in a general form is a rather complicated task. However, we can

consider a combined solution of (49) and (58) - (60).

All the particles that move along y-axis are not affected by the deviated forces such as

E0 ×××× B0 ≡ 0 (65)

Electrons and ions that have a perpendicular velocity component obtain gyration

influence, which increases with the angle between the velocity and the y-axis. In other

words, the constant magnetic field that is parallel to the applied electric field confines

the charged particles in the plasma. Thus, a flat plasma band can be obtained on the

following conditions: applying a hot cathode electron source, using a collimating

aperture, and applying an external magnetic field directed in parallel to the applied

electric field.

2.3.8 Sputtering in the plane discharge.

A sputtering process begins with supplying a high negative voltage to the

target. This process, as shown above, consists in the ion bombardment of the

(64) 0Bevdt

dvm ×= ⊥⊥

28

sputtering target. Therefore, a part of the ions leave the positive plasma column. The

sputtering process in a system with plane forced discharge occurs in the direction (z-

axis) perpendicular to the anode-cathode axis (see figure 4). We can write the

equations describing the ion collection process and the sputtering process for this

direction. The concentration of ions in the plasma band is homogeneous in the y and

x-directions since it is defined by the Helmholtz coil pair26. The ions formed by the

discharge in the triode sputtering system are collected by the entire target surface.

This is a very important advantage of the described sputtering process in comparison

with the magnetron sputtering, in which sputtering is not homogeneous11. Also, the

sputtered particles in the magnetron sputtering were shown to be ionized27, which is

excluded with the triode sputtering.

The floating potential of the positive column, which usually amounts to

several tens volts, does not enable any particles to be sputtered from the target

surface. The commonly used sputtering voltage ranges from 500 to 3,000 V, which

makes it possible to neglect the value of floating potential during sputtering. One can

consider (with idealization) the plane discharge and the surface of sputtering target as

two infinite plane-parallel plates, similar to a plane-parallel capacitor. One of the

plane-parallel plates emits charged particles (ions) and has the potential equal to that

of the plasma. The other plate has the sputtering potential Vt (the target potential) and,

in the first approximation, perfectly absorbs all the arriving charged particles. Let us

assume that the plasma plate emits only one type of particles, with charge +q and

mass mi, at zero velocity. The sheath edge formed by the discharge is defined as a

position where the ions reach the ion sound speed. The potential in the described

capacitor is given by the Poisson equation, which we can write, for the sake of

simplicity, in the one-dimensional form:

29

The particle velocity at a point with potential V is then

If the emitted ion current density is J = qiniv, the ion density at point z will be

and the Poisson equation can be written as

Integrating the equation from z = 0 to z = d yields the following relation for the space-

charge-limited flow29:

or, substituting the value of the elementary charge qi as qi = e:

which is the Child-Langmuir 3/2-power law for space-charge-limited current flow

between the two planes separated by a distance d with a potential Vt between them.

)66(nq

dzVd

0

ii2

2

ε

−−−−====

)67 (m

Vq2v

5.0

i

i

=

)68(

m)z(Vq2

q

J)z(n 5.0

i

ii

i

=

)69(m

Vq2Jdz

Vd5.0

i

i

02

2

−

ε

−=

)70(dq2

mJV32 4

1

i

i21

0

43

t

ε

=

)71(d

Vm

e294j 2

5.1t

5.0

i0

ε=

30

The current density obtained in the formula (71) is the target ion current. This current

can be measured during sputtering only.

Now let γt be the yield of sputtered atoms per incident ion and ρt the atomic

density of the sputtering target material. Then the sputtering rate can be written as:

(72) [cm/s] 1ejR

t

its ρ

γ=

31

2.4 Triode sputtering application

2.4.1 Triode sputtering system implementation.

The experiments with the triode sputtering system were carried out on a

laboratory deposition setup equipped by a diffusion pump with the residual vacuum of

about 2×10-5 Torr. The setup (see figure 6) was adapted for triode sputtering.

Fig. 6. External view of the triode sputtering setup.

A schematic diagram of the triode system in use is shown in figure 7. The

thermoionic1 cathode is a heated cathode prepared of the tungsten wire 0.3 mm thick

and braided to a spiral of 2 mm in diameter and 20 mm long. The anode was prepared

of a squared tantalum foil 0.5 mm thick, with an area of about 7×8 cm2. A copper

collimating aperture clips and shapes the electron beam directed to the anode. The

sputtering target is placed onto the water-cooled target holder mounted

perpendicularly to the anode - cathode axis. The substrate holder is located opposite to

the target for fuller utilization of sputtered particles. Two external electromagnet coils

32

of 34 cm in diameter arranged co-axially with the anode - cathode axis were used to

create a homogeneous magnetic field in the discharge space.

Fig. 7. Schematic diagram of the triode sputtering system.

2.4.2 External electromagnet.

Two coils placed at a specified distance (the Helmholtz arrangement) can be

used to produce a very uniform magnetic field26. Figure 8 shows a schematic

Helmholtz coil pair arrangement. Here P is the center point between the coils (z = 0).

Fig. 8. Schematic presentation of Helmholtz coil pair arrangement.

33

The magnetic field Bsign along the axis of a current loop of a single coil turn is given

by39

where R is the radius of a coil and the distance between the two coils, and z is the

distance from the zero point that defines the center of the coil. The value B of the

magnetic field along the axis of a circular coil of N turns is given by

For the case of two coaxial coils with N turns each and the current I running through

the coils, the magnetic fields along the axis are adding:

Figure 9 presents the estimated ideal magnetic field obtained by solving the

equation (75) for the real magnetic coils with a radius 34 cm, 450 turns for the current

of 3 A. The solution was obtained with the help of the MathCad software40. A

precise solution for the realistic case is unavailable: calculation of even a simplest

case of the off-axis magnetic field leads to elliptic integrals41.

In our sputtering system, the magnetic coils were made of a copper wire with 450

turns, 1 mm in diameter. The summary inductance of the two coils was 296 mH.

Since our real arrangement was not an ideal Helmholtz pair, we measured the value of

the axial magnetic field in the sputtering setup experimentally. Figure 10 presents the

(74) )zR(

12

NIRNBB 2/322

20

sign +µ

==

(75)

2RzR

1

2RzR

12

NIRBBB 2/32

2

2/322

20

21

++

+

−+

µ=+=

(73) )zR(2

IRB 2/322

20

sign +µ

=

34

experimentally measured magnetic field values. The magnetic field was measured

using the Digital Teslameter DTM-132 of Group 3. The current running through the

coils was 3 A.

Fig. 9. Theoretical distribution of magnetic field in the Helmholtz pair.

-2 0 2 4 6 8 10 12 14 16 180

102030405060708090

100110120130140150160170180190

Internal field between coils (z = 0) Edge field of the coil (z = 5 cm) Magnetic field outside of the coil (z = 8 cm) Magnetic field outside of the coil (z = 11 cm)

Axia

l mag

netic

fiel

d, G

s

Distance from the coil center, cm

Fig. 10. Measured magnetic field in the triode sputtering setup.

35

2.4.3 Internal plasma parameters.

The plane gas discharge was studied by the probe method19. A metal probe

inserted in a discharge gap and biased positively or negatively to draw the electron or

the ion current, is one of the most efficient tools for plasma diagnostics. This probe,

called the Langmuir probe, together with its typical voltage-current characteristic is

shown in figure 11. In this picture VB represents the bias potential on the probe, Φf is

known as the floating potential, and Φp is the plasma potential.

Fig. 11. Definition of voltage and current for a Langmuir probe (a)

and a typical I-V characteristic for a Langmuir probe (b).

At the probe voltage VB = Φp, the probe is at the same potential as the plasma and

mainly draws current from the more mobile electrons. This current is designated as

positive current flowing from the probe into the plasma. For VB exceeding this value,

the probe current tends to saturate at the electron saturation current. The saturation

current is defined by the probe geometry. At VB < Φp, electrons are repelled in

accordance with the Boltzmann relationship, until at Φf the probe is sufficiently

negative with respect to the plasma, so that the electron and ion currents are equal,

and I = 0. For VB < Φf, the current is mainly ion current (negative with respect to the

plasma) and tends to saturate at the ion saturation current, which may also vary with

the applied voltage and due to changes in the probe effective collection area. The

value of the ion saturation current is much lower than that of the electron saturation

36

current due to a much greater ion mass. The electron temperature may be estimated

from the probe current graph at Φf < Vb < Φp region in fig. 11 using the Boltzmann

distribution for electrons116.

The measuring scheme and the probe construction are dictated by the experimental

conditions and the plasma parameters. The type of a model to be best used for

estimation of the plasma characteristics is defined by the interrelation between three

basic parameters: rp (the probe radius), λD (Debye radius, see formula (28)), and λ

(the mean free path, see formula (3))19. As mentioned above, the probe is surrounded

by a layer of volume charge. This layer thickness is very important for the

measurement. One can describe three typical cases:

1. λ >> rp >> λD, thin layer of volume charge;

2. λ >> λD >> rp, collisionless thick layer of volume charge;

3. λD >> λ >> rp, thick layer of volume charge with collisions.

Figure 12 presents a side view of the Langmuir probe used in our experiments.

The probe made of tungsten wire 0.25 mm in diameter was placed in a ceramic tube.

The active part of the probe was 3.4 mm long.

Fig. 12. A side view of the applied Langmuir probe.

If the pressure in the vacuum chamber does not exceed the level of 1*10-3

Torr, the mean free path according to formula (3) will be equal to λ = 5 cm. Weakly

ionized plasma contents approximately 0.01 % of ionized particles in the volume

unit21. The gas molecule concentration can be obtained from the known formula7:

p = ngkT (76)

37

Here p is the pressure in the vacuum chamber, ng molecule concentration, and T

the gas temperature. Therefore, the electron concentration in the argon atmosphere

with the pressure of 1*10-3 Torr, is approximately 3.56*109 cm-3. If we assume that

the electron temperature of the weakly ionized plasma is ~ 1 - 10 eV4, 115 (see page

13), the Debye radius will be equal to approximately λD = 0.125 0.395 mm (from

formula (28)). Comparison of the parameters rp, λD, and λ shows that they satisfy the

second case: λ >> λD >> rp: collisionless thick layer of the volume charge. In this case

we can estimate the electron temperature in the plasma from the experimental I-V

characteristics of the Langmuir probe, using the known formula19:

In such a way, the electron temperature is equal to the slope angle of the Langmuir

probe characteristic in the semi-logarithmic scale. This ideal characteristic ignores

the perturbation processes such as bombardment of the probe by high-energy

electrons, emission of secondary electrons from the probe, and etching away of the

probe.

The collisionless condition allows us to use the Bohm approximation45 to

estimate the ion concentration in the plasma:

where jis is the probe current saturation density and vB is the Bohm velocity.

(77) dV

JlndkeT

-1

B

ee

=

(78) mkT

envenji

eiBiis =≈

38

2.4.4 Practical Langmuir probe measurement.

Figure 13 presents the electrical schematic diagram of Langmuir probe measurement.

The measuring probe was placed in the plasma and supplied by the bias voltage

ranging from 40 V up to + 40 V through the resistor R = 5.7 Ω.

Fig. 13. Principal electrical scheme of the Langmuir probe measurement.

Measurements were taken in various points of the plasma at different pressures in the

vacuum chamber. The distance a between the Langmuir probe and the plasma axis

(see figure 14) was also varied in the experimental trials.

Fig. 14. Dimensions of the setup interior arrangement (mm)

39

The plasma in our deposition setup is sustained by the thermo-emission

current. The anode current, or the plasma current, depends on this thermo-emission

defined by the cathode temperature. This dependence is shown in figure 15.

Fig. 15. Dependence of anode current on thermo-emission cathode current.

As seen from the figure, the plasma current is almost directly proportional to the

thermo-emission cathode current. This fact follows from equations (1) and (2), where

the plasma current is defined by the 3/2 law and is limited by the cathode emission

described by the Richardson equation. Thus, the plasma density is dependent on the

cathode current. The vacuum level was found to be nonessential in the studied interval

of pressures.

5.3 5.4 5.5 5.6 5.7 5.8

40

60

80

100

120

140

160

180

200

220

240

260

280

300

320

340

360

PAr = 0.5 mTorr PAr = 1 mTorr PAr = 1.5 mTorr

Anod

e cu

rrent

, mA

Cathode current, A

40

Figure 16 presents typical I-V characteristics of plasma measured with the

Langmuir probe. The characteristics were measured for the four pressure values: 0.49

mTorr, 0.8 mTorr, 0.96 mTorr, and 1.5 mTorr. Such plasma parameters as the cathode

current, the anode voltage and the current of the external electromagnet were

maintained constant. The figure shows that the floating plasma potential decreases

with the pressure in vacuum chamber.

Fig. 16. Langmuir probe current vs applied bias.

We can easily see that in the described plasma, there are many different independent

parameters, such as: the external magnetic field, the cathode current, the plasma

voltage, and the vacuum level. Figure 17 illustrates the effect of the external magnetic

field (the electromagnet coil current) on the Langmuir probe current. The insertion

shows the characteristics in the semi-logarithmic scale. As demonstrated, the slope

angle decreases slightly with the magnetic field reduction. Therefore, the electron

temperature decreases also.

-40 -20 0 20 40-2000

0

2000

4000

6000

8000

10000

12000

14000

Ic = 5.49, AIh = 2 AUa = 35 V

1.5 mTorr 0.96 mTorr 0.8 mTorr 0.49 mTorr

Lang

mui

r pro

be c

urre

nt, µ

A

Applied probe bias, V

41

-40 -20 0 20 40

-2000

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

Argon pressure, 1 mTorrCathode current, 5.3 AAnode voltage, 100 V

Ih = 2.5 A Ih = 2 A

Lang

mui

r pro

be c

urre

nt, µ

A

Applied probe bias, V

Fig. 17. External magnetic field effect on the Langmuir probe current.

This fact can be explained by a lower confinement of electrons undergo with the

weaker magnetic field that leads to decreasing the electron lifetime. The measured

data from figure 17 were used to estimate the electron temperature distribution. For

this purpose, we considered the range between the floating potential point and the

plasma potential (the positive probe bias)19. In this region, the formula (77) transforms

to the practical form:

The estimated results are shown in figure 18. The analysis of the characteristics

shows, on the average, a nearly linear rise of the electron temperature with the

positive probe bias. The electron temperature is lower for higher magnetic field

values. This can be easily explained by more collisions between the particles

occurring under stronger magnetic field.

(79) uu

jlnjlnkeT

1

n1n

n1ne

−

+

+

−−

=

10 15 20 25 30 35 4020.1

54.6

148.4

403.4

1096.6

2981.0

8103.1

Prob

e cu

rrent

, µA

Applied probe bias, V

42

Fig. 18. Electron temperature dependence on the external magnetic field.

The following four figures show the Langmuir probe characteristics measured

in different points of the interelectrode space. All the characteristics were obtained for

the same magnetic field, at three pressure levels. The basic variable parameter was the

distance between the Langmuir probe and the plasma axis (the anode-cathode line).

This parameter a was shown in fig. 14.

-40 -20 0 20 40

-1500

-1000

-500

0

500

1000

1500

Ic = 5.4 AIh = 2 AUa = 70 Va = - 10 mm

PAr = 0.5 mTorr PAr = 1 mTorr PAr = 1.5 mTorr

Lang

mui

r pro

be c

urre

nt, µ

A

Applied probe bias, V

Fig. 19. Langmuir probe characteristics for a = -10 mm.

5 10 15 20 25 30 35 40 450

1

2

3

4

5

6

7

8

9

Argon pressure, 1 mTorrCathode current, 5.3 AAnode voltage, 100 V

Ih = 2.5 A Ih = 2 A

Elec

tron

tem

pera

ture

, eV

Positive probe bias, V

43

-40 -20 0 20 40

-1200

-1000

-800

-600

-400

-200

0

200

400

600

800

Ic = 5.38 AIh = 2 AUa = 70 Va = 0

PAr = 0.5 mTorr PAr = 1 mTorr PAr = 1.5 mTorr

Lang

mui

r pro

be c

urre

nt, µ

A

Applied probe bias, V

Fig.20. Langmuir probe characteristics for a = 0.

-40 -20 0 20 40

-10000

0

10000

20000

30000

40000

50000

Ic = 5.81 AIh = 2 AUa = 45 Va = 10 mm

PAr = 0.5 mTorr PAr = 1 mTorr PAr = 1.5 mTorr

Lang

mui

r pro

be c

urre

nt, µ

A

Applied probe bias, V

Fig. 21. Langmuir probe characteristics for a = 10 mm.

44

-40 -20 0 20 40-2000

0

2000

4000

6000

8000

10000

Ic = 5.83 AIh = 2 AUa = 70 Va = 20 mm

PAr = 0.5 mTorr PAr = 1 mTorr PAr = 1.5 mTorr

Lang

mui

r pro

be c

urre

nt, µ

A

Applied probe bias, V

Fig. 22. Langmuir probe characteristics for a = 20 mm.

These graphs characterize the plasma in our sputtering installation. The plasma

density was estimated from the experimental data (see figures 19-22) substituted into

formula (78). It is of interest to present the plasma density and the electron

temperature obtained from the experimental data and equations (77) - (78) in the form

of three-dimensional graphs. Figure 23 presents the electron temperature distribution

as the function of two variables: the vacuum pressure and the position of the probe in

the interelectrode space.

45

Fig. 23. Electron temperature of plasma in the interelectrode space.

0.500.75

1.001.25

1.50 -10

0

10

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

Plas

ma

dens

ity, *

1011

cm

-3

Distan

ce fr

om th

e

plasm

a axis

(a),

mm

Argon pressure, mTorr

Fig. 24. Effect of argon pressure on plasma density.

0.500.75

1.001.25

1.50 -10

0

10

2.02.53.03.54.04.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0El

ectro

n te

mpe

ratu

re, e

V

Distan

ce fr

om th

e

plasm

a axis

(a),

mm

Argon pressure, mTorr

46

Both plasma density and electron temperature grow with the pressure. At the same

time the influence of the pressure is weaker than that of the position in the vacuum

chamber. The electron temperature is at its maximum on the plasma axis.

2.4.5 Thin film deposition by the plane discharge.

Titanium and silicon targets were chosen as experimental sputtering materials.

These targets had a form of a disk 42 mm in diameter and 1 mm thick and were

placed on the water-cooled target holder perpendicularly to the plasma plane. The

experimental triode sputtering system is shown in figure 25.

Fig. 25. Applied triode sputtering system.

In this system the thermo-cathode is supplied by AC voltage through a transformer.

The negative pole of the DC anode supply is connected to the cathode. The anode

voltage can be varied in the range of 0 to 100 V. The sputtering target is supplied with

a high negative voltage of 500 - 3000 V. The positive pole of the supply is connected

to the tungsten ring electrode placed into the plasma. This electrode, called the

support electrode, serves as a reference electrode for the sputtering target. It takes on

the plasma potential Vp, so that the entire energy of the target power supply is applied

47

to the ions between the plasma and the target. This electrode has also an additional

purpose: it collects electrons from the plasma in the form of a conductive cylinder and

thus confines the ion flow to the target. This results in a more homogeneous

deposition of sputtered atoms on the substrate. The support electrode is a novel

element of the developed triode system.

As shown in figure 25, all electrodes and power supplies are isolated from the

ground. Since the vacuum chamber and other installation parts are connected to the

ground, stray (parasitic) discharges are excluded in our system, and the applied power

is used efficiently. Typical parameters of the sputtering are presented in Table 1.

Table 1. Typical sputtering parameters of the triode deposition system.

Number Parameter Symbol Unit Value

1 Residual pressure Pres Torr 3×10-5 2 Argon (work) pressure PAr Torr 5×10-4 3 Electromagnet current Ih A 2 4 Cathode current Ic A 5 5 Anode voltage Va V 50 6 Anode current Ia A 0.9 7 Sputtering voltage Vt V 1000 8 Target current It mA 40 9 Substrate temperature Ts °C 50

All the parameters appearing in the table, except the anode current and the target

current, are independent, so that our system includes 7 independent variables.

Therefore, any technology to be developed for deposition of a specific coating would

require an enormous number of experiments. The method for decreasing the trial and

error number is presented in the next part of the Thesis.

The I-V characteristics of the plane discharge in our setup appeared in figure 26.

These graphs show that the plane discharge is an abnormal discharge2 and can be

successfully used for sputtering.

48

Fig. 26. Discharge current as a function of the applied voltage.

The discharge current is influenced by the argon pressure and the cathode current.

Fig. 27 shows the effect of the argon pressure on the discharge current for two

different values of the cathode current.

Fig. 27. Discharge current as a function of argon pressure.

The plasma (discharge) current is defined by the electron emission from the cathode

and the concentration of argon atoms in the vacuum chamber.

10 20 30 40 50 600

200

400

600

800

1000

PAr = 1.1 mTorr

Ic = 3.22 A Ic = 3.50 A Ic = 3.49 A

Dis

char

ge c

urre

nt, m

A

Applied discharge voltage, V

0.6 0.7 0.8 0.9 1.0 1.1 1.2500

600

700

800

900

1000

Ic = 3.50 A Ic = 3.49 A

Dis

char

ge c

urre

nt, m

A

Argon pressure, mTorr

49

2.5 Tetrode sputtering application

The triode sputtering method is controlled by several independent physical

parameters. Introducing an additional level of freedom can improve the process

controllability. The novel independent parameter we introduce for this purpose is the

distance between the sputtering target and the plasma. In our sputtering system, this

parameter is varied independently. This additional parameter enables controlling the

ion current while maintaining the same target potential and the same sputtering

particles energy. Figure 28 presents the vacuum chamber equipped with a device

enabling us to move the plane plasma discharge in the target direction. This is done

with the help of a magnetic screen installed around the target. This magnetic screen

serves as the fourth electrode.

Fig.28. Tetrode sputtering system.

The magnetic screen is shown in figure 29. The screen is asymmetric and is

assembled from permanent magnet plates with poles on their largest planes. If the

North poles of the screen are directed in parallel to the anode-cathode axis (as shown

in figure 28), the plasma moves to the target. Rotating the screen with respect to the

50

target axis around the target holder varies the distance between the plasma and the

target.

Fig. 29. External view of the magnetic screen.

The dependence of the target current on the magnetic screen direction is shown in

figure 30.

1 2 330

32

34

36

38

40

42

44

46

48

1 - without magnetic screen2 - S-S position of the magnet3 - N-N position of the magnet

Targ

et c

urre

nt, m

A

Magnetic screen position

Fig. 30. Target current versus magnetic screen position.

This influence can be explained by interaction of two magnetic fields: the external

homogeneous magnetic field and the asymmetrical magnetic field of the control

screen. Plasma as a conductive substance is affected by magnetic fields. The

51

intersection of the magnetic fields serves as a trap for the plasma body and moves it in

the corresponding direction.

We have evaluated the mechanical and electrical properties of the Si and Ti

thin films deposited by the described sputtering method. The sheet resistance was

measured with a standard four-point probe method, and the film thickness was found

by a Tencor Instruments profilometer. The film surface structure was studied on a

Leica Stereoscan 430 Scanning Electron Microscope (SEM) operating in 20 KeV. The

composition of the deposited films was evaluated by the Energy Dispersive

Spectrometry (EDS) method, with the help of the add-on EDS device mounted on the

STEREOSCAN-430. The microhardness of deposited films was measured using a

standard tester PMT-3 manufactured in Russia65, 66. The indentation period was 15

seconds; 5-10 indentations were taken for each specimen. The glass substrates were

cleaned with isoprophanol.

All the films were semi-transparent (they were thin), since the sputtering

duration did not exceed 15 min in all the trials. The method has revealed high

deposition rates. Table 2 presents the deposition rates compared to the data obtained

by ion beam sputtering method and reported by J.P. Greene43.

Table 2. Sputtering deposition rates with Argon, Å/min (Vt = 1000 V).

Element Ref [43] Ref [44] Our results Titanium 50 21 60 Silicon - 35 100

The deposition of Ti films was carried out with the sputtering voltage of 1000 V and

the Argon pressure of 1×10-3 Torr. The target current was 46 mA, and the control

magnetic screen was placed in the N-N position (see fig. 30). The obtained Ti film

was 600 Å thick and its microhardness was ~ 315 kG/mm2.

52

The target current shows a linear dependence on the argon pressure as seen

from figure 31. This dependence, as well as the directly proportional relationship

between the sputtering rate and the target current, is clear from formula (72). In

contrast to the magnetron sputtering11, in the presented system the sputtering target

sputters homogeneously. It is of interest to note that although not controlled, the

substrate temperature did not exceed 60° C after 15 min of sputtering. The

temperature was measured by a standard thermocouple of K-type (Cromega-

Alomega). This warming was due to the thermo-cathode radiation only, since the

plasma was not in contact with the substrate.

Fig. 31. Target current vs argon pressure.

Figure 32 presents a typical relationship between the sputtering voltage and the target

current.

0.4 0.5 0.6 0.7 0.80

5

40

45

Targ

et c

urre

nt, m

A

Argon pressure, mTorr

53

Fig. 32. Target current vs sputtering voltage.

Figure 33 presents the sheet resistance measured for the Ti films deposited

under different deposition conditions. Obviously, the deposition rate affects the film

thickness, and the sheet resistance is defined by the film thickness and density.

Fig. 33. Sheet resistance of Ti films measured by four-point probe method.

200 400 600 800 1000 1200 1400 1600

20

25

30

35

40

45

Targ

et c

urre

nt, m

A

Sputtering voltage, V

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.60

5

10

15

20

25

Ti film, S-S position Ti film, N-N positionM

easu

red

volta

ge, m

V

Applied current, mA

54

The sheet resistance was measured for Si films as well. These data appearing in figure

34 also show a directly proportional I-V dependence.

Fig. 34. Sheet resistance of Si films measured by four-point probe method.

These films were deposited with two different sets of deposition parameters, as shown

in table 3.

Table 3. Sputtering deposition parameters used for the Si films deposition.

Pres PAr Ic Ih Va Ia Ut It t El/m R# Units Torr mTorr A A V mA V mA min Ω/sq.N1 4×10-5 0.7 3.5 2 55 930 1000 36 10 - 23.96N2 4×10-5 0.5 3.1 2 55 740 1500 32 15 N-N 41.16

The data appearing in Table 3 shows a good possibility for fine control of the

deposited film properties. The microhardness of the silicon semi-transparent film was

found to be 465 kG/mm2. This value is sufficiently high for thin Si films, as compared

to the bulk crystalline silicon, which has the microhardness of ~1000 kG/mm2.

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2

3456789

10111213141516171819

Si film, N1 Si film, N2

Mea

sure

d vo

ltage

, mV

Applied current, mA

55

The silicon film was also studied for its composition. Figure 35 shows an EDS

spectrum for the sample N1. As seen, the film mainly consists of silicon and also

contains several additional elements in little portions sputtered from the target holder.

Fig. 35. EDS spectrum of the silicon sample N1.

An external view of this Si film is shown in figure 36.

56

Fig. 36. Electron micro-photography of the Si sample on the glass.

We see the silicon film to entirely repeat the substrate structure, while it is rather

difficult to reveal the inherent film structure. Therefore, this method provides a good

possibility of growing a film repeating the substrate structure.

In the presented work, we have shown a novel sputtering deposition method

and an apparatus for its realization. Now, a table presented in the review of R.

Reuschling 46 (see Table 4) can be successfully completed with our novel method. We

conclude that the triode and tetrode sputtering processes can be efficiently applied in

the microelectronics technology.

Table 4. Different deposition processes and their basic parameters.

Process Type Typical Particle Energy Vacuum (Pa) Evaporation < 1 eV 0.001

Triode (tetrode) Sputtering 500 … 3000 0.05 Magnetron Sputtering 10 100 0.5

Diode Sputtering 1000 10

57

3. OPTIMIZATION AND IMPROVEMENT OF THE DEPOSITED FILMS

3.1 Mathematical modeling of the sputtering process with the random section

method

Thin film vacuum deposition processes are influenced by many independent

input parameters (variables), including the source, the wall and the substrate

temperatures, the residual, total and partial pressures, the power source strength, the

vacuum chamber dimensions, the process duration, etc. Post-deposition treatment,

such as thermal annealing, can also be considered as one of the independent

deposition parameters. There are several different methods used to empirically study

the relationships between one or more of the measured response functions, on the one

hand, and a number of input parameters, on the other hand. One of the methods that

can be used for the optimization of multiparameter systems and complex processes is

the multisimplex method47, or the Taguchi approach48.

The simplex methods are based on the initial design of k+1 trials, where k is

the number of variables. A k+1 geometric figure in a k-dimensional space is called a

simplex. With two variables, the first simplex design is based on three trials. This

number of trials is also the minimum required for defining the direction of the process

improvement. After the initial trials, the simplex process is sequential, with the

addition and evaluation of one new trial at a time. The simplex searches

systematically for the best levels of the control variable. Therefore, it is a timesaving

and economical way to start an optimization process.

The Taguchi approach is based on the design of experiments to provide near

optimal quality characteristics for a specific objective. The Taguchi method includes

the integration of statistical design of experiments into a powerful engineering

process. The goal is to optimize an arbitrary objective function and to reduce the

58

sensitivity of engineering designs to uncontrollable parameters or noise. The objective

function used and maximized in the present case is the signal-to-noise ratio. This

method involves three steps of the optimization process:

1. System design (development of a system designed to function under nominal

conditions, with a given initial set of input parameters);

2. Parameter design (selection of optimal levels for the controllable system

parameters);

3. Tolerance design (a narrower tolerance range must be specified for those design

factors whose variation imparts a large negative influence on the output

function).

The Taguchi method can significantly reduce the number of experiments needed to

find the response function, by improving the efficiency of generating information

needed for the system design.

The response surface methodology (RSM) is an optimization approach, which

uses mathematical and statistical techniques to search for the best combination of the

process variables. The response surface represents the domain of all feasible solutions

for the process model49. The response surface approach is the best method for an

empirical study of the relationships between one or more of the measured response

functions such as resistivity, transparency, density, on the one hand, and various input

parameters, on the other hand50. After such a study, the RSM method enables the

process optimization with a minimum of trial and errors. If the process model is

presented as mathematical function, one can say that the response surface is a trace of

the response function (the main parameter) in the multi-parameter space. Then it can

be written in the form of F = f(X1,X2,..Xm), where X1, X2, and Xm are the

independently controlled measured input variables of the sputtering process.

59

In the narrow interval of the variation of the input variables, the

response function may be considered as an analytical function. The approximation

process may begin as a polynomial model with linear independent coefficients51.

Therefore, it may be expanded in a Taylor series over the chosen area in the form:

(80) )xx(!n

)x(f)x(f n0

0n

0)n(

−= ∑∞

=

By choosing the first three terms of the above expansion, a second order

approximation is obtained:

(81) )xx(dx

)x(fd21)xx(

dx)x(df)x(f)x(f 2

020

2

00

0 −+−+=

This expression has the form of a parabolic function with linear coefficients:

f(x) = a + bx + cx2 (82)

For a multi-parameter process of m levels, i.e. with m independently controlled

measured input variables, the expression is transformed to a model of the following

type, without the high order (in our case, third order) interaction effects:

F = c0 + c1X1 + c2X2 + c3X3 + + cmX1m + cm+1X1X2 + cm+2X1X3 +

+ + ckXm-1Xm + ck+1X12 + ck+2X2

2 + + ck+mXm2 (83)

where F is the response function, ci are the model coefficients, and Xi are the

processing parameters.

A standard form for a planned series of experiments for evaluating the normalized

process parameters is shown in table 5 52. The parameters are normalized using the

following equation:

Xi* = 2(Xi X0)/∆X (84)

where X0 is the midpoint of the variable range, and ∆X is the variation range.

Table 5. Standard form for model coefficient estimation: F is the response function

and Xi* are the normalized process parameters.

60

N X1* X2

* X3* ... XN

* F 1 +1 +1 +1 ... +1 2 +1 +1 +1 ... -1 ... ... ... ... ... ... 2m -1 -1 -1 ... -1

Optimization of this mathematical model is obtained by moving in the direction of the

functions gradient slope. The function optimum (extreme) is achieved under the

following conditions:

∇f (X1X2Xm) = 0 (85)

and

∂ 2 f (X1X2Xm) < 0 (for the maximum). (86)

Obviously, in order to create a complete multi-factor space from the m

variable system, a minimum of 2m separate experiments are required; the constants

can then be calculated from the 2m resulting equations. In addition, a few more

experiments in the system center (Xi* = 0) are required in order to estimate the error

of the calculated result. However, from a practical point of view, there is no need to

run all of these experiments. It is possible to complete only a few experiments and

then deduce the missing coefficients using the mathematical procedure known as

estimation on partial sections in multi-parameter space53. According to this

procedure, each one of the experiments is actually a partial section of the chosen

multiparameter space. Consequently, the coefficients of the approximation model

must be related for all partial sections. Thus the missing coefficients can be calculated

by varying one parameter in the following quadratic model:

c c x c x F x xc c x c x F x xc c x c x F x x

k k

k n k n

k m k m

1 2 32

1 2 32

1 2 32

++++ ++++ ==== ====

++++ ++++ ==== ====

++++ ++++ ==== ====

++++ ++++

++++ ++++

( )( )( )

(87)

61

where ci is the missing coefficient, x the process parameter (X1,X2,..XN), F the

response function, and m, n are indices. The rest of the missing coefficients describing

the required model are then obtained by solving the coefficient matrix with the

Kramer method.

In the case of a large deviation between the measured data and the model, the

approximation order, i.e. the order of approximation equation, should be increased.

3.1.1 Deposition of In2O3 conductive transparent coatings by DC sputtering

3.1.1.1 Introduction.

Transparent conductive coatings with high electrical conductivity and the

maximum optical transparency attracted much attention in recent years. These

coatings have many practical applications, such as, for instance, transparent electrodes

in microelectronics, optoelectronic devices and photovoltaic systems.

Indium oxide thin films possess the electronic properties of n-type

semiconductors with a band gap in the range of 3.5 to 4 eV. Oxygen vacancies in the

films provide free electrons by acting as doubly charged donors. Since the indium

oxide films generally suffer from oxygen deficiency, the electron gas in the

conduction band is degenerate54. These transparent films are often alloyed with tin

oxide (5 to 10 %); such films are known as indium tin oxide (ITO). These films have

a higher free charge concentration, which increases the electrical conductivity55, 56.

Films without the tin admixture have higher resistivity and lower transparency than

ITO57. The tin is commonly believed to provide free electrons by substituting for the

indium atoms and acting as a singly charged donor at the indium site. Other alloys of

indium oxide with elements from the IV, V or VI columns of the periodic table have

62

been reported58. An attempt to deposit by DC magnetron sputtering the insulating

indium oxide films for creating tunnel junction barrier layers has also been reported59.

Most of the unalloyed indium oxide films were prepared by different

evaporation methods. These methods had one common characteristic: injection of

oxygen in the vacuum chamber (reactive evaporation)60, 61. The influence of oxygen

on the partial pressure of the structure of the reactively deposited indium oxide films

has been described in ref. [62]. In ref. [63], the influence of the substrate temperature,

deposition rate, base and background pressures, and alloy constituents concentration

were studied. However, the films with the maximum transparency and minimum

resistivity have been prepared by sputtering 64.

Most of the literature in this field has concentrated on the physical analysis of

ITO coatings. In our work, we present a different approach to the manufacturing of

transparent conductive thin films. In addition to considering the physical properties of

the final product, we developed a mathematical model of the processing stages based

on the above mentioned response surface methodology (RSM). Using our process

model enabled us to optimize the technological parameter values more efficiently,

with a minimum number of trial runs. The modeling method was applied and tested to

the known magnetron sputtering53, 82, 117.

3.1.1.2 Experimental procedure.

The experiments were carried out on a laboratory magnetron sputtering

vacuum station (see figure 37), equipped with a diffusion pump. A principle

schematic diagram of the magnetron sputtering source is shown in figure 38. The

residual pressure prior to the deposition process was less than 4×10-5 Torr and the

sputtering was done in a pure argon atmosphere (99.996%). The substrate temperature

was varied during the deposition process and the samples underwent a post-deposition

63

heat treatment, while still in vacuum. The indium oxide films were deposited on a

borosilicate glass plate 0.13-0.17 mm thick and on an optical glass 1 mm thick. The

sputtering target was in a shape of a round disk, 50 mm in diameter and 3 mm thick,

made of pure indium oxide (99.999%). The substrate-to-target distance was taken as 5

cm.

Fig. 37. Experimental magnetron vacuum station.

We evaluated the optical and electrical properties of the resultant In2O3 films.

The optical transparency of the films and absorption for 250 < λ < 880 nm was

determined on a Uvicon 941 Plus spectrophotometer. Transparency in the IR

spectrum was measured in the range of 400 to 2500 nm. A computerized metallurgical

microscope with magnification × 80 to × 1600 was used for the microtopography

evaluation. The microhardness of the resulting films was measured using a tester

PMT-3. The indentation period was 15 seconds; 5 to 10 indentations were taken for

each specimen. The substrates were cleaned with alcohol in the ultrasound heater for

5 min.

64

Fig. 38. Magnetron sputtering source, schematic diagram

The most significant properties of In2O3 films are their transparency and

resistivity. These properties were defined as the response functions. The active

technological parameters for the sputtering model were taken as follows: PAr - argon

pressure (1.5 ÷ 2 Torr), Vt - DC voltage to the target (500 ÷ 750 V), Ts - substrate

temperature (150 ÷ 250° C), and t - deposition duration (20 ÷ 40 min), and the model

was written in the form: F = f(P,V,T,t). A complete factor experiment of the order of

24 yields the following regression equation with 15 coefficients:

2

152

s142

t132

Ar12s11t10

st9Ar8sAr7tAr65s4t3Ar21

taTaVaPatTatVa

TVatPaTPaVPataTaVaPaaF

++++++

+++++++++= (88)

3.1.1.3 Practical model implementation.

Figure 39 illustrates an example of a bi-parameter space. The coordinate axes

X and Y represent two active technological parameters. Axis Z, in its turn, represents

the response function, which can also be named an object function. The response

function draws a surface in the multi-parameter space, called the response surface.

The response function can be studied with the help of the RSM method and then

65

optimized. A standard orthogonal plan for this experiment can be presented in a

tabular form (see Table 6), which is also convenient for numerical calculations50, 52.

Fig. 39. Example of bi-parameter experiment space.

This table contains the selected normalized active technological parameters: P*, V*,

T*, t* and the defined response functions. The normalization was done by formula

(84).

Table 6. Normalized active technological parameters.

N P* V* T* t* D RS 1 +1 +1 +1 +1 2 +1 +1 +1 -1

16 -1 -1 -1 -1

It is obvious that in order to create a complete 15-constant parameter space

from a 4-variables system, 15 separate experiments should be carried out. These

constants will be then calculated from the 15 obtained equations. In addition, it is

necessary to perform a few more experiments in the system center in order to obtain

an estimation of the relative experiment error. A virtual example of a multi-parameter

space experiment and a random partial section of it is shown in figure 40. Each one of

the experiments is actually a partial section of the chosen multi-parameter space.

66

Fig. 40. Illustration of random partial sections of the multi-parameter space.

These random partial sections are described by the same equations and belong to the

analytical response function. Consequently, the coefficients of the approximation

model must be equal for all partial sections. Thus the missing coefficients can be

calculated by varying one parameter using the equations (87). The rest of the missing

coefficients describing the required model are then obtained by solving the coefficient

matrix. One should always bear in mind that only the first few resultant coefficients

were experimentally obtained, while the rest of the coefficients were derived

mathematically from following matrix (i is an index):

( ) ( ) ( )ii2i

2i

2i

2iiiiiiiiiiiiiiiii Fa tT V P tT tV TV tP TP VP tT V P 1 ⋅= (89)

At this point the optimization model of equation (85) can be implemented.

Thus, the optimization problem is transformed to the following gradient relation:

0

tT V P

)t,T,V,P(F)t,T,V,P(F s

t

Ar

stAr**** =

∂∂∂∂

⋅∇=∂ , (90)

where F(P*,V*,T*,t*) is the required extreme. Analysis of this relation yields 16

possible combinations (24) between the four variables. Similarly, analysis of the

second derivative yields the type of extreme: a maximum for the transparency and a

minimum for the resistivity.

67

3.1.1.4 Experimental results.

The adhesion of the obtained films to glasses was extremely good. A

qualitative examination by steel scribing showed the adhesive forces to correlate with

a transparency peak on the wavelength of 400 nm. On this wavelength, the adhesion

increases with increasing of transparency. The film morphology shows a very

homogeneous smooth surface without hillocks usually observed on evaporated films.

The diagram in figure 41 shows the dependence of the partial oxygen pressure and the

substrate temperature on the film phase composition62.

Fig. 41. Evaporated ITO film composition diagram.

This diagram is drawn up for the films obtained by thermal evaporation of pure

indium in the oxygen atmosphere. The diagram shows the variation range of the

technological parameters that enable the indium indium oxide mixture: the substrate

temperature T = 60 to 250 °C and the partial oxygen pressure Po ≤ 1.5×10-4 Torr.

In our case, there are two important distinctions: the sputtered particles are

partially excited27 atoms of In and O with energies in tens times higher than those of

evaporated particles. Besides, the oxygen partial pressure was lower than 1×10-5 Torr

(residual gas only). Therefore we can assume that our films are mixtures of different

oxides of indium (InO + In2O + In2O3), which have different temperatures of

formation and heat capacity64. Under the conditions of free oxygen atom deficiency,

68

complete oxidation of all sputtered particles is impossible. Fig. 42 shows a typical

transparency characteristic near the absorption edge.

Fig. 42. Transparency characteristics of In2O3 on the borosilicate glass.

The adhesion increasing with the transparency shift to UV direction can be explained

by ordering of the deposited particles in the growing film while the sputtered particle

energy increases. The band gap of the grown films increases with further In oxidation.

In addition, the covalence bonds between the absorbed particles and the surface

change the Van-der-Vaalse forces, which looks like increased adhesion. As shown in

figure 42, the maximum transparency of the coating reaches up to 98%, which yields

an antireflection effect. The energy band gap of the obtained In2O3 films was found

by plotting the relation

2/1g )Eh(A)h( −ν=να (91)

where Eg is the band gap, for direct impurity transitions60. The least square method

was used for extrapolation and finding the intersection point of the function (91) and

300 400 500 600 700 800 900

0

20

40

60

80

100

glass with film glass

Tran

spar

ency

, %

Wavelength, nm

4 3.5 3 2.5 2 1.5Energy, eV

69

the abscissa axis, i.e. the band gap59. The maximum band gap value was found to be

4.08 eV for the film with the maximal adhesion.

The electrical properties of the obtained films were stable enough. Figure 43

illustrates a typical four-point measurement on the obtained films.

Fig. 43. Typical measurement of surface resistance

This characteristic representing a linear function shows that under the normal

conditions, all the carriers participate in the current transfer.

The active technological parameters affect the electrical, mechanical and

optical properties of the deposited films during deposition. Figures 44, 45 illustrate

this influence. Figure 44 shows the influence of the type of substrate and of the

growth rate on the resistivity of deposited films. Figure 45 presents the influence of

the growth rate on the microhardness and transparency, on the wavelength of 550 nm.

0.1 0.2 0.3 0.4 0.5 0.60.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

Mea

ssur

ed c

urre

nt, m

A

Applied voltage, V

70

Fig. 44. Resistivity of In2O3 film for different types of substrates.

Fig. 45. Influence of growth rate on In2O3 film properties.

Table 7 summarizes the experimental data. Each row represents a trial and is a

random section of the multi-parameter space. Based on these results, the coefficients

were calculated for best transparency and resistivity, using the quadratic model (87).

00.050.1

0.150.2

0.250.3

0.350.4

0.450.5

0.80 1.24 1.58 2.50Growth rate, A/sec

opticborosilic

1 2 3

82

84

86

88

90

92

Growth rate, A/sec

Tran

spar

ency

, %

Transparency

550

600

650

700

750

Microhardness, kG

/mm

^2

Microhardness

71

Table 7. Experimental data on In2O3 film deposition.

Vt, V V* PAr,Torr P* Ts, °C T* t, min t* RZ, Ω/sq F, %

400 -1.8 2 1 20 -3.6 30 0 463714.2 90.7 700 0.6 2 1 150 -1 30 0 5142.007 86.02 750 1 1.83 0.32 250 1 30 0 5176.905 88.97 750 1 1.67 -0.32 250 1 20 -1 15391.73 90.9 600 -0.2 1.96 0.84 150 -1 40 1 10656.96 91.9 500 -1 1.5 -1 150 -1 40 1 4706.079 93.47

The obtained partial model for the F parameter response (transparency) is:

(92)

t38.5t5.002.86F

T71.0T47.178.86F

V37.3V04.683.90F

P04.0P72.377.89F

2

2ss

2tt

2ArAr

++=

++=

−−=

+−=

According to these expressions, the model coefficients are obtained in the form of a

regression equation, by the standard procedure of the least square method. In this

calculation, the second order interference effects should be considered as well. The

required expression is derived in the following form:

F = 88.96 - 0.93PAr + 0.13Vt - 1.5Ts + 0.37t + 0.005PArVt + 0.005PArTs + 0.006PArt +

+ 0.006VtTs + 0.005Vtt + 0.005Tst + 0.38PAr2 + 0.006Vt

2 + 0.006Ts2 + 0.005t2 (93)

Calculations of the optimal deposition parameters were done using the steep rise

method (90). The optimization was done for two parameters: transparency at λ = 550

nm and sheet resistance. The resultant optimized technological values for these two

parameters are shown in Table 8.

Table 8. Optimized technological parameters.

Vt, V PAr, Torr Ts, °C t, min d, Å RZ, Ω/sq F, % (550nm) 1000 2.03 172 15 720 2438.22 88.23 750 1.76 179 34 2525 1704.03 90.97

72

A SEM microstructure analysis of the films indicated good homogeneity with grain

dimensions smaller than the microscope resolution. Figure 46 represents the

transparency of the film in the large wavelength range.

Fig. 46. The transparency of In2O3 film in the IR wavelength range.

The absorption edge of the film corresponds to the transparency of the In2O3 film at

the energy gap of ~ 3.88 eV. Variation of the process parameters enabled us to control

the shift in the absorption edge value, the value of maximum transparency and the

transparency curve shape. The model error for (82) was evaluated at the point (-1,-1,-

1,1), and the relative error in the chosen space of active parameters was:

(94) %2%10063.91

63.9147.93%100F )1,1,1,1(

=−

=∆

=δ−−−

This small error proves that the proposed model is adequate and can be used to set

magnetron deposition parameters. The designed model coefficients (regression

coefficients) can be presented as coefficients of Taylor expansion around the points of

interest. This approximation method is very convenient, since the accuracy of the

0 500 1000 1500 2000 25000

10

20

30

40

50

60

70

80

90

100

Tran

spar

ency

, %

Wavelength, nm

73

solution system can be enhanced by increasing the degree of the original Taylor

series.

The main disadvantage of this model lies in the need of quadratic

approximation for multi-parameter processes, due to inaccuracy of the linear

approximation. It should be noted that even the quadratic approximation does not

precisely represent the situation since the real function may be too complicated. The

obtained parameters of our In2O3 films are presented in Table 9.

Table 9. Parameters of films after optimization.

Optimization Thickness, Å Transparency, %(for λλλλ = 550 nm)

Sheet resistance,ΩΩΩΩ/sq

Resistivity, ΩΩΩΩ××××cm

Before 3000 90.9 15391.73 0.461 After 2525 90.7 1704.03 0.043

From the practical point of view, the described method is of great interest.

Using this method, good results may be obtained without performing a large number

of trial and error experiments in order to develop a specific coating technology.

Furthermore, it is no longer necessary to have complete knowledge of all process

parameters. This model is most useful for cost reduction in industrial multi-parameter

processes. For example, the cost of a complete factor experiment comprises those of

the vacuum station, manpower, measurements, data processing and adaptation. Most

of these factors are rather time consuming. Reduction in the number of trial runs

evidently lowers the cost of R&D. Assume the laboratory cost to be $100/hour. A

complete factor experiment requires in the order of 24 = 16 trial runs. One trial run

takes 4 hours including processing. Therefore the full cost of preliminary work will be

USD 6,400. Halving the number of trial runs saves half of the cost (USD 3,200).

74

3.1.2 Deposition of Ta resistive coatings with specified conductivity

3.1.2.1 Introduction.

Tantalum films were widely used as resistive coatings since the early sixties.

These films feature a negative temperature resistance coefficient (NTC),

superconductivity at low temperatures (4 K for the bulk tantalum and 40-50 K for the

tantalum oxide thin films), and a wide range of resistivity at low temperatures (10 to

1000 µΩ⋅cm at 0 to 30 % oxygen content)68. During the nineties, tantalum films were

successfully applied in microelectronic applications owing to their attractive physical

properties such as high dielectric constant (~20-40), transparency in the spectrum

range of 0.4 µm to 8 µm, high durability, corrosion resistance, and excellent

mechanical protection to the layers underneath69.

All of these advantages of tantalum oxide thin films make them usable for a

wide variety of electrical and mechanical applications. Tantalum oxide films were

applied in capacitors and in MOS transistors in dynamic random access memories

(DRAM)70, 71, in MOS memory film devices70, as photosensitive materials in Braggs

gratings for optical waveguides72, and as a protection layer in photosensitive device

systems. High dielectric constant (greater than 20) and high material strength make

the tantalum oxide films a potential candidate for replacing the silicon oxide films in

integrated circuits71. Replacement of SiO2 with Ta2O5 in VLSI capacitors results in a

six-fold increase of the capacity69, owing to its higher dielectric constant. Ta2O5 films

were applied as membranes in biomedical systems73 due to their biological

compatibility with blood and other biological matters.

The deposition techniques for tantalum oxide films and their conductivity

mechanisms were studied in several reports68,69,74. In these and some other

researches75-77 the tantalum oxide films were presented as a mixture of metallic Ta

75

and insulating Ta2O5. In these papers, a non-linear dependence between the

conductivity (σ) and the inverse temperature value (1/T) was obtained. One of the

models75 presents the electrical conductivity of the Ta + Ta2O5 system as a function of

the fractional volume of tantalum particles in the tantalum oxide matrix. However, the

experiments showed the conductivity to decrease much faster than predicted by the

model. Then the authors suggested an activation model of the process, i.e. σ ∝ exp(-

∆E/T), where ∆E is the activation energy. Therefore, the non-linearity was suggested

to be the result of a combination of metallic and tunneling conductivity. Another

proposed explanation implied the excess tantalum to form donor levels along the band

gap of Ta2O5.

Hall effect measurements showed the major carriers in the Ta+Ta2O5 system to

be of positive sign76. Using this result, F.C. Livermore et al.77 showed that the

electrical conductivity mechanism consists of both types of conductivity: metallic and

tunneling. The metallic contribution to the conductivity process acts in parallel with

the activation mechanism. To explain the conductivity mechanism, the authors of this

paper77 suggested a layer model of oxygendoped tantalum. In this model the Ta

layers varied in their oxygen content. It was shown that the two-layer model explains

the TCR sign inversion with respect to the temperature increase68. It was found that

the resistivity of both layers together could be represented by a linear function whose

slope was positive or negative in accordance with the prevailing layer.

Tantalum is a refractory (high-melting) metal. Its vapor pressure reaches 10-2

Torr at its melting point, 3000 °C. The tantalum thin films are usually formed by

sputtering methods78 or by electronbeam evaporation72. Resistive tantalum films are

usually solid solutions of (Ta + Ta2O5) or (Ta + Ta2N). The type of deposition usually

used for these films is reactive sputtering.

76

In our work we tried to use the above ideas for mathematical modeling of the

magnetron sputtering. Here, we used such modeling for the deposition of thin film

resistors consisting of tantalum and its oxides. A feasibility of depositing tantalum

films with the specified resistance values controlled by variation of the technological

parameters was successfully proved.

3.1.2.2 Experimental procedure.

The experiments were done using the same laboratory magnetron sputtering

vacuum station11 (see fig. 5). The glass microscope slides of 25*75 mm2 were used as

substrates in all experimental trials. Residual gas was used as an active oxidizing

reactive agent. By the end of the process the samples were thermally treated with the

help of a novel method known as Vacuum Photothermal Processing (VPP)79, for 2

minutes, at about 250-300°C, in a vacuum of 3-5*10-5 Torr.

The following five independent technological parameters were varied in the

deposition experiments:

- V, sputtering voltage: 1.5 to 2.5 kV;

- PAr, argon pressure: 70 to 300 mTorr;

- Pres, residual air pressure: 2.5*10-5 to 7.5*10-5 Torr;

- Ts, the substrate temperature: 200 °C;

- t, deposition duration: 1.5 to 2.5 hours.

A standard four-point probe method was used for the sheet resistance

measurement. Each sample was measured in several points, and the averaged results

were used for further estimations. The surface structure of the obtained films was

studied on a computerized metallurgical microscope with optical magnification up to

×1600. Film thickness was measured on the homemade computerized micro-

77

interferometer80. Figure 47 shows an optical schematic diagram of the micro-

interferometer.

Fig. 47. Contactless thickness measurement scheme.

A typical photograph of a surface prepared for the thickness measurement is shown in

figure 48. The samples were prepared by scratching the substrate with the thin film as

deposited and adjustment in the micro-interferometer.

Fig. 48. Typical sample prepared for thickness measurement.

The obtained pictures were evaluated with suitable standard software. The thickness

was estimated by the bending of interference lines. The estimation error did not

78

exceed 2.5%. The measurement precision was approximately 20 A°, which is no

worse than the accuracy of a standard Tenkor Instruments profilometer (100 A°).

3.1.2.3 Practical model implementation.

A magnetron sputtering process can be described in the form of the equation F

= f(x1,x2,..xi,..xn), where xi are the process parameters. The process analysis enables

us to use only several parameters, while taking other parameters as insignificant. For

the first level approximation, we chose the following equation

F = f (Pres, PAr, V, Ts, t), (95)

where Pres is the residual pressure in the vacuum chamber, PAr the working argon

pressure, V the sputtering voltage, Ts the substrate temperature, and t the

deposition duration.

In order to provide complete evaporation of the air and water molecules from the

substrate, we chose Ts = 200 °C for all the experimental trials. This temperature was

not essential for the metal film growth81, but enabled better adhesion to the substrate.

The deposition duration was two hours. With sufficiently thick tantalum layers, we

can neglect the initial stage of film growth. The long deposition duration also allows

us to neglect the influence of the residual pressure, due to a relatively high pressure of

argon gas in the course of the 2-hour process.

Therefore, for such a deposition process the following approximation model can be

applied:

σ = f(PAr, V) (96)

Here the conductivity σ represents a response function of two variables. Suppose that

the response function is analytical, so that it can be expanded in a Taylor series82. The

first three terms represent a classical second-degree equation. Therefore, the resulting

equation can be written as:

79

σ = a1 + a2PAr + a3V + a4PArV + a5PAr2 + a6V2 (97)

3.1.2.4 Experimental results.

Table 10 presents the variable deposition parameters and the measurement results. As

seen, the parameters were varied in a random way. Therefore, each experimental trial

represents a partial section of the chosen multi-parameter space82, and the required

model can be built with the missing coefficients.

Table 10. Variable deposition parameters for Ta films.

N Pres, Torr PAr, mTorr V, kV t, hours d, µm RZ, Ω σ, Sim 1 4.5*10-5 300 1.5 2.0 0.08 ∞ 0.01 2 4.5*10-5 70 1.5 2.0 0.45 9.9 2244.67 3 4.5*10-5 70 2.5 2.0 3.00 3.1 1075.27 4 4.5*10-5 300 2.5 2.0 0.80 37.7 331.56 5 4.5*10-5 300 2.3 2.0 0.20 20900 2.39 6 4.5*10-5 300 2.0 2.0 0.74 ∞ 0.01

1-1 5.5*10-5 110 2.0 2.0 1.20 16.0 520.83 1-2 5.5*10-5 110 2.0 2.0 1.10 18.8 483.56 2-1 2.3*10-5 110 2.0 2.0 0.7 24.7 578.37 3-1 5.5*10-5 90 2.0 2.0 0.42 19.3 1233.65 4-1 5.5*10-5 110 1.5 2.0 0.40 47.1 530.79 5-1 7.5*10-5 130 1.5 2.5 0.55 46.3 392.70 6-1 7.5*10-5 90 2.5 1.5 2.15 7.2 645.99

Table 10 can be processed into Table 11, thus representing the tantalum film

conductivity as a function of two variables: the argon pressure and the sputtering

voltage.

Table 11. Tantalum conductivity as function of two variables.

V \ PAr 70 90 110 130 300 1.5 2244.67 530.79 392.7 0.01 2.0 1233.65 520.83 0.01 2.3 2.39 2.5 1075.27 645.99 331.56

The tantalum films shown in figure 49 were deposited with different argon pressures.

It can be observed that the surface roughness of the obtained films increases with

80

higher pressures. The surface roughness usually reflects the presence of spot defects

or second phase inclusions. It is reasonable to assume that these defects are crystal

inclusions of the tantalum oxide film in the metal tantalum matrix. Therefore, when

the number of these inclusions (spot defects) exceeds the defined threshold level, the

activation type of conductivity become prevailing over the metallic type.

(a) (b)

© (d)

Fig. 49. Tantalum thin films photographs (X 160):

a Ar pressure of 90 mTorr, b Ar pressure of 110 mTorr,

c Ar pressure of 130 mTorr, d Ar pressure of 300 mTorr

This reasoning originates from figure 50, which represents in a graphical form the

dependence of conductivity on the working pressure, for various sputtering voltages.

It is easily observed that the graphs σ = f (PAr) are similar to each other for various

sputtering voltages. As shown in these graphs, conductivity decreases with the

increase in the working pressure. The area on the graph with a sharp bending, in

81

which the tantalum electrical properties drastically change, is clearly observed. At this

threshold level, marking the change in the tantalum films composition structure, the

activation type of conductivity becomes prevailing over the metallic type.

Fig. 50. Conductivity of tantalum films as a function of two variables: argon pressure

and sputtering voltage.

As seen from table 11 and figure 50, the experimental data contain missing points. In

other words, a number of experimental trials were insufficient for the table 11 filling.

In our case, the missing points and the model coefficients can be calculated using a

polynomial interpolation method83. Let us consider, for example, the first row in table

8. The response function value at the point PAr = 90 mTorr is absent. As in the

Newton Divided-Difference Interpolation Polynomial method, the following formula

similar to the Taylors expansion can be applied:

σ (PAr) = f(x) = Pn(x) + Rn(x), (98)

where f(x) is the response function, Pn(x) is the interpolation polynomial and Rn(x) is

the approximation error. Pn(x) equals f(x) in the specified points: x0, x1,xn,

therefore, the values of x = PAr are the interpolation centers.

50 100 150 200 250 300

0

500

1000

1500

2000

2500

1.5 kV 2.0 kV 2.5 kV

Con

duct

ivity

, Sim

Argon pressure, mTorr

82

The 3rd degree interpolation polynomial, in our case, can be written in the form of:

Pn(x) = f[x0] + (x - x0)f[x1,x0] + (x - x0)(x - x1)f[x2,x1,x0] + (99)

+ (x - x0)(x - x1)(x - x2)f[x3,x2,x1,x0]

where

This model can be re-written in the Lagrange form, for higher levels of interpolation.

Thus, the coefficients of the process model can be calculated from the following

formula:

Estimation of the required coefficients gives the following specific model:

P3(x) = -2.55*10-3x3 + 1.39x2 250.81x + 14216 (102)

Figure 51 represents this P3(x) ≈ σ(PAr) for the sputtering voltage of V = 1.5 kV.

( ) (101) )xx(

)xx()x(f

xPn

0in

ij0j

ji

n

ij0j

ji

n ∑∏

∏

=

≠=

≠=

−

−

=

(100) )xx(

)x(f)xx()xx(

)x(f...

)xx()xx()x(f

)xx()xx()x(f

)x,...,x,x(f

n

0in

ij0j

ji

i

0n1nn

0

0n1nn

1n

0n1nn

n01nn

∑∏=

≠=

−

−

−

−−

−=

−⋅⋅⋅−++

+−⋅⋅⋅−

+−⋅⋅⋅−

=

83

50 100 150 200 250 300

0

500

1000

1500

2000

2500

Data calculated by interpolationfor x = 90: y = 1044

Sputtering voltage V = 1.5 kV

Con

duci

vity

, Sim

Argon pressure, mTorr

Fig. 51. Tantalum film conductivity, calculated value.

As shown, the estimated point fits the experimental graph without disfiguring it.

Generalizing the interpolation model in 3-D84 using the Lagrange form yields the

response surface shown in figure 52.

Fig. 52. 3-D presentation of tantalum films conductivity as a function of technological

parameters.

84

This graph enables the deposition of tantalum films with pre-defined conductivity in

the chosen areas of the technological parameter variations. Thus, we have

demonstrated the possibility to successfully apply the proposed engineering method to

experiment planning and development of various deposition technologies.

85

3.2 Vacuum Photothermal Processing (VPP).

3.2.1 Introduction

Sputtering processes modify in various ways the surfaces with deposited films

and the growing thin film itself. Such post-deposition methods as annealing relax

the obtained films and stabilize their physical properties. Since the post-deposition

treatment may be considered as an additional degree of freedom for the multi-

parameter space of the sputtering deposition process, the development and

investigation of the this kind of treatment becomes a significant part in fine tuning of

the sputtering process.

The ever-growing development of sub-micron devices such as ULSI, micro-

machines, multi-quantum well solar cells, etc., demands a continuous reduction in the

quantity of macroscopic and microscopic defects during the manufacturing process.

These defects are caused by various atomic mechanisms. For example, ion

implantation leads to distortion and damages of the crystal lattice, so that by the end

of the implantation process it is necessary to re-locate the implanted ions and

displaced atoms onto the active lattice sites. This relocation is done in what is called

a post-implantation annealing process.

The traditional annealing process, a furnace process, is carried out by

heating the silicon wafers in a furnace at ~1000 °C for up to a few hours. Not only is

this process time-consuming, it is inappropriate for complex semiconductors such as

III-V or II-VI compounds, which have comparatively low dissociation temperatures

and high vapor pressures characteristic of V or VI elements. The furnace processing

had been used for a long time and was included into the production line of the

semiconductor manufacturing industry. However, modern devices require low

86

temperature processing, low thermal and residual stress, low defect concentration, and

faster processes with high yields and throughputs.

The fundamental disadvantages of furnace processes had been overcome with

the introduction of Rapid Thermal Annealing (RTA). The first RTA process was

demonstrated in 1975 by using pulsed lasers to anneal ion implanted GaAs devices85.

Many different types of RTA methods for transient annealing were proposed since

then, although the first industrial RTA application was described in 1980 86,87. All of

these methods were based on the application of different types of heating lamps or

lasers on the annealed substrate. Further we demonstrate the experimental results of a

novel approach called Vacuum Photothermal Processing (VPP) based on fast direct

irradiation of thin films in vacuum. This approach is based on the RTA process and

relates the changes in the electric properties of the semiconductor films to photonic

rather than to thermal effects.

3.2.2 Development and investigation of VPP treatment.

3.2.2.1 Basic Principles of Rapid Thermal Processing.

The basic idea behind the RTA is the exposure of semiconductor samples to a

radiation flux generated by arc or halogen lamps. A schematic diagram of this

technique is shown in figure 53. One or two banks of halogen lamps above and below

irradiate the wafer while providing nitrogen atmosphere in the reaction chamber in

order to stabilize the vapor pressure88.

Fig. 53. Typical Rapid Thermal Annealing (RTA) [88].

87

The power to the water-cooled lamps is computer controlled via a negative feedback

loop. In the old RTA configuration, a thermocouple on a calibration wafer was used

as a temperature sensor. Most of the modern RTA systems use a pyrometer for direct

measurement of the heated wafer temperature. A typical RTA processing period for

implants in GaAs is 2 to 10 seconds in the temperature range of 800 to 1050° C.

Modifications of this technique may include several numbers of lamp banks and

several numbers of lamps in the bank. Other arrangements of the lamps structure may

also affect the final properties of the treated samples89.

The Rapid Thermal Processing (RTP) is based on incoherent radiation from

tungsten or halogen lamps90. This procedure combines the annealing process with

photonic influence. The furnace processes are caused by photons of infra-red range

with wavelengths longer than 0.8 µm. Irradiation of samples with short wavelength

(visible and UV ranges) photons permits to decrease the treatment period as well as

the annealing temperature of samples. In the case of the RTP the short wavelength

photons influence becomes more significant than heating waves. According to the

Plank radiation law for black bodies16, the maximum of the spectral radiation

distribution shifts to the shorter wavelengths part of the spectrum for higher radiation

intensities. The tungsten filament in the halogen lamps is heated up to about 3000° K

so that most of the RTP treatment is done by the shorter wavelength photons, that is,

those with energy ranging from the ultraviolet to the visible range. Obviously, the

influence mechanism of long wavelength photons on semiconductors differs from that

of the short wavelength photons. The authors of the fundamental paper91 showed that

owing to quantum effects, the incoherent light results in a short RTP impact, which

brings the reduction in the activation energy and the surface roughness, and the

improvement of the annealing uniformity.

88

Halogen and deuterium lamps when used as sources for incoherent light

decrease the influence of thermal effects and increase the role of the quantum effects

on the sample being processed. It was also shown91 that application of the deuterium

lamp only (vacuum UV source) with intensity of 10 mW/cm2 did not increase the

wafer temperature (purely quantum effects). In addition, the RTP leads to cardinal

transformations in the structure and type of contacts in the metal-semiconductor

system of Ti-Al-GaN. It was shown92 that the RTP in Ar atmosphere changes the

contact type from non-linear to ohmic with a specific contact resistance of

approximately 10-2 Ω⋅cm2, owing to reaction between Al and Ti.

3.2.2.2 Experimental procedure.

Figure 54 presents a schematic diagram of our experimental setup. A standard

tungsten coil evaporator93 was used as a non-coherent light source. This tungsten coil

was made of four braided pure W wires, each of 1 mm in diameter and ~ 37.5 cm

long. The treated samples were placed 30 mm away from the coil.

Fig. 54. Schematic diagram of Vacuum Photothermal Processing setup.

During the VPP the temperature of the samples was measured by a

thermocouple (K-type - Cromega-Alomega). The current in the tungsten coil was

measured with an ACA Leakage Tester of DL-6054 type. The voltage drop on the

tungsten coil was measured as well. These data were used for estimation of the energy

source temperature. The composition of the deposited films was evaluated using the

89

Energy Dispersive Spectrometry (EDS) add-on mounted on the STEREOSCAN-430

(LEICA Scanning Electron Microscope operating on 20 keV). The film thickness was

measured with the automatic Tencor Instruments profilometer. The surface views and

structure of samples were studied on the Canon microscope with magnification up to

x1600. The fine surface roughness of the samples as deposited and after the VPP

treatment was measured with the Digital Instruments DimensionTM 3100 Scanning

Probe Microscope.

The I-V characteristics of the thin film systems prior to and after the VPP

processing were compared. In addition, the I-V characteristics of the VPP treated

samples were compared to those of the control samples. The characteristics were

obtained with the help of the Keithley bench station. Figure 55 illustrates the

measuring system.

Fig. 55. Schematic diagram of the I-V measurement.

The potential barrier between the metal electrode and the silicon was calculated from

the I-V characteristics 18, 94. The Shockley equation for the saturation current density

is:

(103)

kTq

expTAJ B2rS

Φ−=

90

where Ar is the effective Richardson constant, q the elementary electron charge, k

the Boltzmann constant, T the ambient temperature, and ΦB the potential barrier.

Equation (103) was then transformed to express the potential as:

(104) JTA

logq

kT3.2JTA

lnq

kT

S

2r

S

2r

B ==Φ

The Richardson constant was assumed to be 120 A⋅cm-2K-2. In order to obtain the JS

values, the experimental I-V characteristics were built in semi-logarithmic

coordinates. These values were obtained from the estimation of the intersection points

of the ordinate axis with the tangent to the I-V curve.

3.2.2.3 Sample preparation.

The VPP experiments were carried out with the following samples (deposited

on the glass and silicon substrates):

• glass-Al-Al2O3-Ge-Au;

• glass-In2O3;

• Au-Si-Au;

• Au-Si-Ti;

• Au-Si-Ni.

Figure 56 presents a typical thin film structure of a measured sample on glass.

This structure consists of crosswise located electrodes with semiconductor films

between them. The active surface of the metal-dielectric-semiconductor-metal

(MDSM) structure was 4 mm2 large.

Fig. 56. Typical MDSM structure for measuring I-V characteristics.

91

The samples with transparent conductive coating of In2O3 on the glass substrates were

of 25×75 mm2 large. The measured metal-semiconductor-metal (MSM) samples were

prepared as follows: P-type silicon wafers [111] were cut to pieces (samples) of about

10*20 mm2. Each sample contained more than 15 contact structures. The upper

electrodes were 1, 1.5 or 2 mm in diameter. Figure 57 presents a typical thin film

structure of a measured MSM sample.

Fig. 57. Schematic upper and front view of the treated MSM samples.

Prior to thin film deposition, all samples had been cleaned with a proper

chemical procedure. First, the silicon samples were cleaned in a buffer solution

(HF:NH4F = 1:7) for 1 min at 25° C in order to remove the native silicon oxide from

the sample surface. Then the samples were rinsed in soft water. After cleaning the

samples with trichlorethylene and an additional rinse in deionized water, the samples

were ready to the pre-treatment. The samples were then washed in the RCA-2 solution

(H2O:H2O2:HCl = 20:1:1) for 20 min. at 50° C and then in the ultra-sonic cleaning

bath. The final rinse was done with isoprophanol. After drying in air, the backsides of

all of the samples were coated with thin film Au electrodes. Then, the front sides of

six samples were coated with Ti electrodes and those of the other six were coated with

Ni electrodes. The Au-Si-Au control systems were also prepared, for comparison of

the electrical characteristics. Glass substrates were degreased in the trichlorethylene in

ultrasound at 30° C for 5 min and rinsed in the iso-prophanol.

92

All metal films were deposited on the laboratory evaporation vacuum setup

with the vacuum of approximately 3*10-5 Torr. The substrate temperature during

evaporation was ~200° C. A thin film of Al2O3 is a natural oxide that grows on the

lower electrode while exposing the vacuum chamber to air. The germanium thin films

and semiconductor indium oxide thin films were grown by sputtering using a

laboratory magnetron vacuum station95, 96. The grown Ge films were about 1.5 µm

thick. The thickness of the In2O3 transparent conductive thin films was found to be

about 250 nm82.

3.2.2.4 Control experiments.

It was found that the temperature of the samples during the VPP raises up to

about 400° C (see figure 58).

Fig. 58. Typical Vacuum Photothermal Processing.

Our goal in this work was to verify the type of the treatment mechanism. There are

several factors of influence on the VPP: ultra-violet photons, thermal effects, and

electrons emitted from the heated tungsten wire. The examination was done in three

ways. In the first control experiment, the similar samples were heated up by a hot

93

plate to 400° C for the same period of time as in the VPP experiment, and the

electrical properties of these samples were compared to those of samples treated with

the VPP. In the second control experiment, a glass with absorption edge at 380 nm

was introduced between the tungsten coil and the samples. This glass blocked the UV

light and emitted the electron flux reaching the sample; therefore the samples were

exposed only to thermal effects and to photons in the visible range. Like in the first

control experiment, the electrical properties of these samples were compared to those

of the VPP treated samples. Figure 59 presents a schematic diagram of the third

control experiment. A thin (0.2 mm diameter) tungsten filament was placed between

the energy source and the substrate holder. This filament was connected to a power

supply, which provided a positive or negative voltage with respect to the light source.

The voltage supplied to the coils during the VPP treatment was -5, -1, 0, +1, or +5

Volt.

Fig. 59. Schematic diagram of VPP treatment setup with control electrode.

The VPP treatment was conducted under the same conditions for all the trials.

In the third control experiment, the first sample was made of Au-Si-Ti and the second

one of Au-Si-Ni. The treatment was done for 90 seconds at a vacuum of ~ 2*10-5

Torr with the maximum allowable current (75 A) through the tungsten coils. In order

to isolate the influencing parameters, only one parameter was changed from one trial

94

to another. The voltage on the control electrode was changed from - 5 V up to 5V,

while the residual pressure and the process duration remained the same. These

parameters are shown in table 12.

Table 12. Parameters of experimental trials.

NN Variable parameter Constant parameters V bias, V Pressure, Torr VPP duration, sec 1 - 5 3*10-5 90 2 - 1 3*10-5 90 3 0 3*10-5 90 4 + 1 3*10-5 90 5 + 5 3*10-5 90

3.2.2.5 Experimental results.

The temperature of the energy source is highly important for the VPP process.

It is well known that tungsten coils can be heated up to ~3000° K in vacuum for long

periods97, since the melting point of W is ~ 3683 K and the boiling point is ~ 6200

K98. Figure 60 illustrates the tungsten steam pressure for high temperatures2.

Fig. 60. Tungsten steam pressure2.

One can see that tungsten practically does not evaporate up to its melting point (the

steam pressure is less than 10-2 Torr). The tungsten coil temperature was estimated

95

with the help of the current and the voltage drop values. A typical time dependence of

the source temperature is shown in figure 61. The average measured values of the

current intensity and the voltage drop were I = 52.4 A and U = 6.61 V, respectively.

Estimation of the tungsten coil temperature gave the value of ~ 3363 K on its surface.

Fig. 61. Tungsten coil temperature during VPP.

In our calculations, we used the standard data on the temperature dependence of

tungsten resistivity99. These data together with the interpolation formula are shown in

figure 62.

Fig. 62. Tungsten resistivity dependence on temperature99.

96

The approximate quantity of photons irradiated from the annealed tungsten coil was

evaluated as the ratio of irradiation intensity (Iλ) and energy of photons (Eλ). The

irradiation intensity, in its turn, was calculated in accordance with the known Planck

relationship100:

Here h is the Planck constant, c is the velocity of light, and λ the wavelength. The

estimated data is presented in figure 63.

Fig. 63. Tungsten spectral efficiency at different temperatures.

The comparison of the tungsten coil to the known calibrated sources, such as, for

example, that of McPherson Inc.101, shows our process to be simpler than a standard

lamp application, while providing higher power (~ 350 W). The tungsten Halogen

light source of the above company is a high-energy unit with a 100 W lamp. This

source produces continuous spectrum of radiation in the range of 400 nm to 2

microns. Therefore, one can say that when using tungsten coils free of glass jacket,

(105) 1e

1hc2kT/hc5

2

−λπ

=ℜ λ

97

the process may become more efficient, as radiation with higher intensities then takes

place. There are, of course, various special light sources for vacuum ultraviolet and

ultraviolet, such as deuterium lamps. These sources provide a continuous spectrum of

radiation in the wavelength range from 0.115 до 0.37 microns. However, in this case

the thermal effects are completely excluded. Figure 64 shows the thermo-emission

current density in a heated W wire.

Fig. 64. Thermo-emission current density of Tungsten.

These data may be used for the electron flux estimation. Also, these data become

significant under the VPP conditions, that is, under irradiation without glass jacket in

vacuum.

2000 2200 2400 2600 2800 3000-2

0

2

4

6

8

10

12

14

16

Ther

moe

mis

sion

cur

rent

den

sity

, A/c

m2

Temperature, K

98

3.2.2.6 Mechanical properties of the films

Figure 65 presents the microscopic photography of a multiplayer glass-Al-

Al2O3-Ge-Au system including the lower (Al) electrode, a semiconductor Ge film and

the upper (Au) electrode. The germanium thin film surface obtained by sputtering

Fig. 65. Microscopic view of the Al-Al2O3-Ge-Au system (× 80).

represents a very non-homogeneous surface as shown in figure 66 96. This nano-

crystalline structure was obtained with the following technological parameters:

• Residual pressure ~ 3×10-5 Torr

• Argon pressure ~ 130 mTorr

• Sputtering DC voltage - 2 kV;

• Substrate temperature ~ 200° C;

Fig. 66. 3D presentation of Ge thin film surface (AFM microscope).

99

This surface was found to have many surface states, which essentially influenced the

electron transport mobility in the studied system. A rapid thermal processing step has

been shown to be a cure mechanism to produce precise structural phase

transformations, including carrier traps removal 102,103.

The three main parameters that most influence the thin film system properties

during thermal processing are the film material and thickness, and the thermal

treatment duration. The vacuum photo-thermal processing was studied for various

processing intervals, materials and film thicknesses. The treatment period was chosen

according to the known data2, ranging from 15 to 90 seconds. The influence of the

thermal processes becomes significant after about 120 seconds of treatment due to the

temperature increase of the sample, resulting in higher diffusion rate. For

temperatures higher than 400°C, the Au atoms penetrate the semiconductor system

and the entire structure is destroyed. Figure 67 presents comparative external views of

some samples treated by the VPP and by the conventional furnace annealing

treatment.

Fig. 67. External view of the processed MDSM samples.

100

All samples designated 2b were heated up to 400°C and were destroyed in a real-time

measurement due to the different thermal expansion/contraction coefficients of the

semiconductor layers and the glass substrate. The thin films were wrecked and part of

Au atoms were diffused into the Ge layer. On the other hand, all of the VPP treated

samples (1b) kept their shape and form.

A view of the MSM samples before and after the VPP treatment is presented

in figures 68 and 69. These figures demonstrate the significant changes in the

electrode surfaces. Figures 68a and 68b show the Ni electrodes with magnification

×150 and ×3000, respectively.

(a) (b)

© (d)

Fig. 68. Experimental photographs of Ni electrodes on silicon for as deposited and

VPP treated samples, in two magnifications:

(a) as deposited film, *150; (b) as deposited film, *3000;

© VPP treated film, *150; (d) VPP treated film, *3000.

101

The same Ni electrode after VPP treatment is shown in the same magnification in

figure 68c and figure 68d, respectively. Similar photographs for Ti electrodes are

presented in figure 69.

(a) (b)

© (d)

Fig. 69. Experimental photographs of Ti electrodes on silicon for as deposited and

VPP treated samples for two different magnifications:

(a) as deposited film, *150; (b) as deposited film, *3000;

© VPP treated film, *150; (d) VPP treated film, *3000.

It is clearly demonstrated that the surface homogeneity of both Ni and Ti films

increases after the VPP treatment. The metal electrode films cover all the defects in

the crystal plane on top of the silicon surface. Following the VPP treatment, these

defects disappeared and the global number of surface defects was dramatically

decreased. K.P. Lee et al. examined the effect of UV illumination on the surface

102

morphology of dry etched Ta2O5 by atomic force microscopy method104. They have

shown that there was no significant difference in root mean square surface roughness

between the two cases: with UV irradiation and without it, for different chemistry

used. In our experiments, however, the significant metamorphoses of the treated

surfaces were found. Figure 70 presents the sample temperature versus processing

time for three types of treatments: filtered VPP, hot plate, and full VPP.

0 20 40 60 80 100

50

100

150

200

250

300

350

400

450

Filtered VPP Furnace VPP processing

Tem

pera

ture

, OC

Time of processing, sec

Fig. 70. Sample temperatures vs. processing time for various thermal treatments.

We see that after 40-50 sec of processing the temperature is approximately equal in all

three cases. In the same time the influence on the electrical properties is significantly

higher in the case of the full VPP. The longer duration of the hot plate treatment leads

to a fatal destruction of the semiconductor film and of the measuring system located

on its surface. Also note that unlike the RTP that uses tungsten, halogen or deuterium

lamps as heat sources, in our case there is no jacket to the irradiating lamps.

Therefore, more ultra violet (UV) photons together with the emitted electron flux are

involved with the annealing process.

103

The 3D atomic force microscope (AFM) pictures of these treated Ti films are

presented in figure 71. Comparison between figures 71a and 71b emphasizes the

smoothing effect of the VPP treatment. The second picture presents the flat plane

surface of the metal electrode. Following the VPP treatment the metallic grains

became lower and the roughness decreased. The AFM section analysis of the

roughness is shown in figure 72. It can be seen that the mean roughness on the VPP

treated samples decreases to less than a half and the maximum surface grain height

decreases from 6.67 nm to 1.83 nm.

(a) (b)

Fig. 71. 3D AFM photographs of as deposited and VPP treated Ti films.

(a) as deposited Au-Si-Ti system; (b) VPP treated Au-Si-Ti system.

(a) (b)

Fig. 72. AFM section analysis of roughness for as deposited and VPP treated Au-Si-

Ti systems.

(a) (a) as deposited Au-Si-Ti system; (b) VPP treated Au-Si-Ti system.

104

This effect is conditioned by the combined influence of two fluxes: the electron beam

and the photon irradiation emitted from the tungsten coil in vacuum. Obviously, this

effect is impossible without one of the two components. For example, the electron

beam was absent in the paper by K.P. Lee et al.104 since all the samples in their study

were DC biased.

3.2.2.7 Electrical properties of the films.

In this section we demonstrate the VPP influence on the electrical properties

of different thin film systems for all types of experiment trials. It was found that in the

MDSM Al-Al2O3-Ge-Au system, a rectifying barrier is generated, with properties that

can be affected by the VPP process.

Figure 73 presents a typical I-V characteristic of the glass-Al-Al2O3-Ge-Au

system (sample 2a) measured after the deposition and after the hot plate control

experiment. It is easily seen that after a brief treatment there is no significant change

in the electrical properties of the thin film system.

Fig. 73. Electrical properties of the glass-Al-Al2O3-Ge-Au system before and after

traditional furnace annealing treatment.

Similar measurements were made before and after the VPP process with a glass filter

placed between the W irradiation source and the samples. These measurements are

105

shown in figure 74. The most significant modifications in the electrical properties (as

compared to the hot plate control experiment) are mainly due to the added W coil

irradiation in the visible range.

Fig. 74. VPP with glass filter (sample 1b).

Figure 75 shows the I-V characteristics before and after a VPP process. The

experiment was done on several typical glass-Al-Al2O3-Ge-Au samples (1A). All

samples were grown under the same conditions, and the resulting Ge layer was about

~1.5 µm thick for all the samples. We conclude from these results that the factors

responsible for essential modification in the film properties are shorter wavelength

irradiation together with the electron flux emitted by the W coil during the VPP.

Fig. 75. VPP processing of a glass-Al-Al2O3-Ge-Au system.

106

In this figure, the VPP results in an increase of the thin film resistance and of

the built-in potential barrier height. This phenomenon can be explained by excitation

of free bonds in the boundary between the metal and the nano-crystalline germanium

layers owing to the absorbed short-wavelength photons. On the other hand, the

excited surface states are saturated, and the potential barrier increases. In other words,

the number of free charge carriers in the semiconductor and in the interface between

the metal electrode and the semiconductor film decreases. Furthermore, after the VPP,

the breakdown points were cured and the breakdown voltage increased, which also

resulted in reduced number of free charge carriers.

As shown in figures 76 and 77, the VPP reduces the sheet resistance value in

the glass-In2O3 systems, while improving its uniformity. This effect can be explained

by the redistribution of the partially oxidized indium atoms in the oxide film and the

partial reduction of indium oxide at the expense of excited electrons under the ultra-

violet photon impact. Therefore, the VPP is found to improve the quality of

transparent conductive layers.

The sheet resistance measurements were carried out on the surfaces of the

samples by the four-point method. These measurements have shown a significant

decrease in the sheet resistance of the film, from about 1,000 Ω/Sq to an average

value of 120 Ω/Sq. The uniformity of the sheet resistance across the sample was

improved as well.

107

Fig. 76. Sheet resistance distribution in the glass-In2O3 system.

Fig. 77. Sheet resistance distribution in the VPP processed glass-In2O3 system.

108

3.2.3 VPP influence on the potential barrier height

A special attention was devoted to the MSM Au-Si-Metal structures,

representing the rectifying diodes. The I-V characteristics of the samples were taken

with the applied voltage in the range of -10 to +10 V. Figure 78 presents the

symmetrical I-V characteristics of the as deposited (non-annealed) and the VPP

treated Au-Si-Au systems.

Fig. 78. I-V characteristics of as deposited and VPP treated Au-Si-Au systems.

This figure shows the reduction in the built-in potential barrier in the Au-Si-Au

system after the VPP treatment. The built-in potential barrier value was estimated

according to expression (104). The obtained values for the same system were ΦB =

0.384 eV for as deposited Au-Si system and ΦB = 0.341 eV for the VPP treated films.

These values are similar to the data presented in ref. [105], which confirms the

estimation method.

109

Figure 79 presents the I-V characteristics of the Au-Si-Ti system. This data

was obtained for a VPP treatment in identical samples with different bias on the

control electrode (according to table 12). The I-V characteristics clearly show a

serious influence of the electron beam emitted from the hot tungsten coil.

Fig. 79. I-V characteristics of as deposited and VPP treated Au-Si-Ti systems with

bias voltage on the control electrode.

Similar characteristics for the Au-Si-Ni systems are presented in figure 80.

Fig. 80. I-V characteristics of as deposited and VPP treated Au-Si-Ni systems with

bias voltage on the control electrode.

110

In all these figures, both with and without VPP, the non-symmetrical contact

properties are well presented. These figures enable us to evaluate the direct influence

of the electron beam and to separate it from the non-coherent irradiation of the

filament. The magnitude of the built-in potential (rectifying barrier height) was

estimated for each experiment using the equation (104). Figure 81 presents the

evaluation results of the built-in potential barrier vs. the bias voltage for the control

electrode.

Figure 81. Influence of biased VPP treatment on potential barrier values

in Au-Si-Metal systems.

The influence of the control bias voltage on the potential barrier value is clearly

demonstrated. The trend of potential barrier variation is similar for both contact

systems: Si-Ti and Si-Ni. Furthermore, it was found that the VPP also affects the

potential barrier value due to simultaneous action of the electron beam emitted from

the tungsten coil and the non-coherent irradiation. A positive bias causes reduction of

the rectifying barrier. A negative bias on the control electrode also changes the

111

potential barrier value. However, there are two factors that define this modification:

the primary state of metal-silicon interface and the metal nature.

All our experiments were carried out in realistic conditions. The silicon

samples were cleaned as described above and did not come from an ideal

environment. Prior to the metal layer deposition, all samples came in contact with air.

The oxides formed at the surface and the oxygen remnants remain no matter how

carefully the semiconductor is treated afterwards. Therefore, the interface traps

(between metal and silicon) of various kinds have to be formed and are formed in

reality because the metal layers are not lattice matched to the silicon106. We are rather

in doubt whether this can be a universal explanation of these complicated systems.

We tried to describe the obtained results in a pure qualitative way and we believe our

explanation to be confirmed by the calculations and additional measurements of the

following experiments.

As shown in figure 81, control of the electron beam enables variations in the

potential barrier height in a wide range of values. The potential barriers in Ti-Si and

Ni-Si were changed from 0.605 eV to 0.785 eV and from 0.571 eV to 0.672 eV,

respectively. This result may be explained as follows. A positive voltage on the

control electrode intercepts part of electron beam from the tungsten coil during the

VPP. The work function of tungsten equals to 4.54 eV 107. If the voltage on the

control electrode is +1 V, only a small part of the electron beam emitted from the W

coil will be intercepted. In this case we can see the decrease in the potential barrier

value at the expense of the structure ordering and the dangled bonds filling. Low

negative voltage acts similarly on the treated samples; however, the electron beam

here is confined. In the case of a high positive voltage (that is, a beam of electrons

with energy of more than 9.5 eV) only high-energy electrons can arrive at the sample.

112

These electrons may penetrate through the metal electrode and result in a significant

decrease in the potential barrier. This reduction may be explained by the interface

ordering, that is, ordering of both the silicon surface defects and the metal growing

defects. As shown in figures 46 47 and 49 - 50, the roughness of samples decreased

with the controlled VPP treatment. On the contrary, using UV irradiation without

electron beam has no significant effect on the roughness of treated surfaces104. High

negative voltage acts similarly on the various metals, however, in a reverse trend to

the positive voltage. We tried to explain this fact by taking into consideration the fine

metal structure. The metals that we used in our experiments (Ti, Ni) are referred to as

transition metals108, whose upper d-shell electrons should be considered as the valence

electrons109. Nickel and titanium have two electrons in the upper orbit. However, two

electrons are missing in the upper d-shells of nickel atoms and eight electrons in

those of titanium. This difference in the energetic structure of metals explains the

difference in behavior between the VPP treated Si-Ti and Si-Ni samples. Under a high

energetic electron current beam, the Si-Ti system first fills the electron states in the

unfilled electron orbit, and only then begins to fill the external orbit. Although the

system Si-Ni behaves similarly, the number of free electron states in this system is

less than in Si-Ti. The gold atom, on the contrary, has fully complete upper d-shells.

So, in the Si-Au system, the gold atom behaves oppositely to the described VPP

treatment mechanism. In this system the potential barrier value decreases following

the VPP from 0.384 eV to 0.341 eV.

It is of interest to note a similar behavior of Ti and Ni metals under the

influence of the light and electron beam irradiation during the VPP, as shown in

figure 81. We believe this behavior to be specified by the common metal properties of

these elements. The difference between them is determined by the different number of

113

electrons in the upper d-shell. We also suppose the similar behavior of other metals

with similar structures, which fact it will be of interest to study experimentally.

114

4. CONCLUSIONS AND FUTURE DIRECTIONS

4.1 Conclusions.

Vacuum deposition methods are very important to the up-to-date

microelectronics technology. Sputtering is one of the most useful methods for thin

film deposition. In the present work we have demonstrated a novel sputtering

deposition method and the apparatus for its implementation. The basic idea derives

from a mathematical approach describing a complex system as multi-parametric with

several levels of freedom. Each independent parameter should be considered as a

level of freedom. Such approach enables a fine control of the deposition process. In

our work, we have achieved the following results:

1. An artificial plane plasma discharge was obtained in a comparatively high

vacuum; a special construction for moving the plane plasma was implemented and

tested.

2. Electron temperature and ion concentration distribution in the plane plasma was

studied with the help of Langmuir probe measurement.

3. A novel multiparameter sputtering deposition system based on the plane plasma

discharge was developed.

4. The novel sputtering system enables the deposition of metal and semiconductor

thin films in high vacuum (0.2 - 5 mTorr) with high and fine-controllable

deposition rates (Si and Ti films were deposited and tested).

5. In this deposition system, the substrate is totally isolated from the plasma; the

substrate temperature is not affected by the plasma.

6. The target material sputters homogeneously over the entire target surface (in

contrast to the magnetron sputtering); an additional support electrode confines the

sputtering ions and decreases the loss of sputtered atoms.

115

7. A modeling method for the deposition processes based on the theory of

multiparameter systems was developed; the method of partial sections in a

multiparameter space was applied for decreasing the number of trial-and-error

attempts in the course of development of a specific deposition technology.

8. This modeling method was used for the development of practical sputtering

deposition technologies for specific materials, such as transparent conductive

In2O3 films and Ta resistors with predefined resistivity.

9. A novel method, the Vacuum Photo-Thermal Processing (VPP), for heat treatment

of deposited films, was developed. This method can be considered as an additional

level of freedom in the multiparameter system. The VPP consists of a

simultaneous irradiation of the treated sample in vacuum with electron flux and

non-coherent light, mainly from the UV and VUV spectrum.

10. The VPP provides for stabilizing the deposited coatings, improved homogeneity,

and curing the irreversible electrical breakdown in thin-film systems; it also

decreases the number of surface states and improves the roughness of surfaces and

interfaces between the semiconductor and the metal film coating.

11. A method for controlling the electron flux during VPP was developed. It was

found that the value of the electron flux affects the potential barrier height in the

rectifying diodes. The possibility of operating the potential barrier height in the

metal-silicon contact systems was studied.

12. It was found that the potential barriers for the pairs Si Ti and Si Ni behave in

the same way and the physical changes on the interface are identical in nature.

116

4.2 Future Directions.

In extension of our investigations in the field of the plane plasma discharge, it

seems important to study the electron temperature distribution more precisely when

implementing this plasma for reactive sputtering deposition. This application should

extend the DC sputtering for deposition of dielectric films and various semiconductor

films. We also suggest to use the proposed sputtering method for the case of complex

intermetallic compounds such as NiCr and YBa2CuO3. A high RF voltage applied to

the target support electrode pair will enable direct sputtering of dielectric materials.

Since the sputtering in our experiments was carried out in a comparatively high

vacuum (collisionless case), we believe that the deposited films will be of

stoichiometric composition. We believe that the novel sputtering method

demonstrated in our work can be used for numerous types of materials and

applications.

In this work we developed a novel type of thermal post-deposition treatment

(VPP). At present, only the effect of the VPP on the Ti and Ni has been studied. We

are also planning to study the VPP influence on other metals, as well as on the

interfaces of metals with a number of semiconductors, to begin with Ge and SiC.

Moreover, it seems highly important to investigate the specific nature of VPP

influence on the metal-semiconductor interface in order to find whether the VPP can

be used for repair of neutron-irradiated semiconductor devices. The physics and

chemistry of VPP and participation of hydroxyl groups, hydrogen, and water

molecules in these process also will be subject of future works.

117

REFERENCES

1. S Ismat Shah, Handbook of Thin Film Process technology, IOP Publishing Ltd.,

1995.

2. L.I. Maissel, R. Glang, Handbook of Thin Film Technology, McGraw-Hill,

N.Y., 1970.

3. A. Wynveen, Fan, J., Kakalios, J., Shinar, J., Studies of light soaking stability in

rf sputter-deposited a-Si:H,

http://cmp.ameslab.gov/cmp/Abstracts/shinar_21.html

4. M.A. Lieberman, A.J. Lichtenberg, Principles of plasma discharges and materials

processing, John Wiley & Sons, Inc, N.Y., 1994.

5. S. Minomura, Metallic modified material of intermetallic compound, Pat. USA

N 4238232,1980.

6. R. Kossowsky, Surface Modification Engineering, CRC Press, Boca Raton,

1989.

7. N. St. J. Braithwaite, Introduction to gas discharge, Plasma Source Sci. Technol.

9 (2000) 517 527.

8. T. Mousel, W. Eckstein, H. Gnaser, Energy spectra of sputtered species under

sub-keV ion bombardment: experiments and computer simulations, Nucl. Instr.

and Meth. B 152 36-48, 1999.

9. S.M. Rossnagel, Sputter deposition for semiconductor manufacturing, IBM

Journal of Research and Development, Vol.43,No ½, 1999, 163-180.

10. I.I. Beilis, M. Keidar, R.L. Boxman, and S. Goldsmith, Interelectrode plasma

parameters and plasma deposition in a hot refractory anode vacuum arc, Physics

of Plasmas, V. 7, n. 7, 2000, 3068 3076.

118

11. G. Golan, A. Axelevitch, Ring etching zones on magnetron sputtering targets,

Thin Solid Films, Vol.300, pp.72 - 77, 1997.

12. Shablov, V.I., Bercushkina, E.I., Sharova, Z.M., The triode sputtering equipment

for development and preparing of electron-gun and photo-electric devices,

Review of electron techniques, series 8 (139), (Russian), 1973.

13. A. Axelevitch, The method of production of signal plate for vidicon, Certificate

of invention USSR N 616,944; Intl. Cl. H01j 9/20, 1978.

14. Golan, G., Axelevitch, A., Rabinovitch, Bilayer structure of Pd-Al2O3 for UV

vidicon applications, Thin Solid Films, Vol. 283, pp.235 - 238, 1996.

15. E.G. Kalashnikov, Electricity, Science, Moscow, (Russian), 1977.

16. M. Born, Atomic Physics, Blackie and Son, London, 1963.

17. R.A. Powell, S.M. Rossnagel, Thin Films (PVD for Microelectronics: Sputter

Deposition Applied to Semiconductor Manufacturing), Vol.26, Academic Press,

San-Diego, 1999.

18. A. Bar-Lev, G. Golan, Semiconductors, The Open University of Israel, Tel-

Aviv, (Hebrew) 1996.

19. B.V. Alekseev, B.A. Kotelnikov, Plasma diagnosis by the probe method,

Energoatomizdat, Mosow, (Russian), 1988.

20. D.A. Frank-Kamenetsky, Lectures on the Plasma Physics, Atomizdat, Moscow,

(Russian), 1964.

21. R.J. Stokes, D.F. Evans, Fundamentals of Interfacial Engineering, Wiley-VCH,

N.Y., 1997.

22. S.V. Berezhnoj, C.B. Shin, U. Buddemeier, I. Kaganovich, Charged species

profiles in oxygen plasma, Appl. Phys. Lett., 77(6,7) 2000, 800 802.

119

23. C.M. Ferreira, J. Loureiro, Electron kinetics in atomic and molecular plasmas,

Plasma Sources Sci. Technol. 9 (2000) 528 540.

24. S.O. Kasap, Principles of electrical engineering materials and devices,

McGraw-Hill, Boston, 1997.

25. T. Smy, L. Tan, S.S. Winterton, S.K.Dew, M.J. Brett, Simulation of sputter

deposition at high pressures, J. Vac. Sci. Technol. A 15(6) 1997, 2847 2853.

26. C. Walsh, Measurement of the Magnetic Field of a Helmholtz Coil Pair,

http://sensor.phys.dal.ca/minilabs/helmholtz/helmholtz.htm

27. C. Christou, Z.H. Barber, Ionization of sputtered material in a planar magnetron

discharge, J. Vac.Sci. Technol. A 18(6), 2000, 2897 2907.

28. A. Möller, Probe measurements of Fluctuations and Transport in Reversed-Field

Pinches, Dissertation thesis, Royal Institute of Technology, Stockholm, 1998.

29. R.H. Huddlestone, S.L. Leonard, Plasma Diagnostic Techniques, Academic

Press, N.Y., 1965.

30. L.D. Landau, E.M. Lifshitz, Fundamental Physics, Vol. 2, Field Theory,

Science, Moscow, (Russian), 1988.

31. Course of Semiconductors, University of Kiel, Faculty of Engineering,

http://www.techfak.uni-kiel.de/matwis/amat/semi_en/

32. Fundamental of Chemical Vapor Deposition,

http://www.timedomaincvd.com/CVD_Fundamentals/plasmas/plasmaTOC.htm

33. D. Ramshaw, C.H. Chang, Ambipolar Diffusion in Multicomponent Plasmas,

Plasma Chem. Plasma Process. 11, 395 (1991).

34. E. Velmre and A. Udal, Influence of Excitonic Scattering on Charge Carrier

Ambipolar Diffusion in Silicon, http://www.essderc.org/papers-97/235.pdf

120

35. D. Streb, G. Klem, P. Kiesel, G.H. Dohler, Investigation of the fast ambipolar

diffusion in GaAs n-i-p and n-i-p-i structures,

http://www.tpl.physik.uni-erlangen/allgemein/diff.html

36. ECE 423: Gaseous Electronics and Plasmas,

http://nigelz.ece.uiuc.edu/classes/ece423/handouts.html

37. M. Ashfold, MSc Physics of Advanced Semiconductor Materials, Plasma

Diagnostics: Mass Spectroscopy,

http://www.chm.bris.ac.uk/~paulmay/misc/msc/msc3.htm

38. D. Wood, Optoelectronic Semiconductor Devices, Prentice Hall, N.Y., 1994.

39. The Helmholtz Coil, Physics Laboratory Manual, Part two, Department of

Physics, New Jersey Institute of Technology, University Heights,

http://physics.njit.edu/classes/physlab/laboratory121/lab210/lab210.htm

40. R.C. Schaefer, Helmholtz Coil Problem and Solution Mathcad Application

Files, http://www.mathsoft.com/appsindex.html

41. D.B. Montgomery, J. Terrell, Some useful information for the design of air-core

solenoids, 1961.

42. P. May, MSc Physics of Advanced Semiconductor Materials, Plasmas and

Plasma Processing, http://www.chm.bris.ac.uk/~paulmay/misc/msc/msc4.htm

43. J.P. Greene, G.E. Thomas, Nucl. Instr. And Meth. A (1997) Argon beam

sputtering deposition rates using Argon,

http://www.phy.anl.gov/targetlab/sputtabl.htm

44. Fine Beam Saddle Fied Ion Source, Information Sheet B, Atom Tech Ltd., Island

Farm Ave., West Molesey, Surrey KT8 2UZ, England.

45. Ion Flux to Surfaces: the Bohm Velocity,

http://www.timedomaincvd.com/CVD_Fundamentals/plasmas/ion_flux.html

121

46. Sputter Coating of SEM Samples by Ralf Reuschling,

http://www.boc.com/evt/products/tfd/emsys/ralf_reuschling.htm

47. MultiSimplex, Design of Experiments Optimization Strategies,

http://www.multisimplex.com/strategy1.htm

48. Resit Unal, Edwin B. Dean, Taguchi Approach to Design Optimization for

Quality and Cost: an Overview,

http://techreport.larc.nasa.gov/ltrs/91/conf-13-ispa-unal%2bdean.refer.html

49. E.B. Dean, Taguchi Methods, http://mijuno.larc.nasa.gov/dfc/tm.html

50. G. Box, W. Hunter, J. Hunter, Statistics for Experimenters: An Introduction to

Design, Data Analysis, and Model Building, N.Y., John-Wiley, 1978.

51. V.A. Voznesensky, Experiment planning statistical methods for technical and

economical investigations, Statistics, Moscow, (Russian), 1974.

52. L. Moshinsky, Epoxy resins and hardeners, Arcady-Press, Tel-Aviv, (Russian),

1995.

53. G. Golan, A. Axelevitch, E. Rabinovitch, A Linear Model Application for the

Design of Transparent Conductive In2O3 Coatings, Microelectronics Journal, 29,

689 - 694, 1998.

54. J.R. Bellingham, W.A. Phyllips and C.J. Adkins, Electrical and optical

properties of amorphous indium oxide, J. Phys.: Condens. Matter, 2, 1990, 6207

- 6221.

55. Y. Shigesato and D.C. Paine, A microstructural study of low resistivity tin

indium oxide prepared by D.C. magnetron sputtering, Thin Solid Films, 238,

1994, 44 - 50.

122

56. A.H. Khalid, A.A. Rezazadeh, Fabrication and characterization of transparent-

gate field effect transistors using indium tin oxide, IEE Proc.-Optoelectron., 143

(1), 1996, 7 -11.

57. A. Murkherjee, The deposition of transparent conductive oxide thin films onto

large sheets of glass, acrylic and polycarbonate, Vacuum, 39, 1989, 537 - 540.

58. T. Maruyama, T. Tago, Germanium- and silicon-doped indium oxide thin films

prepared by radio-frequency magnetron sputtering, Appl. Phys. Lett., 64 (11),

1994, 1395 - 1397.

59. S. Kasiviswanathan and G. Rangarajan, Direct current magnetron sputtered

In2O3 films as tunnel barriers, J. Appl. Phys., 75 (5), 1994, 2572 - 2577.

60. S. Naseem, M. Iqbal, K. Hussain, Optoelectrical and structural properties of

evaporated indium oxide thin films, Solar Energy materials and Solar Cells, 31,

1993, 155 - 162.

61. J.I. Jeong, J.H. Moon, J.H. Hong, J.S. Kang, and Y.P. Lee, Low resistivity

transparent In2O3 films prepared by reactive ion plating, Appl. Phys. Lett., 64,

10, 1994, 1215-1217.

62. S. Muranaka, H. Hirooka, Y. Bando, Influence of oxygen pressure on the

structure of reactively deposited indium oxide films, J. Mater. Chem., 3 (3),

1993, 237 - 240.

63. S. Mirzapour, S.M. Rozati, M.G. Takwale, B.R. Marathe and V.G. Bhide,

Dependence of structural and electrical properties of undoped spray-deposited

indium oxide thin films on deposition temperature, Materials Letters, 13, 1992,

275 - 278.

64. R.N. Joshi, V.P. Singh, J.C. McClure, Characteristics of indium tin oxide films

deposited by r.f. magnetron sputtering, Thin Solid Films, 257, 1995, 32 - 35.

123

65. G. Golan, E. Rabinovitch, A. Axelevitch, L. Klibanov, A. Seidman, N. Kroitoru,

Microhardness analysis of diamond like carbon films, Diamond Films and

Technology, 7, 4, 257 - 266, 1997.

66. E. Rabinovitch, A. Axelevitch, B. Sigalov and G. Golan, Microhardness

Measurements of Indentation Size Effects in Sputtered Thin Films, The 17th

Israel Vacuum Society Conference, Tel Aviv, P5, December 1997.

67. Handbook of Chemistry and Physics, 48th Edition, The Chemical Rubber Co.,

N.Y., 1967.

68. W. D. Westwood, N. Waterhouse, P.S. Wilcox, Tantalum Thin Films,

Academic Press, New York, 1975.

69. K. Chen, M. Nielsen, S. Soss, E. J. Rymaszewski, T.-M. Lu, and C.T. Wan,

Study of Tantalum Oxide Thin Film Capacitors on Metallized Polymer Sheets

for Advanced Packaging Applications, IEEE Transactions on Components,

Packaging, and Manufacturing Technology Part B: Advanced Packaging, May

1997, 20, 02, p. 117.

70. Tantalum Ethoxide and Niobium Ethoxide,

http://www.cerac.com/pubs/proddata/ethoxds.htm

71. Y. Takaishi, Process of fabricating dynamic random access memory device

having storage capacitor low in contact resistance and small in leakage current

through tantalum oxide film, United States Patent 5,726,083, March 10, 1998.

72. Tantalum oxide, Ta2O5 for Optical Coating,

http://www.cerac.com/pubs/proddata/ta205.htm

73. A.S. Poghossian, The Mechanism of Blood-Compatibility of Some Materials

Used in Biomedical Sensors. Proceedings of the East Asia Conference on

Chemical Sensors, October 12-13, Supplement B, 1993.

124

74. M. Mukaida, Y. Imai, A. Watanabe, K. Osato, T. Kameyama, K. Fukuda,

Electrical Property of Tantalum oxide Films Prepared by KrF Excimer Laser

Chemical Vapor Deposition, Tran. Mat. Res. Japan, 14A, 767-770, 1994.

75. D. Gerstenberg, and C. J. Calbick, J. Appl. Phys., 35, 402, 1964.

76. K.E. Hennings, and U.D. Strahle, Schweigz. Archiv Fer Angewandte Wissensch.

und Tech. (Switzerland), 35, 365, 1969.

77. F.C. Livermore, P.S. Wilcox, and W.D. Westwood, J. Vac. Sci. Technol. 8, 155,

1971.

78. Resistor technology assessed at microwave frequencies,

http://www.kditriangle.com/pdfiles/APPNOTE5.pdf

79. G. Golan, A. Axelevitch, E. Rabinovitch, and R. Margolin, Vacuum

Photothermal Processing (VPP) for Complex Thin Film Systems Curing, Journal

of Optoelectronics and Advanced Materials, Vol. 1, 4, December 1999, p. 69-80.

80. A. Axelevitch, R. Margolin, E. Rabinovitch, G. Golan, Sputtered Ta Oxide

Films as Precision Resistors, AGIL-2000, 3rd Conference of the Israeli Materials

Society, Jerusalem, Israel, 8-9 November 2000.

81. G. Golan, A. Axelevitch, Novel Sputtering Method For Pd-Al2O3 UV

Transparent Conductive Coatings, Microelectronics Journal, Vol. 31, 6, June

2000, 469-473.

82. G. Golan, A. Axelevitch, and E. Rabinovitch, General Technological Modeling

Method for the Design of Transparent Conductive In2O3 electrodes, J. Vac. Sci.

Technol. A 16(4), 2614-2618, 1998.

83. B. Cernahan, H.A. Luther, J.O. Wilkes, Applied Numerical Methods, John

Wiley & Sons, N.Y., 1969.

125

84. B.P. Demidovich, I.A. Maron, Calculus Mathematics Fundamentals, (in

Russian), Science, Moscow, 1970.

85. G.A. Kachurin, N.B. Pridachin, and L.S. Smirnov, Annealing of Radiation

Defects by Laser Radiation Pulses, Sov. Phys. Semicond., 9, 1975, 946.

86. K. Nishiyama, M. Arai and N. Watanabe, Radiation Annealing of Boron-

Implanted Silicon with a Halogen Lamp, Jap. J. Appl. Phys., 19 (10) October

1980, L563 - L566.

87. A. Gat, Heatpulse annealing of Arsenic Implanted Silicon with a CW Arc

Lamp, IEEE Electron. Dev. Lett., EDL-2, 4, April 1981, 85-87.

88. David Wood, Optoelectronic Semiconductor Devices, Prentice Hall, 1994.

89. R. Singh, V. Vedagarbha, S.V. Nimmagadda, S. Narayanan, Comparative study

of back surface field contact formation using different lamp configurations in

rapid thermal processing, J. Vac. Sci. Technol. B, 16 (2) 1998, 613 618.

90. R. Singh, S. Sinha, R. P. S. Thakur, and P. Chou, Some photoeffect roles in

rapid isothermal processing, Appl. Phys. Lett., 58 (11) March 1991, 1217 - 1219.

91. R. Singh, S. V. Nimmagadda, V. Parihar, Y. Chen, and K. F. Poole, Role of

Rapid Photothermal Processing in Process Integration, IEEE Transactions on

Electron Devices, 45, 3, 1998, 643 - 653.

92. H. Cordes, Y. A. Chang, Interfacial Reactions and Electrical Properties of Ti/n-

GaN Contacts, MRS Internet Journal of Nitride Semiconductor Research,

http://nsr.mij.mrs.org/2/2/complete.utf.html

93. Vacuum Evaporation Sources Catalog, R.D. Mathis Company, 1999.

94. S.O. Kasap, Principles of electrical engineering materials and devices,

McGraw-Hill, Boston, 1997.

126

95. A. Axelevitch, B. Sigalov, E. Rabinovitch and G. Golan, Heterostructures of Si-

Ge multilayer prepared by sputtering techniques, The 17th Israel Vacuum Society

Conference, Tel Aviv, Israel, P6, December 1997.

96. A. Axelevitch, E. Rabinovitch and G. Golan, Sputtered Nanocrystalline

Semiconductors, Workshop on Science and Technology of Nanostructures

(STN-98), Rehovot, Israel, April, 1998.

97. S.O. Kasap, http://Materials.Usask.Ca/server/kasap/WorkExample/example2-2

98. Refractory metals catalog, Molypress Company, 1999,

http://www.molypress.com

99. University of Chicago, Dept. of Geophysical Sci., Prof. D. Archer, PS 134, 1998,

http://geosci.uchicago.edu/~archer/PS134/LabManual/lab.thermal.html

100. D. Halliday, R. Resnick, Physics, part II, John Wiley & Sons, New York,

1962.

101. Light Sources Catalog, McPherson Inc., 1998,

http://www.mcphersoninc.com/lightsources/lightsources.htm

102. Laboratory for the Integration of Sensors and Actuators of Ecole Polytechnique

of Montreal, Catalog of facilities, 1996,

http://lisa.polymtl.ca/Equipement/Fabrication.htm

103. C. W. Nam, S. Ashok, T. Sekiguchi, Thermal anneal activation of near-surface

deep level defects in electroncyclotron resonance hydrogen plasma-exposed

silicon, J. Vac. Sci. Technol. B, 15 (2) 1997, 226 231.

104. K.P. Lee, H. Cho, R.K. Singh, S.J. Pearton, C. Hobbs, and P. Tobin, Ultraviolet

light enhancement of Ta2O5 dry etch rates, J. Vac. Sci. Technol. B 18 (1), 2000,

293 - 295.

105. H.L. Kwok, Electronic Materials, PWS Publishing Company, Boston, 1997.

127

106. K.Hess, Advanced Theory of Semiconductor Devices, IEEE Press, N.Y. 2000.

107. I.S. Grigoriev, E.Z. Meilihov, Handbook of Physical Values, Moscow,

Energoatomizdat, 1991, (Russian).

108. L. Pauling, General Chemistry, W.H. Freeman and Company, 1954.

109. N.W. Ashcroft, N.D. Mermin, Solid State Physics, Holt, Rinenart and

Winston, N.Y., 1976.

110. Effects of Operating Parameters on Capacitive Plasmas,

http://www.timedomaincvd.com/CVD_Fundamentals/plasmas/Cap_parameters.html

111. J.L. Vossen and W. Kern, Thin Film Processes, Academic, N.Y., 1978.

112. H. Conrads and M. Schmidt, Plasma generation and plasma sources, Plasma

Sources Sci.Technol. 9 (2000) 441 454.

113. Plasma Damage,

http://www.timedomaincvd.com/CVD_Fundamentals/plasmas/plasma_damage.html

114. F. Debal, J. Bretagne, J.P. Dauchot, M. Heck, and M. Wautelet, On the role of

plasma-surface interactions in dc magnetron discharges in Ar-N2 gas mixtures,

Plasma Sources Sci.Technol. 10 (2001) 30 -37.

115. J.D. Huba, "Plasma Formulary", Naval Research Laboratory, 2000.

116. F. Mulally, "Plasma Probe Investigation", December 2000,

http://www.netsoc.ucd.ie/~fergalm/plasma/plasma.html

117. G. Golan, A. Axelevitch, R. Margolin, and E. Rabinovitch, Novel Approach to

Sputtered Tantalum Film Resistors With Controlled Pre-Defined Resistance,

Microelectronics Engineering, Vol.32, No 1, 2001, pp.61-67.

118. C. Niu, K. Shepherd, D. Martini, J.A. Kelber, D.R. Jennison, and A. Bogicevic,

Cu Interactions with α-Al2O3(0001): Effects of Surface Hydroxyl Groups vs.

Dehydroxylation by Ar Ion Sputtering, Surface Science, v. 465, 163, 2000.

אביב-אוניברסיטת תל

ש איבי ואלדר פליישמן''הפקולטה להנדסה ע

פיתוח שיטות התזה מתקדמות "דוקטור לפילוסופיה"חיבור לשם קבלת התואר

'אלכסנדר אקסלביץ

אביב-הוגש לסנאט אוניברסיטת תל 2001דצמבר

אביב-אוניברסיטת תל

ש איבי ואלדר פליישמן''הפקולטה להנדסה ע

פיתוח שיטות התזה מתקדמות

'אלכסנדר אקסלביץ "דוקטור לפילוסופיה"חיבור לשם קבלת התואר

אביב-הוגש לסנאט אוניברסיטת תל

א בפקולטה להנדסה''עבודה זה נעשתה באוניברסיטת ת ר גדי גולן'' ודיוסי רוזנווקס' בהדרכה פרופ

2001דצמבר

תקציר

. ומתכות) מ"מל( גובר והולך העניין בשכבות דקות של מוליכים למחצה ות האחרונשניםב

. מ של שכבות דקות''מתקדמות מקבלות עדיפות ביצירת מערכות מל (Sputtering)שיטות התזה

מלאה ועצמאית מאפשרת שליטה אלה אינה למטרות המקובלת כיום מגנטרונית ההתזה השיטת

כל אחד מן הפרמטרים .הפיזיקליים המשתתפים בתהליך יצירת השכבה הדקהבכל הפרמטרים

, מתח ההתזה: כמו למשל, הפיזיקליים של תהליך גידול השכבה באמצעות התזה יונית

ככל שמספר דרגות החופש . הם דרגות חופש בתהליך', וכו, לחץ האדים בפעמון, טמפרטורת המצע

שליטה עצמאית על כל אחת מדרגות . גידול השכבהגדול יותר כך מסובכת השליטה על תהליך

. החופש הללו עשויה להביא לשיפור בתכונות השכבה המגודלת

מספר רב של דרגות בעלת , מתקדמת התזה תשיטהמטרה העיקרית של עבודה זו היא פיתוח

ם ליישומים אלקטרוני,ליצירת שכבות דקות של מוליכים למחצה ומתכות מסוגים שונים, חופש

בגז זוהרת פריקה יצירת הוא זו שיטה חדשה עיון המרכזי העומד מאחרי רה. םאלקטרוני-ופטווא

הניתזים כך שאטומים ,בין המצעהניתוז ל תבין מטרהממוקמת ,)וחהטפלזמה ש(וח טבמבנה ש

בעבודה הפלזמה שנבנה , ואכן. הדדיותהתנגשויותלעבר המצע יוכלו לעשות דרכם ללא מהמטרה

מרחק בין כך גם הודיו פלזמה להיות קטן ה של קיר ועוביעל . מימדי- במבנה דו דקקירזו נראית כ

. גזהשל אטומי (Mean Free Path)חופשי כם הלהיות פחות ממהלעל מרחק זה . מצע למטרהה

כל את . למקומהודורש ריתוק יעיל של הפלזמהבפעמון ארגון התנאי זה מגביל את לחץ גז

שדה . עצמאיתמספקת אלקטרונים מקתודה הי פריקת גז מלאכותית "עקבל ל ניתן ל''התנאים הנ

כמו סליל קתודה זו נראית . קתודההה של תאת הפלזמה בצורמקבע אלקטרומגנטי חיצוני

פלזמה הפרמטרי בצורה זו ניתן לקבל שליטה עצמאית על . היא כקיר דקה תצורוארוך טונגסטן

. למשנהואחד בין מרכיב ללא תלות

. יישום מעשי של שיטת ההתזה החדשהה מתאר את המודל המוצע וון של העבודהחלקה הראש

. טריודהמסוג דיודה וואקום לניתוז יוני מסוג של מערכת טניפשו מודל איכותי בתחילה מוצג

והמודל הפיסיקלי הפלזמה השטוחהמנגנון יצירת תיאור בסיסי כמותי של בשלב זה מובא

גז החופשי של מולקולות המהלך קח בתהליך זה היה ההפרמטר הקריטי שנל. המתאר אותה

גז של הרמת הלחץ את והמותזים קים יחלקמעבר של האת סוג הפרמטר זה מגדיר . בואקום

שטוחה בתנאי העברה הפלזמה המאפייני התנהגות בחינה נעשתה לגבי . ואקוםפעמון הב

של הפעפוע גם תהליך בון ים בהם נלקח בחשגבוליבתנאים ו) הדדיותהתנגשויותללא (יםבליסטי

. המותזיםאטומים ה

בה אומשבתור בסיס למערכת הניתוז החדשה שימשה משאבת דיפוזיה מעבדתית לואקום גבוה

ואילו לוחות כקתודהנימת להט מסוג טונגסטן שימשה . ואקום-בתור פורלואקום נמוך מכנית

שנפלט מן םאלקטרוניהף שט. לאנודה ניתן מתח גבוה ביחס לקתודה. אנודהטנטלום שימשו כ

שדה על ידי ההפתח הכווני שליד הקתודה ובהיותו מוגבל על ידי , אנודההוכוון לעבר הקתודה ה

סודיות של הפלזמה השטוחה יה היתכונות. בקירוב ארסטד100 בעצמה של חיצוניטי המגנה

מדדו נעמון בפאלקטרונים וריכוז היונים ה תורטטמפר. פלזמהלב הלאנגמיר בדיד נלמדו בעזרת מ

מיליטור 0.2 -מהנע החל מדידות בוצעו בלחץ ארגון ה. לאנגמירמדיד של I-V -אופייני הבדיקת ב

במהלך הניסויים נמצא . פלזמההציר ולאורך ואקוםהבתא ים מיליטור במקומות שונ5 -עד לו

. הציר של הפלזמהזור מצויים באוולט - אלקטרון7-7.5 של תמרביבעלי אנרגיה שאלקטרונים

בלחץ בתלות על ציר של הפלזמה , cm-3 6×1011 עד cm-3 3.5×1011בין נע חיובים היונים הריכוז

.הגז

במקביל למשטח המטרה הוצבה . במיםכולו מחזיק מיוחד המקורר גבי על מטרת הניתוז הועמדה

,אך מצידה השניבאותו מרחק מהפלזמה מוקם מחזיק המצע . מ ממנו" ס5במרחק של והפלזמה

ן מוגן מהיה המצע . על המצעלשוב ולהקרין הקתודה ן מולקטרונים שנפלטלמנוע מא מנת על

במשך שעה 60°Cה על ת המצע לא עלתטמפרטור. של הקתודההפלזמה ומקרינה אלקטרונית

הטונגסטןטבעתאלקטרודת לביחס מטרה ההשלילי הופעל על הגבוה המתח . אחת של התזה

. יחסיתהספקים נמוכיםמהיר בידול קצב געובדה זו גרמה ל. תומכתה

100 - וA/min 60 :קצבי הגידול לחומרים אלה היו. טיטניום וסיליקון נבחרו כמטרות להתזה

A/min, 50עם הספק שלא עלה על , בהתאמה W .על אדמה ן הכל המקורות חשמליים בודדו מ

.ואקוםהבתא טפיליים מפני מטענים גונן למנת

, הנלווהיונים ההתזה וזרם המתח את שרת לשנות באופן עצמאי התזה הנוכחית מאפהשיטת

כמערכת במערכת התזה רגילהכמובן לביצוע לא ניתן שינוי עצמאי שכזה . בזמן ההתזה

זרם היונים יכול , שפותחהבמערכת . שבפלזמהקצב ההתזה נקבע על פי זרם היונים. המגנטרונית

מסך מגנטי מלבני . זמה לבין מטרת ההתזהפלהמרחק בין שינוי הלהשתנות בהתאמה עם היה

מסך מגנטי זה . ניע את קיר הפלזמה סביב מחזיק המטרהמסוגל להיוצר שדה מגנטי לא הומוגני ה

. )טטרודה (משמש כאלקטרודה רביעית במערכת

:אחרותניתוז ונידוף לבין שיטות שיטת הניתוז שפותחה השוואה בין ציגה הטבלה הבאה מ

סוג השיטה ית חלקיקים אופייניתאנרג (Pa)ואקום 0.001 < 1 eV נידוף תרמי )טטרודית(התזה טריודית 3000 … 500 0.05 התזה מגנטרונית 100 … 10 0.5 התזה דיודית 1000 10

מספר הניסויים של מאפשרת הקטנה משמעותית הזה מוקדש לשיטה יהתעבודת של החלק השני

ראשון של המתוארת בחלק ה ,שיטת זאת. ות הדקותת השכבתהליך יציראת לייעל על מנת הדרוש

הטיפול בפרמטרים הללו נעשה . פרמטרים בלתי תלוייםהלא פחות משבעמביאה בחשבון ,זהיהת

הרעיון העיקרי מאחורי ". חי תגובהט משתמתודולוגי"בשיטת ו" יםיאקראהחיתוכים השיטת "-ב

נסות לקבל משתנים ול-רב) משוואה(דל פרמטרי כמו-רבהתהליך ההצגה של נעוץ בגישה זאת

יושמה בעבר זוגישה . בצעיםמתהניסויים הבהתאם למספר , סופיפתרון מספרי באמצעותו

נגדים תחמוצת אינדיום ושכבות סוג ופות ומוליכות מקשכבות שלשיפור יעילות הנידוף של

. העשויים מטנטלום

:שכבות דקות בשםהרפיה חומני בול פ שיטה חדשה לטיתארוזה מתיהתעבודת של חלק השלישיב

"Vacuum Photothermal Processing" (VPP) . יהן שלונותכתאת לשפר בשיטה זו ניתן

של המצע המנודף באורכי גל מתחום תזמני- בוהתהליך כולל הקרנה. השכבות הדקות המנותזות

נלקח VPP -ה. חמה הנפלטת מן הקתודה תבקרן אלקטרוני, ובנוסף, IR -הנראה ומתחום ה

של ןייצובאת מאפשרת VPP -שיטת ה. פרמטרי רבהבחשבון כרמה נוספת של חופש במרחב

, כמו כן. הרוסיםp-nבצמתי החזרת המתח המובנה ו, ןהומוגניות שלהבשיפור , השכבות הנבנות

חספוס בממשק שבין צמצם את הלשטח והמצבי של םמספראת קטין להVPP -ביכולתה של ה

שינוי מאפשרת VPP-הביצוע במהלך םאלקטרוניהשליטה על קרן . תכתה תשכבלמ ''מלשכבת ה

בניסוי נתגלה ). מגעי מתכת(לבין המוליך ) סיליקון(מ "בין המלשפוטנציאל המחסום כו של ערב

- ומeV 0.785 - לeV 0.605 - מVPP -הבעזרת עלה Ni-Si - ובTi-Si - בםמחסום הפוטנציאליש

0.571 eV0.672 - ל eV,מה בהתא.