A Plasma

Plasma deposition of nanocomposite thin films
Process concept and realisation
Proefschrift
ter verkrijging van de graad van doctor
aan de Technische Universiteit Eindhoven,
op gezag van de Rector Magnificus, prof.dr. R.A. van Santen,
voor een commissie aangewezen door het College
voor Promoties in het openbaar te verdedigen op
maandag 10 mei 2004 om 16.00 uur
door
Gregory Robert Alcott
geboren te Basingstoke, Engeland
Dit proefschrift is goedgekeurd door de promotoren:

prof.dr.ir. M.C.M. van de Sanden
en
prof.dr.ir. D.C. Schram
The work described in this thesis was supported by The Netherlands Ministry of
Economic Affairs within the framework of the Innovation Directed Research
Program (IOP Oppervlaktetechnologie, IOT 99005).
CIP-DATA LIBRARY TECHNISCHE UNIVERSITEIT EINDHOVEN

Alcott, Gregory Robert
Plasma deposition of nanocomposite thin films: Process concept and realisation /
by Gregory Robert Alcott. Eindhoven : Technische Universiteit Eindhoven, 2004.
- Proefschrift.
ISBN 90-386-1905-7
NUR 926
Trefwoorden: plasma / plasma processing / plasma depositie / plasma chemie /
stoffig plasma / dunne film karakterisatie/ nanogestructureerde dunne films /
hybride deklagen
Key words: plasma / plasma processing / plasma deposition / plasma chemistry /
dusty plasma / thin film characterisation / nanocomposite materials / hybrid
coatings
Printed and bound by Universiteitsdrukkerij Technische Universiteit Eindhoven
Cover realisation by Paul Verspaget.
II
Table of contents
Chapter 1: General introduction
1.1 Introduction
1.1.1 Wet chemical routes
1.1.2 Current state of the art and applications
1.2 PVD processes
1.3 CVD processes
1.3.1 Plasma activated (enhanced) CVD
1.4 Hybrid films via gas phase processes: a literature review
1.4.1 Silicon-based polymers
1.4.2 Gas phase polymerisation of hydrocarbons
1.4.3 Gas phase particle synthesis
1.5 Conclusions
1.6 Scope of this thesis
1
1
2
3
7
7
8
10
11
14
15
18
19
Chapter 2: Anisotropic plasma etching of SiOxCyHz layers

2.1 Introduction
2.1.1 Reactive ion etching
2.2 Experimental
2.2.1 Ion saturation current measurements
2.3 Results
2.3.1 DC bias measurements
2.3.2 Etch results
2.4 Conclusions
25
25
26
28
30
32
33
37
39
Chapter 3: Synthesis and characterisation of SiOxCyHz layers

3.1 Introduction
3.1.1 Problem description
3.2 Process design and reactor concept
3.2.1 Diagnostics and substrate preparation
3.3 Polymer synthesis in the capacitive plasma (Zone II)
3.4 Inorganic synthesis in the inductive plasma (Zone I)
3.5 Film adhesion
3.6 Mechanical/optical properties of hybrid layers
3.7 Discussion
3.8 Conclusions
Appendix A: Vapour pressures of precursors for the CVD of
silicon-based films
Appendix B: Cross cut tape test
41
41
41
44
47
47
48
49
52
53
53
Synthesis, monitoring and characterisation of

nanoparticles for incorporation into nanocomposite layers
4.1 Introduction
4.1.1 Particle formation in r.f. discharges
55
61
Chapter 4:
63
63
64
III
4.2 Forces on particles

4.3 Experimental
4.4 Infrared analysis
4.4.1 Rayleigh scattering
4.5 Physical characterisation of particles
4.6 Discussion and conclusions
Appendix C: Infrared analysis of particles
65
75
77
80
81
83
86
Chapter 5: Reaction kinetics and mechanisms in the plasma

deposition of SiO2 from a TEOS/O2 gas mixture
5.1 Introduction
5.2 Molecular processes in TEOS/O2 discharges
5.3 Experimental
5.4 Mechanisms in the TEOS/oxygen plasma
5.5 Experimental results
5.6 Conclusions
Appendix D: Table of infrared absorption frequencies
91
91
92
94
96
100
102
105
Chapter 6:
Deposition of nanocomposite layers for ultra

low dielectric applications
6.1 Introduction
6.2 Reactor design
6.2.1 Experimental
6.3. Experimental results
6.3.1 Dielectric constant
6.4 Conclusions
Appendix E: Optical emission spectroscopy
Appendix F: Dielectric constant determination
107
107
111
113
114
115
116
118
122
Chapter 7: General Conclusions

7.1 Scratch resistant layers
7.2 Particle monitoring
7.3 Plasma chemistry
7.4 Potential application as low dielectric layers
123
125
125
126
126
Symbols and Abbreviations
127
Summary
129
Samenvatting
131
Acknowledgement
133
Curriculum Vitae
135
Publications related to this work
137
IV
Chapter 1
General introduction
Chapter 1: General introduction

Abstract
Relatively new developments in materials technology, fuelled by the
growing interest surrounding nanotechnology have given rise to a new
breed of materials known as nanocomposites. Nanocomposite are a
subgroup of hybrid materials and represent a two phase material in
which one phase appears in the form of nanometre sized clusters or
particles. The concept of hybrid materials and their origins from
polymer and wet chemical synthetic techniques is presented. The
suitability of these materials in some key applications is reviewed and
the current state of the art regarding material properties discussed.
Advantages and disadvantages of the wet chemical versus gas phase
depositions techniques are given, followed by a brief introduction to gas
phase deposition techniques with particular attention paid to plasma
processes.
1.1 Introduction
In general, the worlds of ceramic and polymer science have developed
independently from one another with little overlap between the two areas.
Traditionally, the field of ceramics concerns itself with the preparation of inorganic
matrix glass materials through sintering or firing processes while the area of
polymers relates to the synthesis and processing of polymerizing network forming
high molecular weight organic molecules. It is quite evident that ceramic and
organic polymer materials strongly differ in chemical and mechanical properties,
and it is for precisely this reason why combining these materials is of great interest.
Typical advantages of organic polymers include flexibility, low density, toughness,
and formability whereas ceramics have excellent mechanical and optical properties
such as surface hardness, modulus, strength, transparency and high refractive
index. If these materials can be effectively combined, a new class of high
performance and highly functional organic-inorganic hybrid materials may be
achieved. The field of hybrids and nanocomposites covers a wide range of
materials and scientific disciplines ranging from polymer chemistry to plasma
physics. An introduction into the subject must therefore start with some definitions.
The terms and classifications given to hybrid materials differ depending on which
discipline they are studied. Classification schemes vary depending on chemical
interaction between the polymeric and ceramic phases, intrinsic dimensions and the
preparation technique used. For clarification, the hybrid materials described in this
thesis are grouped into two main classifications as pictorially presented in Fig. 1.1.
Chapter 1
R
R
Organically modified ceramic: Derived from

adding organic functionality to a ceramic
material.
Nanocomposites: Particles mixed into a

polymeric or ceramic material. Might or
might not be chemically bonded.
= Ceramic matrix
= Polymer chain
= Particles
R = Organic functionality
Figure 1.1: Classification of material forms as described in this thesis.

Organically modified ceramics are principally ceramic materials that have
been modified with organic compounds so that some intrinsically organic
attributes, such as hydrophobicity, are bestowed upon them. As the material is
predominantly ceramic it exhibits similar mechanical and physical properties of
ceramic materials such as high hardness and mechanical strength but poor
flexibility. As the organic content increases the material properties evolve from
those of ceramic to those attributed to plastics, such as toughness and flexibility. In
this study the term organically modified ceramic is also extended to include plasma
polymers deposited from siloxane precursors, as is described in chapter 3, despite
the fact that the structure of the plasma siloxane films is sometimes closer to that of
silicone polymers. Ceramic functionality has also been imparted into polymeric
materials by the addition of nanometre sized ceramic particles. These materials,
herewith referred to as nanocomposites, exhibit improved toughness and reduced
gas permeability than the original plastics. Terms such as CERAMER and
NANOMER have not been adopted in this study due to the inconsistency of use
and wide variety of materials that they have been used to describe in the literature.
In addition to this, phrases such as nano-structured and nano-crystalline are
often used in the context of thin film morphology [1, 2] where the term nano
refers to the presence of nanometre sized crystallites within one material
composition (cf. c-Si [3, 4]). Such materials are not included within the scope of
this work and the term nanocomposite refers to materials composed of nanometre
sized ceramic particles embedded in polymeric layers.
1.1.1 Wet chemical routes
The traditional synthetic route to hybrid materials is the sol-gel process. Sol-gel is a
method to produce ceramic materials by preparation of a colloidal suspension of
precursors (sol), gelation of the sol and removal of the solvent. The precursors are
typically metal alkoxide compounds and derivatives thereof. One of the most
studied sol-gel chemistries is that of the alkoxide tetraethoxysilane (TEOS) [5, 6, 7,
8]. Under aqueous conditions, the ethanoic groups on TEOS are hydrolysed to form
2
Chapter 1
silanol groups. Application of heat then drives the reaction to completion, liberating
H2O and forming SiO2. This reaction is depicted in Fig. 1.2.
Si OR + H 2O Si OH + ROH
Si OH + HO Si Si O Si + H 2O or
Si OH + RO Si Si O Si + ROH
Figure 1.2: Hydrolysis of TEOS, R represents an arbitrary hydrocarbon

chain, in the case of TEOS, R = C2H5.
During the gelation phase the sol-gel solution can easily be cast into a
complex shape or used to coat suitable substrates. Coating is achieved via spinning
or dipping techniques where the final coating thickness is determined from
spinning/dipping speeds and the viscosity of the gel solution. After casting or
coating, the sol-gel solution is cured to complete the chemical reactions and drive
out the remaining water or alcoholic by products and densify the material. Curing is
typically achieved by baking the sample in an oven at temperatures up to and
exceeding 400C, but exposure to plasma or UV radiation from a lamp have also
shown to be effective curing methods [9]. For an extensive review of sol-gel
technology see J. Brinker et al. [10].
1.1.2 Current state of the art & Applications

The sol-gel process can be modified via one of two routes to produce hybrid
materials. Functionalised alkoxide precursors such as (3-glycidoxypropyl)
trimethoxysilane (Glymol) [11, 12] or isocyanatopropyl triethoxy silane [13] are
used in place of, or in conjunction with conventional alkoxides. These react to form
materials containing organic bridging units. Considering Glymol as an example,
hydrolysis reactions occur on only three of the available positions surrounding the
silicon atom and polymerization occurs at the epoxide group located at the end of
the carbon chain. The resulting silica structure, containing organic bridging units, is
depicted in Fig. 1.3.
R
O
Si
Si
R O O Si
Si O
Si O
Si
Si
Si
Si
Figure 1.3: Theoretical structure of organically modified ceramic

synthesized from (3-glycidoxypropyl) trimethoxysilane.
3
Chapter 1
In addition to improving the flexibility and toughness of the final material, the
organic bridging units also lower the curing temperature necessary to achieve
material densification. This is believed to be because the added pliancy of the
material enables it to relax into a denser structure with lower energy loads [14].
An alternative to using a modified precursor is to prepare a precursor solution
where the organic and inorganic components are simultaneously synthesized from
two different precursors. Materials synthesized from two component solutions (a
precursor such as TEOS for the inorganic phase and glymol for the organic matrix)
have been developed by the Fraunhofer Institute and are know by the trade name
ORMOCER [15]. Other authors have referred to these materials as
nanocomposites, perhaps partly due to the copyright, but also as slight phase
separation can occur during curing, resulting in the formation of SiO2-like clusters
embedded in the predominantly organic matrix.
Inherently weak polymers are also reinforced by blending particulate fillers
into the polymer resin prior to it being cured into a network structure [16, 17, 18,
19, 20, 21]. The disadvantages of this approach are the invariable coalescence of
the filler particles into large aggregates and poor chemical interaction between the
filler and polymer. This results in poor dispersion and consequently limited
improvements in mechanical properties. Since the first nanocomposite materials
were prepared shortly after the major interest in synthetic plastics started in the
1950s, a variety of materials have been investigated as fillers including, clays [22],
carbon nanotubes and cellulose whiskers.
Composite materials also have important applications in areas such as
catalysis and electronics. In a recent paper Mohallem et al. [23] synthesized
magnetic NiFe2O4/SiO2 nanocomposite films using sol-gel techniques. The films
consisted of 6 to 10 nm sized particles of crystalline NiFe2O4 dispersed in an
amorphous SiO2. Yano et al. [24, 25] compared various mechanical properties,
including stress-strain curves, sorption isotherms and dynamic viscoelastic
properties of hydroxypropyl cellulose (HPC) and poly (vinyl acetate) combined
with silica prepared via sol-gel synthesis (with TEOS) to blended samples with
untreated silica particles (diameter 20-40m). They discovered that even with
minimal amounts of silica, the sol-gel prepared samples exhibited drastic
improvements in mechanical properties, whereas a comparative test with blended
samples showed only minimal improvements. A decrease in overall tensile strength
of the blended samples as the silica content increased, was attributed to aggregation
of the silica particles with in the HPC matrix to form weak points, a phenomena
which has also been observed by others [26, 27, 28]. Other chemistries such as (nbutyl methacrylate)/titania [29] and poly(vinyl alcohol)/alumina [30] have also
been evaluated with respect to mechanical properties with similar results.
Improvements in material properties and particle dispersibility have been
demonstrated by subjecting the particles to a chemical pre-treatment, before mixing
with the polymer matrix [31].
More recently, nanometre scale structuring of organic/inorganic hybrid
materials has been synthesized using Polyhedral Oligomeric Silsesquioxane
(POSS) based precursors for the inorganic component [32]. The prefabricated silica
cage structures of the POSS compounds (Fig. 1.4) provide more control over the
4
Chapter 1
structure and content of the SiO2 phase. The organic side groups (R) can be varied
so as to maximise bonding to the organic network (Fig. 1.4). An overview of
optimum mechanical properties and applications for wet chemically synthesised
hybrid thin films is given in Table 1.1.
Figure 1.4: Structural representation of Polyhedral

Oligomeric Silsesquioxane (POSS). R can easily be tailored to
maximise bonding to organic network.
Table 1.1: Optimal scratch resistant properties achieved for different wet
chemically synthesized hybrid materials.
Reference
Material
After 500 taber cycles
Application
Li [33]
HPO/TMOS
>98 %
Glazing Hard
Coats
Barrel abrasion
Ophthalmic
>99 %
95-96 %
<97 % (200 cycles)
Lacquers
Urreaga [34] TEOS/MPS/MMA

Haas [35]
Al(iPA)3/glymol
Ti(iPA)4/glymol
Zr(C3H7O)4/glymol
These materials have been evaluated as low dielectric constant (low-k) layers
in semiconductor devices [36, 37]. Porous SiO2 layers synthesized from various
POSS precursors [38, 39, 40] and prepared using sol-gel deposition techniques
have produced thin films with dielectric constants typically between 2-3. Sol-gel
processes with siloxane precursor have also been extensively studied [41, 42, 43,
44], with dielectric constants as low as 1.7 [45] being reported using TEOS as a
precursor. These studies show that the film porosity and dielectric constant
decrease as the curing temperature and time increase. Gas phase technologies are
an attractive alternative to wet chemical processes as they utilize the same vacuum
reactors and toolsets as are already used in the semiconductor industries. Synthesis
of porous SiO2 layers from siloxane precursors using expanding thermal plasma
[46], PECVD [47, 48] and reactive evaporation of SiO [49] have all been
investigated.
5
Chapter 1
Sol-gel processes are a popular choice for fabricating low-k films because they
inherently produce porous materials. Dielectric layers with void volume fraction as
high as 95 % have been reported [44]. Although such materials have dielectric
constants around 1, their susceptibility to copper diffusion causes problems when
attempting to implement them into semiconductor processes. On the other hand,
gas phase deposition techniques inherently deposit dense materials and can easily
be implemented into existing semiconductor processes. The complimentary nature
of these two techniques has been recognised by a number of researchers who used
PECVD [50] or PVD [51] technology to encapsulate wet chemically synthesized
xerogels. Several authors have investigated the mechanical properties and other
issues concerning the implementation of these porous materials in semiconductor
processes [52, 53, 54, 55].
Wet chemical synthesis of hybrid films typically involves several process
steps. A coating solution needs to be prepared, the substrate coated via a dip or
spinning technique followed by post-deposition curing procedure. Solvents are
usually required to prepare the coating solution, and high temperatures (or other
energy load) to complete the curing. Gas phase deposition processes are typically
one-step solvent free processes and are hence more attractive for the large-scale
coating of hybrid thin films. Advantages and disadvantages of wet chemical and
gas phase deposition processes are summarized in Table 1.2.
Table 1.2: Advantages and disadvantages of gas phase and wet chemical
processing techniques for hybrid materials.
Process
Advantage
Disadvantage
Use of potentially harmful
solvents
Wet
chemical
Good control of
process chemistry
Relatively low
investment costs
Single step
deposition process
Gas Phase
Additional processing of
nano-powders necessary to
achieve dispersion and
bonding to plastic matrix
Poor control over process
chemistry
No solvents
No post deposition
curing
Long curing time and high

temperatures
Relatively high investment

cost
Chapter 1
Gas phase deposition processes are generally classified as either Chemical

Vapour Deposition (CVD) or Physical Vapour Deposition (PVD) processes.
1.2 PVD processes
Classical PVD processes differ from CVD processes in that no chemical reactions
occur, instead a sample ingot of the desired composition is atomised (either by ebeam, plasma sputtering, thermal evaporation or some other method),
transported/processed and deposited onto a suitable substrate (Fig. 1.5). Low
pressures are used (10-5 to 10-8 Torr) to ensure mean free path lengths are in the
order, or greater than the reactor dimensions and thus sputtered or evaporated
particles reach the substrate without being scattered to the reactor walls by the
background gas. Film properties and homogeneity are affected by the application of
an electric field (causing ionisation of the gas) and varying the process pressure by
the introduction of inert or sometimes reactive gas. In the plasma version of the
technique, magnetic fields are used to enhance plasma density and thus deposition
rate [56]. As no gas collisions occur, PVD technologies are line of sight processes
and therefore difficult to implement for three dimensional substrates, requiring the
use of mechanised substrate holders. CVD techniques differ from PVD techniques
in that chemical reactions occur so that the final film is of a different composition
to that of the precursors.
0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000 000 000 000 000 000 000 000 Substrate
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0holder
000000000000000000000000
00 00 00 00 00 000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000000000000000000000000000000000000000000000000000
Molten00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000000000000000000000000000000000000000000000000000
pool 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Pumping
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 0 0 0 0 0 0 0
port
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e-beam
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000000000000000000
Separation
plate
Ingot
rod
Pumps
e-beam
source
Vacuum
chamber
Figure 1.5: Schematic of e-beam evaporation apparatus.

1.3 CVD processes.
Chemical vapour deposition is a process whereby solid films are synthesised from
chemically reactive gas mixture. The energy required to drive the chemical
reactions can be supplied in a number of different forms the most common of
7
Chapter 1
which are heat (thermal CVD), electrical discharge (plasma enhanced or activated
CVD), photons (photo CVD) and even sound (acoustic CVD). A typical CVD
system consists of three components, a gas handling system where suitable
precursors are vaporized and dosed, a reactor vessel where precursors react and an
exhaust system where harmful reaction waste products are scrubbed and the
harmless gases vented. The physical and chemical properties of the precursor are
one of the most important considerations when designing a CVD system. A suitable
CVD precursor would have the following characteristics:
Volatile and stable during transport to the CVD reactor.

Readily decompose under reaction conditions to form the desired coating.
Low toxicity.
Cost effective.
Not damaging to deposition apparatus.
The choice of activation method depends on the substrate material used. For
thermally stable substrates, thermal activation is preferred as it usually produces
superior coating properties for all but polymer-type films.
1.3.1 Plasma activated (enhanced) CVD
Plasma is (at least for physicists) ionised gas consisting of a mixture of free
electrons, ions and neutral atoms (depending on the degree of ionisation) and is
formed when sufficient energy is put in to a gas. More than 99 % of matter in the
universe exists in the plasma state though nature rarely produces plasma on the
earth's surface. In nature plasma is formed when gas is exposed to high
temperatures (inside of stars) or high energy radiation (UV radiation in the
ionosphere). Under laboratory conditions however, where such high energies are
not available, ionised gas is generated using high electric fields and is known as
cold plasma or electric discharge. Electrical discharge is the most practical means
of creating and sustaining low temperature plasma in the laboratory and many
methods of coupling electrical energy into gases to generate plasma have been
developed using both d.c. as well as a.c. power sources. The two most common
methods for coupling electrical energy into a gas discharge are via capacitive
coupling (CCP) (Fig. 1.6a), or via induction as is done with inductively coupled
plasma (ICP) (Fig. 1.6b). Inherently different plasma conditions are created with
CCP and ICP discharges and the choice of ICP or CCP discharge depends on the
application.
Chapter 1
Gas inlet
a.
r.f.
~
Gas inlet
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000 0000000 000000 000000000
Pumps
b.
Pressure
Gauges
Substrate
table
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ~ r.f.
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0 0 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
000000000000000000000000000
Pumps
Figure 1.6: Typical apparatus for CVD experiments, a. direct

capacitively coupled parallel plate reactor, b. remote inductively
coupled plasma.
Capacitively coupled plasmas are characterised by relatively low plasma
density but high energy ion bombardment of the substrate surface. Most of the
chemistry occurs on the surfaces within the reactor due to the high energy ion
bombardment. High energy ions causes relatively unselective fragmentation of the
surface adsorbed species and deposition or etch processes result depending on the
power and molecules involved. These plasmas are typically used for thin film
deposition/modification and, due to the high energy uni-directional ions, are
particularly useful for high aspect ratio anisotropic etching. Remote plasma
processes (such as ICP [57], electron cyclotron resonance (ECR) [58], helicon [59,
60], corona [61] and expanding thermal plasmas [62] sources) differ in that they
produce higher density plasmas and the precursors are injected downstream of the
plasma source. High density and absence of high energy ions at substrate surface
(unless additional biasing is applied) results in a higher proportion of the chemistry
occurs in the gas phase. These sources are useful in applications where high
densities and temperatures are needed without high levels of ion bombardment to
the surface.
Gas discharges are intrinsically none equilibrium systems and consequently
need to be characterised by several parameters. Plasma is usually described in
terms of electron and ion temperatures, Te, Ti, and the electron and ion number
densities (number of species per cubic metre of gas) ne and ni. As electrons are
point charge carriers with no vibrational or rotational energy levels, their
temperature directly equates to their translational energy and is measured in
electron volts (eV)*. Low temperature plasma can only be sustained by the
continuous power input to compensate for losses. In the case of an electrical
*
Where 1 eV = e/kB = 11604 K, in which e is the electron charge and kB the Boltzman
constant.
9
Chapter 1
discharge, this energy is used to sustaining electric fields that accelerate and hence
heat the free electrons and ions. If the gas pressure is low enough and hence the
mean free path (mfp) long enough, the electrons reach sufficient energy that they
either undergo ionising collision with background neutral gas or go straight to the
vessel walls where they recombine with positive ions to reform the neutral gas. If
no chemical reactants are injected into the plasma then a majority of the electrical
energy put into the discharge leaves the system in the form of heat, sputter
processes at the vessel wall and light emission from the plasma glow. For a more
thorough description of plasma processing see Chapman [63] or Lieberman et al.
[64].
1.4 Hybrid films via gas phase processes: a literature review
A review of literature pertaining to the deposition of hybrid thin films via gas phase
technology is sought. Due to the broad project description and the variety of
scientific disciplines it encompasses, a strategy in which to approach the literature
is imperative. For this reason literature reviewing thermal and plasma activated gas
phase deposition techniques, synthesis of ceramic and polymeric materials and the
current state-of-the-art regarding nano-composites is put into a literature database.
From this database any presence of an overlap in process techniques and condition
for the synthesis of ceramic and polymeric phases can be found. If however no
overlap in process conditions exists, then it will be necessary to develop a dual
source process whereby the ceramic and polymer phases can be independently
activated before being combined to form the hybrid film. This approach to the
literature is schematically represented in Fig. 1.7.
Deposition of
organic/polymer
coatings
One set of
process
conditions
Precursors
Deposition system
Plasma ( Type )
Thermal
Deposition conditions
Coating characteristics
Structure
Properties
Overlaps in conditions?
Deposition of
inorganic
coatings
Multiple
process
conditions
Figure 1.7: Flowchart representing approach to the literature review

Literature specific to gas phase processing of 3 main material groups is
therefore reviewed, namely,
10
Chapter 1
Silicon based polymers. Gas phase polymerisation of alkoxy silicon

precursors as possible synthetic routes to silica/silicone hybrid materials.
Gas phase polymerisation of hydrocarbons. Gas phase synthesis of
compounds from purely carbon based precursors.
Gas phase particle synthesis. Synthesis and control of particles in gas
phase processes.
1.4.1 Silicon based polymers
The plasma polymerisation of siloxane precursors under various conditions is a
subject that has received a great deal of attention by researchers and has been
extensively reported in the literature. Many authors however, especially those
seeking to deposit SiO2, do not classify the plasma polymerised siloxane thin films
as organically modified ceramics despite the presence of carbon and hydrogen.
Plasma deposition processes can be separated into two classes, direct and remote
plasma (see Fig. 1.6). When a siloxane monomer is introduced into plasma, it is
activated and fragmented by high-energy electrons and ions. At low pressure and
hence long mean free path, activated molecular fragments and ions are more likely
to collide with the vessel walls then with other gas phase species (unless they are
negatively charged) and hence absorb and react on surfaces within the reactor to
form thin films of similar composition and structure to the injected monomer. This
process has been used to deposit hybrid films from a variety of organometallic
precursors with applications ranging from biomedical films for implants [65],
barrier coatings for food packaging [66, 67] to dielectric layers in integrated
circuits [68, 69]. The most popular precursors for the deposition of organically
modified ceramics are hexamethyldisiloxane (HMDSO) and tetramethoxysilane
(TEOS) often in combination with oxygen and/or a variety of dopants. The physical
and chemical properties of the resulting films depend heavily on the process
parameters, such as, the way in which the power is coupled into the gas
(inductive/capacitive), precursor injection point, the power per unit monomer mass
flow*, and whether or not an oxidant is added. Coatings produced in this way from
siloxane-based precursors are known as Plasma Polymer (PP) coatings and have the
general composition SiOxCyHz where x 2 and y and z can be anything up to the
stoichiometry of the injected precursor. If the deposition is performed at relatively
low discharge power and without the addition of oxidant, then the film generally
has a similar elemental composition to the precursor. A number of studies have
been performed which compare film composition to process conditions and
precursors [70, 71, 72]. A comparison between the IR spectra of the pure 1,2bis(trimethyl)siloxyethane (TMSE) precursor and subsequent plasma polymer thin
film can be seen in Fig. 1.8. At higher discharge powers, Fracassi et al. [73] found
that while the elemental composition of the film was similar to that of the
precursor, the concentration of specific molecular groups (C-H, C-O, O-C-O) and
hence functionality differed.
*
Calculated by W/FM where W = power input (Watts), F = monomer mass flow (g/s) and M
= relative molecular mass of monomer.
11
Transmission (0-1)
Chapter 1
1
Flim
0.9
0.8
0.7
Liquid
0.6
0.5
0.4
0.3
0.2
3800
3500
CO 2
C-H
Si-CHx
3200
2900
2600
2300
2000
1700
1400
1100
Si-C
800
501
-1
Wavenumber [cm ]
Figure 1.8: Retention of chemical functionality between thin film and

precursor for plasma polymerised created from TMSE (from this work,
see chapter 3).
The chemistry of the siloxane plasma process is complicated by the wide
variety of compounds formed by both carbon and silicon, however some general
observations can be made. Inagaki [74] concluded from a study of purely organic
monomers, unsaturated compounds exhibit higher deposition rates than their
saturated counter parts. The higher deposition rate is attributed to the unsaturated
structure of the monomers, and the deposited films appear to contain some of the
monomers conjugation. The readily polymerising nature of unsaturated
hydrocarbon systems over the saturated counterparts is also true for siloxane
precursors containing unsaturated carbon ligands [65, 70].
Addition of oxygen to the siloxane plasma vastly complicates the process
chemistry and has been observed in films even when no oxygen is deliberately
injected. A majority of authors attribute this to irremovable water molecules
adsorbed on the surface of the substrate and deposition system. However, in some
cases the hydroscopic nature of the precursor itself can be a cause of oxygen
incorporation. A number of authors have reported adding O2 during the plasma
polymerisation of a number of siloxane precursors [65, 71, 75, 76, 77, 78, 79] in
attempts to deposit films with a SiO2 composition. Oxygen is readily fragmented by
the plasma state and forms very reactive atomic oxygen and ozone which react with
the siloxane monomers. At low oxygen concentrations, the action of atomic oxygen
and ozone dramatically increases film growth rate but can cause powder formation
at higher concentrations. Oxygen also readily etches the growing film surface to
remove carbon and retards deposition rate. As the oxygen fraction in the process
gas increases, the resulting coating composition goes from being similar to the
monomer precursor composition, to that of SiO2. However, high purity SiO2 is
difficult to attain because of silanol (SiOH) incorporation. Silanol incorporation is
not a result of incomplete precursor oxidation, but of reactions between growth
precursors and OH radicals that are inevitably present when hydrocarbon and
oxygen chemistries are mixed. As an example of such a complex system, a more
12
Chapter 1
extensive investigation into the chemistry of the siloxane and oxygen plasma
process is presented in chapter 5.
To achieve dense stoichiometric (silanol free) silica from siloxane precursors
requires both high density plasma and high substrate temperatures, or alternatively
high energy ion bombardment is required. Menichella et al. [80] deposited a
gradient layer coating by starting with a pure TEOS gas mixture and gradually
increasing the oxygen concentration during the deposition process. By gradually
increasing the oxygen concentration the deposited the film gradually became more
SiO2 like at the top surface. The intermediate layer between the SiO2-like surface
and bulk plastic substrate provided a smooth transition in thermal expansion
coefficients, while the top SiO2 provided the scratch resistance and hardness
required. Similar processes have been reported where alternate plasma polymerSiO2 layers are deposited to form a multilayer stack. An overview of some silicon
based silica/silicone hybrid coating chemistries and applications are summarised in
Table 1.3 below.
Table 1.3: Precursors used in the deposition of organically modified ceramic
films.
Precursors
Dopant/oxidant
HMDSO [65, 75, 76, 77, 72, 69, 78,

79, 71,81, 82, 83, 84, 85, 86, 87,
88]
O2, C2F4
TMDSO [65, 76, 84]
O2
Properties/
application
Barrier coatings,
Adhesion
promoters
Low-k
Barrier coatings
VpMDSO [84]
O2
Barrier coatings
TMTSO [84]
O2
Barrier coatings
TVTMTSO [84]
O2
CF3(CF2)nC2H4Si(
OCH3)3, n=0, 5, 7
Barrier coatings
Hydrophobic
coatings
Ophthalmic
coatings
TMMOS [89]
Silane [90]
O2
Methyl silane [91]
HMDS [68]
[92]
Low-k
Ophthalmic
DVS [68]
Low-k
TMS [93]
MTS [68]
O2
-
Low-k
Low-k
13
Chapter 1
Metroline et al. [94] have a number of patents for apparatus and processes for
depositing SiO2/Si-polymer-like coatings for protecting metal [95, 96, 97, 98].
1.4.2 Gas phase polymerisation of hydrocarbons
As the plasma will fragment most organic compounds, plasma polymers can be
deposited from almost any organic monomer. Polymeric films have been deposited
from a number of unsaturated and saturated hydrocarbon compounds such as the
styrene monomer [99, 100], ethylene [101], naphthalene [102], ethylene oxide
[103, 104], methane [105] and a mixture of fluorocarbons [105, 103, 104, 106].
Typically an organic precursor will undergo numerous fragmentations in the
plasma, and the resulting film will form from most of them. Unlike wet chemical
synthetic methods, plasma polymers differ in chemical composition from the
monomers, even if the same monomers are used in the two processes. Plasma
polymers are in most cases highly branched, highly cross-linked, insoluble and
adhere to solid surfaces. The chemical and physical properties of the plasma
polymers depend on the precursor gas and the type of discharge in a similar way to
the siloxane chemistry. An overview of some hydrocarbon based coating
chemistries and applications are summarised in Table 1.4 below.
Table 1.4: Some plasma polymers and their applications.
Precursor
Additives
Process
Application
Styrene [100, 107, 99]
O2, H2
a.c./d.c., r.f.
Ethane [101, 108]
r.f.
1-Fluoro naphthalene[109]
r.f.
Hexane [103, 104]
CH2CO/H2
r.f.
Fluoro ethene, Fluoro

ethane, Fluoro butane
[106]
H2
40 kHz
parallel
plate
Hydrophobic
protective
coatings
Methane/tetrafluoromethan
e [105]
SiH4/H2
r.f. ICP
Anti reflection
coatings
Electrical
components
Barrier/protective
coating
Not specified
Biocompatible
coatings
In general, plasma polymers find applications as protective coatings on

metals, hydrophobic coatings and insulating layers in electrical applications. The
use of highly unsaturated hydrocarbon sources such as C2H2, in combination with
high energy plasma, results in the deposition of extremely dense and highly cross
linked films. These materials are known as diamond like carbon films (DLC), and
are extremely hard. The Monsanto Company hold patents on DLC films [110], for
14
Chapter 1
scratch resistant coating applications on glass substrates. Diamonex Inc. also holds
a number of patents regarding the deposition of DLC coatings [111]. A number of
excellent reviews regarding plasma polymers and plasma polymerisation
techniques can be found elsewhere [112, 113, 114].
The only commercial, thermally driven process for applying purely carbon
based polymers is the Parylene processes. Here, one of the parylene precursors
(in the form of a xylylene dimer) is vaporized, pyrolized in an oven at 600C and
transported to a cooled substrate where the xylylene radicals condense and react to
form a layer. The advantages of the parylene process are that deposition can be
done at atmospheric pressure and temperature. The resulting polymers are however,
rather soft and porous compared to plasma polymers, but have nevertheless been
considered for application such as encapsulation layers for poly LED devices [115]
and a substitute for SiO2 as low dielectric in chip manufacturing [116, 117]. In a
unique study Senkevich et al. [118], synthesised and incorporated silica nano
powders into growing parylene thin films. SiO2 particles were generated by thermal
decomposition of diacetoxy-di-tert-butoxysilane (DADBS) in oxygen and
transported, via predominately gas flow, to the deposition chamber and
incorporated into Parylene-C films. Parylene-C was synthesised from dichloro-diparaxylylene (DPXC) and both DPXC and DADBS were simultaneously
decomposed and activated at 615C and 600C respectively prior to entering the
deposition zone, which was at room temperature.
1.4.3 Gas phase particle synthesis
In the CVD process, a feed source gas is first decomposed by heat or plasma,
followed by diffusion of the decomposition products onto the substrate to form a
thin film. The decomposition products can also be used to form particles via either
homogeneous gas phase nucleation or heterogeneous nucleation at surfaces within
the reactor. These particles can then go on to be incorporated into the growing film.
In general particle or powder formation in CVD processes is undesirable, and in the
semiconductor industry great lengths are taken to prevent particles from entering
the reactors [119]. In some processes, however, homogeneous gas phase nucleation
is encouraged as the incorporation of particles into the growing film can offer some
beneficial properties. Particle precipitation during CVD results in porous layers that
can be used as supports for catalysts, as membranes or porous electrodes. If excess
homogenous nucleation occurs, and a majority of the precursor is consumed before
reaching the substrate, then loose powdery deposits are formed instead of a coating.
A summary of some synthetic processes is listed in Table 1.5.
15
Chapter 1
Table 1.5: Precursors and processes used for the gas phase synthesis of
particles.
Particle
Precursor
Process
SiO2
TEOS [120], TRIES [120],

TMOS [120], OMCTS
[120], SiH4 [121] + O2/O3
Thermal excitation
Mo
Mo(CO)6 [129]
Microwave Plasma
CeO2/CexY1-xO2-y
CeCl3, YCl3 + O2 [122]
Thermal excitation (size 980 nm)
Si
SiH4 [128, 123]
CO2 laser excitation
Si3N4
SiH4 + NH3 [124]
SiC
SiH4 + CH4 [124]
Pt-TiO2
Pt, Ti + O2 [130]
Joule heating / Magnetron

sputtering (size 30-150
nm)
Pt-Al2O3
Pt, Al + O2 [130]
Joule heating / Magnetron

sputtering.
TiN
TiCl4 NH3 [125]
Thermal excitation
ZrO2
Zr(CO)4 [126],
Zr(n-C3H7O)4 [127] + O2
Inert gas condensation /

flame pyrolysis (size 0.3-1
m).
A majority of gas phase powder forming processes are in fact designed for
synthesising particles, which are later sintered into bulk ceramic materials. In most
of these methods gaseous precursors are introduced into a reactor system whereby
energy is supplied causing homogenous nucleation reactions to occur which result
in the formation of particles. Energy used to drive the chemical reactions can be
supplied in a number of different forms including high power lasers [128],
microwave discharge [129] and r.f. magnetron sources [130].
E. Whitby et al. [121] found that particle nucleation is more favourable where
high temperature gradients exist. This is a result of high monomer activation in the
high temperature region, followed by high concentrations in the low temperature
region. This principle was used by Okuyama et al. [131] where they used a sixzone furnace to synthesise SiO2 particles. A similar effect occurs in the CO2 laser
excitation processes [128, 123, 124] as depicted in Fig. 1.9. As the precursor gas
flow passes through the CO2 laser beam, rapid thermal excitation of the gas occurs.
16
Chapter 1
The excited gas then leaves the laser beam and rapidly cools, nucleates and forms
particles.
Reaction
flame
Water cooled
copper beam
stop
Filter/Particle
collection
CO2 laser
Laser
Inert gas
inlet
Precursor
inlet
Figure 1.9: Laser activated particle-forming processes, as used by

Cannon et al. [124].
Once formed the behaviour of very small particles (>20 nm) is primarily
governed by diffusion. For larger particles, thermophoretic or electrophoretic forces
begin to govern the particles behaviour. In order to understand how particles
behave in CVD reactors, a number of authors have attempted to model the
processes [132]. Whitby et al. [121] describes a number of plausible explanations
for the formation and movement of particles in a thermal CVD reactor. Amongst
other things their numerical studies indicated that particle nucleation shows a steep
dependency on pressure. Whitby et al. and Fujimoto et al. [120] proposed
mechanisms for the formation/incorporation of particles and the transport of
particles during the deposition of SiO2 films. Bai et al. [122] thermally decomposed
CeCl3 and YCl3 in the presence of O2 to form CeO2 and Yttrium doped CeO2
powders. They characterised the collected powders to find they had a size
distributions of either 9-30 or 30-80 nm depending on whether they collected the
material in a cooled container outside of the reactor zone (9-30 nm), or an actively
cooled susceptor situated inside the reactor (30-80 nm).
In the semiconductor and micro electronic deposition systems, particle
production is an undesirable by-product of some plasma processing steps. Device
failure as a result of dust contamination is a familiar issue and has prompted
research into particle formation and transport mechanisms in the hope that it might
be avoided [133]. A prominent area of literature regarding particle contamination
deals with silane chemistries [134, 135, 136, 137, 138, 139] due to its wide spread
application in industrial processes. Although particles have been observed in TEOS
and HMDSO [140] plasma discharges, very little has been done in elucidating the
mechanism by which they are formed. Fujimoto et al. [141] investigated particulate
17
Chapter 1
formation from four siloxane precursors (triethoxysilane, tetramethylorthosilicate,

octamethylcyclotetrasiloxane and TEOS) in an atmospheric pressure, thermally
activated CVD process and determined TEOS, with the lowest overall reaction
activation energy, would most readily form particles. A more detailed study on the
formation and transport of nanometre sized particles formed in a TEOS/oxygen
plasma is presented in chapter 4.
Thin film and particle properties depend on the synthesis technique. Wet
chemical processes produce very porous soft materials that require some form of
curing to achieve densification. PECVD and PVD materials are inherently denser,
but rarely produce fully dense crystalline materials without additional substrate
heating. Atmospheric pressure, high temperature CVD processes generally produce
the hardest and most crystalline materials. It is therefore reasonable to assume that
particle generated in low pressure discharges will also exhibit some level of
porosity.
Porosity in PECVD synthesized silica and siloxane layers can be controlled by
applying a d.c. bias potential (as demonstrated in chapter 3). The particles however,
are formed from negative ions trapped in the plasma glow and the sheath
surrounding them, and hence the energy of ion impinging on them, is much lower
than the sheath above the electrode (as described in chapter 4). As the formation of
dense fully oxidized SiO2 networks requires high energy input, (as seen with SiO2
deposition in chapter 3) the energy is not available to form dense particles.
1.5 Conclusions
Very little has been reported about the gas phase deposition of nanocomposite
materials, especially where inorganic and organic phases are combined. A great
deal of literature is available regarding the plasma polymerisation of various
siloxane precursors for a wide variety of applications, including scratch resistant
layers and low-k materials.
The synthesis of dense siloxane films is done using capacitively coupled
plasma configurations where self biasing helps densify the layer.
Films composition and structure can be characterised using FTIR
spectroscopy, XPS and Auger spectroscopy.
The bonding of these films has to some extent been revealed using
techniques such as FTIR and X-ray photon spectroscopy. Unfortunately,
elucidation of the exact connection between the inorganic and organic
components of these materials has not been fully characterised.
Although the authors rarely describe their films as hybrid materials, they are
never the less composed of both inorganic and organic components. From a hybridmaterial point of view, a majority of these coatings are perhaps best described as
organically modified ceramics, primarily inorganic with organic functionality. In
cases where no oxygen is added, and relatively low energy plasma are used, the
materials resemble highly cross-linked silicon polymers.
18
Chapter 1
In order to maintain a good control of film composition, the inorganic and

organic phases need to undergo activation separately and the active species
combined in the deposition zone housing the substrate. Using this method, the
inorganic and organic phases can be separately optimised before they are combined
to generate the single coating. The reactor must also be flexible, so that it can be
used to deposit the organically modified ceramic layers as well as nanocomposite
layers as described in Fig. 1.1.
1.6 Scope of this thesis
The primary aim of this work is to develop a process for the synthesis of hybrid
materials using solely gas phase synthetic methods. To achieve this, an
understanding of gas phase deposition technologies is necessary, brief introduction
to which has been presented. From this understanding, process conditions required
to synthesize nanocomposite materials can be identified. The strategy to achieve
this is based on the continuous adaptation of a dual plasma reactor which is used
throughout this study in various configurations. The dual plasma system provides
independent control over conditions for the synthesis of both phases of the final
material.
The thesis is divided into seven chapters. An introduction to hybrid materials
has already been presented detailing the various classifications of hybrid materials,
their origins from wet chemical processes and the most promising techniques from
which they might be deposited using purely gas phase deposition technology. This
is followed in chapter 2, by an introduction into plasma physics in a study of the
helicon etch machine. This work is performed at the Australian National University
in Canberra and investigates the feasibility of etching silicone based hybrid
materials using SF6, CF4 and O2 as active etch gases.
Knowledge obtained from reviewing the literature is applied in chapter 3
where the design of a dual plasma reactor is discussed along with preliminary
deposition results of the polymer and ceramic layers. The synthesis and
characterisation of nanometre sized silica particles for incorporation into polymer
layers is discussed in chapter 4. Particles formed and trapping in an electric field
are monitored using in-situ FTIR. Ex-situ characterisation is performed using XPS
and ESEM and the formation of macroscopic structures on the electrode is studied
with the aid of fluid dynamic models. An in depth look at the intrinsic kinetics of
the TEOS/O2 plasma system is presented in chapter 5 with a view to determining
mechanisms for the formation of SiO2 particles/film and gaseous by-products of the
process.
Particles synthesized in the low pressure discharge turned out to be porous and
offered little to no mechanical enhancement of the hybrid layers. However, reports
of nano-porous materials for as dielectric materials prompted a study into
synthesizing porous layers for low-k applications. A twin capacitive plasma system
was developed for this and is the subject of chapter 6. The thesis is concluded in
chapter 7.
19
Chapter 1
References
[1] Veprek S., Thin Solid Films, 297, 145-153, (1997).
[2] Karvankova P., Veprek-Heijman M.G.J., Zindulka O., Bergmaier A., Veprek S., Surface
and Coatings Technol., 163-164, 149-156, (2003).
[3] Hadjadi A., Beorchia A., Roca I Cabarrocas P., Boufendi L., Thin Solid Films, 403-404,
139-143, (2002).
[4] Huet S., Viera G., Boufendi L., Thin Solid Films, 403-404, 193-196, (2002).
[5] Wu G., Wang J., Shen J., Yang T., Zhang Q., Zhou B., Deng Z., Bin F., Zhou D., Zhang.,
J. Non-Cryst. Solids, 275, 169-174, (2000).
[6] Haereid S., Dahle M., Lima S., Einarsrud M.-A., J. Non-Cryst. Solids, 186, 96-103,
(1995).
[7] Chaudhury N.K., Bhardwaj R., Murari B.M., Current Appl. Phys., 3, 177-184, (2003).
[8] Yu M., Lin J., Wang S.B., J. Alloys Comp., 344, 212-216. (2002).
[9] Medda S.K., Kundu D., De G., J. Non-Cryst. Solids, 318, 149-156, (2003).
[10] Sol-Gel Science: The physics and chemistry of sol-gel processing, Brinker J., Scherer
G.W., Academic Press, (1990).
[11] Innocenzi P., Brusatin G., Guglielmi M., Bertani R., Chem. Mat., 11, 1672-1679, (1999).
[12] Matejka L., Dukh O., Brus J., Simonsick-Jr. W.J., Meissner B., J. Non-Cryst. Solids,
270, 34-47, (2000).
[13] Noell J.L.W., Wilkes G.L. Mohanty D.K., J. Appl. Polym. Sci., 40, 1177-1194, (1990).
[14] Laczka M. Cholewa-Kowalska K., Kogut M., J. Non-Cryst. Solids, 287, 10-14, (2001).
[15] Frohlich L., Houbertz R., Jacob S., Popall M., Mueller-Fiedler R., Graf J., Munk M., von
Zychlinski H., Mat. Res. Soc. Symp. Proc., 726, 349-354, (2002).
[16] Polmanteer K.E., Lentz C.W., Rubber Chem. Technol., 48, 795, (1975).
[17] Dannenberg E.M., Rubber Chem. Technol., 48, 410, (1975).
[18] Wagner M.P., Rubber Chem. Technol., 49, 703, (1976).
[19] Kraus G., Rubber Chem. Technol., 51, 297, (1978).
[20] Boonstra B.B., Polymer, 20, 691, (1979).
[21] The Chemistry of Silica, Iler R.K., Wiley, New York, (1979).
[22] Alexandre M., Dubois P, Mater. Sci. Eng., 28, 1-63, (2000).
[23] Mohallen N.D.S., Seara L.M., Appl. Surf. Sci., 214, 143-150, (2003).
[24] Yano S., Nakamura K., Kodomari M., Yamauchi N., J. Appl. Polym. Sci., 54, 163-176,
(1994).
[25] Yano S., Iwata K., Kurita K., Mater. Sci. Eng., C, 6, 75-90, (1998).
[26] Ning Y.-P., Tang M.-Y., Jiang C.-Y., Mark J.E., J. Appl. Polym. Sci., 29, 3209-3212,
(1984).
[27] Sohoni G.B., Mark J.E., J. Appl. Polym. Sci., 45, 1763-1775, (1992).
[28] Landry C.J.T., Coltrain B.K., Landry M.R. Fitzgerald J.J., Long V.K., Macromolecules,
26, 3702-3712, (1993).
[29] Mauritz K.A., J. Appl. Polym. Sci., 40, 1401-1420, (1990).
[30] Suzuki F., Onozato K., Kurokawa Y., J. Appl. Polym. Sci., 39, 371-381, (1990).
[31] Bauer F., Sauerland V., Ernst H., Glasel H.-J., Naumov S., Mehnert R., Macromol.
Chem. Phys., 204, 375-383, (2002).
[32] Tsai M-H., Whang W-T., Polymer, 42, 4197-4207, (2001).
[33] Li C., Jordens K., Wilkes G.L., Wear, 242, 152-159, (2000).
[34] Urreaga J.M., Matias M.C., Lorenzo V., de la Orden M.U., Mat. Lett., 45, 293-297,
(2000).
20
Chapter 1
[35] Haas K.-H., Amberg-Schwab S., Rose K., Schottner G., Surf. Coat. Technol., 111, 72-79,
(1999).
[36] Jeng S.P., Havemann R.H., Chang M.C., Mater. Res. Soc. Symp. Proc., 35, 337, (1994).
[37] Ryan J.G., Geffken R.M., Poulin N.R., Paraszczak J.R., IBM J. Res., Dev., 39, 371,
(1995).
[38] Liu P.T., Chang T.C., Sze S.M., Pan F.M., Mei Y.J., Wu W.F., Tsai M.S., Dai B.T.,
Chang C.Y., Shih F.Y., Huang H.D., Thin Solid Films, 332, 345-350, (1998).
[39] Baklanov M.R., Mogilnikov K.P., Microelectron. Eng., 64, 335-349, (2002).
[40] Xu J., Moxom J., Yang S., Suzuki R., Ohdaira T., Appl. Surf. Sci., 194, 189-194, (2002).
[41] Chang T.C., Tsai T.M., Liu P.T., Mor Y.S., Chen C.W., Mei Y.J., Sheu J.T., Tseng T.Y.,
Thin Solid Films, 420-421, 403-407, (2002).
[42] Uchida Y., Katoh T., Oikawa M., Mater. Sci. Semicond. Process., 5, 259-264, (2003).
[43] Jain A., Rogojevic S., Ponoth S., Agarwal N., Matthew I., Gill W.N., Persans P.,
Tomozawa M., Plawsky J.L., Simonyi E., Thin Solid Films, 398-399, 513-522, (2001).
[44] Kim R.-H., Park H.-H., Hyun S.-H., Yeom G.-Y., Thin Solid Films, 332, 444-448,
(1998).
[45] Liu P.-T., Chang T.C., Hsu K.C., Tseng T.Y., Chen L.M., Wang C.J., Sze S.M., Thin
Solid Films, 414, 1-6, (2002).
[46] Creatore M., Barrell Y., Kessels W.M.M., van de Sanden M.C.M., Mat. Res. Soc. Symp.
Proc., 766, (2003).
[47] Shaniryan D., Weidner K., Gray W.D., Baklanov M.R., Vanhaelemeersch S., Maex K.,
Microelectron. Eng., 64, 361-366, (2002).
[48] Borvon G., Goullet A., Mellhaoui X., Charrouf N., Granier A., Mater. Sci. Semicond.
Process., 5, 279-284, (2003).
[49] Si J.-J., Show Y., Banerjee S., Ono H., Uchida K., Nozaki S., Morisaki H.,
[50] Schulz S.E., Koerner H., Murray C., Streiter I., Gessner T., Microelectron. Eng., 55, 4552, (2001).
[51] Iacopi F., Tokei Zs., Stucchi M., Brongersma S.H., Vanhaeren D., Maex K.,
[52] Volinsky A.A., Vella J.B., Gerberich W.W., Thin Solid Films, 429, 201-210, (2003).
[53] Travaly Y., Eyckens B., Carbonel L., Rothschild A., Le Q.T., Brongersma S.H., Ciofi I.,
Struyf H., Furukawa Y., Stucchi M., Schaekers M., Bender H., Rosseel E.,
Vanhaelemeersch S., Maex K., Gaillard F., Van Autryve L., Rabinzohn P.,
[54] Lanckmans F., Maex K., Microelectron. Eng., 60, 125-132, (2002).
[55] Lee H.J., Oh K.S., Choi C.K., Surf. Coat. Technol., 171, 296-301, (2003).
[56] Jehn H.A., Surf. Coat. Technol., 112, 210-216, (1999).
[57] Zhang J.L., Yu S.J., Ma T.C., Vacuum, 65, 327-333, (2002).
[58] Bulkin P., Bertrand N., Drevillon B., Thin Solid Films, 296, 66-68, (1997).
[59] Boswell R.W., Chen F.F., IEEE T. Plasma Sci., 25, (1997).
[60] Hayden D.B., Juliano D.R., Neumann M.N., Allain M.C., Ruzic D.N., Surf. Coat.
Technol., 120-121, 401-404, (1999).
[61] Thyen R., Weber A., Klages C.-P., Surf. Coat. Technol., 97, 426-434, (1997).
[62] Kessels W.M.M., Severens R.J., Smets A.H.M., Korevaar B.A., Adriaenssens G.J.,
Schram D.C., van de Sanden M.C.M., J. Appl. Phys., 89, (2001).
[63] Glow Discharge Processes, Chapman B.N., Wiley & Son, New York, (1980).
[64] Principles of plasma discharges and materials processing, Lieberman M.A., Lichenberg
A.J., Wiley, New York, (1999).
21
Chapter 1
[65] Garbassi F., Occhiello E., Macromol. Symp., 139, 107-114, (1999).
[66] Mutikainen R., Thin Solid Films, 238, 248-257, (1994).
[67] Schuster F., Peit G., US patent Nr. US5437725, (1995).
[68] Nguyen V.S., Underhill J., Fridmann S., Pan P., J. Electrochem. Soc.: Solid-State Sci.
Technol., 132, 1925-1932, (1985).
[69] Shirafuji T., Miyazaki Y., Nakagami Y., Hayashi Y., Nishino S., Jap. J. Appl. Phys., 38,
4520-4526, (1999).
[70] Gengenbach T.R., Griesser H.J., Polymer, 40, 5079-5094, (1999).
[71] Alexander M.R., Short R.D., Jones F.R., Michaeli W., Blomfield C.J., Appl. Surf. Sci.,
137, 179-183, (1999).
[72] Grundmeier G., Stratmann M., Bunsenges B., Phys. Chem., 11, 1387-1392, (1995).
[73] Fracassi F., dAgostino R., Favia P., J. Electrochem. Soc., 139, 2636-2643, (1992).
[74] Plasma surface modification and plasma polymerisation, Inagaki N., Technomic
Publishing Company Inc., (1996).
[75] Korzec D., Traub K., Werner F., Engemann J., Thin Solid Films, 281-282, 143-145,
(1996).
[76] Morra M., Occhiello E., Garbassi F., J. Appl. Polym. Sci., 48, 1331-1340, (1993).
[77] Agres L., Segui Y., Delsol R., Raynaud P., J. Appl. Polym. Sci., 61, 2015-2022, (1996).
[78] Taylor C.E., Boerio F.J., Ward S.M., Ondrus D.J., Dickie R.A., Brutto M.M., J. Adhes.,
69, 237-261, (1999).
[79] Taylor C.E., Boerio F.J., J. Adhes., 69, 217-236, (1999).
[80] Menichella S., Misiano C., Simonetti E., De Carlo L., Carrabino M., Society of Vacuum
Coaters, 37th Annual Technical Conference Proceedings, 37-40, (1994).
[81] Poll H.-U., Meichsner J., Arzt M., Friedrich M., Rochotzki R., KreyBig E., Surf. Coat.
Technol., 59, 365-370, (1993).
[82] Kusabirake M., Aozasa M., Thin Solid Films, 281-282, 533-535, (1996).
[83] Fujii T., Hiramatsu M., Nawata M., Thin Solid Films, 343-344, 457-460, (1999).
[84] Yu Q.S., Yasuda H.K., J. Polym. Sci.: Part A: Polym. Chem., 37, 1577-1587, (1999).
[85] Bonnar M.P., Wilson J.I.B., Burnside B.M., Reuben R.L., Gengenbach T.R., Griesser
H.L., Beamson G., J. Mater. Sci., 33, 4843-4855, (1998).
[86] Turner R.H., Segall I., Boerio F.J., Davis G.D., J. Adhes., 62, 1-21, (1997).
[87] Creatore M., Barrell Y., Kessels W.M.M., van de Sanden M.C.M., Mat. Res. Soc. Symp.
Proc., 766, (2003).
[88] High Rate Plasma Deposition of Silicon Oxide Like Films, van Hest M., PhD Thesis,
Technische Universitiet Eindhoven, (2002).
[89] Takai O., Hozumi A., Sugimoto N., J. Non-Cryst. Solids, 218, 280-285, (1997).
[90] Klug W., Gegenwart R., Herrmann R., Schneider R., SPIE Hard Mater. in Optics, 1275,
135-141, (1990).
[91] Nault M., Weidman T., Sugiarto D., Mui D., Lee C., Yang J., J. Vac. Sci. Technol. B, 16,
3730-3733, (1998).
[92] Samson F., Surf. Coat. Technol., 81, 79-86, (1996).
[94] Metroline website http://www.metrolineindustries.com.
[95] US Patent 5,298,587, issued March 29, (1994).
[96] US Patent 5,320,875, issued June 14, (1994).
[97] US Patent 5,433,786, issued July 18, (1995).
[98] US Patent 5,494,712, issued February 27, (1996).
[99] Sawa G., Yamanaka S., Nakamura S., Yamaguchi S., Jap. J. Appl. Phys., 20, 201-204,
(1981).
22
Chapter 1
[100] Bui A., Carchano H., Sanchez D., Thin Solid Films, 13, 207-211, (1972).
[101] Ooij W.J. van, Sabata A., Zeik D.B., Taylor C.E., Boerio F.J., Clarson S.J., Symposium,
6-42, (1993).
[102] Kamo H., Yudasaka M., Kurita S., Matsui T., Kikuchi R., Ohki Y., Yoshimura S.,
Synth. Met., 68, 61-63, (1994).
[103] Schue F., Clarotti G., A. Ait Ben Aoumar, Sledz J., Mas A., Geckeler K.E., Gopel W.,
Orsetti A., Marcromol Reports, A31, 1161-1175, (1994).
[104] Clarotti G., A. Ait Ben Aoumar, Schue F., Sledz J., Macromol. Chem., 192, 2581-2590,
(1991).
[105] Vanhulsel A., Dekempeneer E., Wegener W., Meneve J., Smeets J., Soc. Vac. Coat.
42nd Annual Technical Conference Proceedings, ISSN 0737-5921, (1999).
[106] Lin Y., Yasuda H., Miyama M., Yasuda T., J. Poly. Sci.: Part A: Polym. Chem., 34,
1843-1851, (1996).
[107] Stuart M., Proc. IEE., 112, 1614-1616, (1965).
[108] Yasuda H., Hirotsu T., J. Polym. Sci.: Polym. Chem. Ed., 16, 313-326, (1978).
[109] Munro H.S., Till C., J. Polym. Sci.: Polym. Chem. Ed., 24, 279-286, (1986).
[110] US Patent, 5,635,245, issued June 3 (1997).
[111] World Patent, WO 95/24275, issued September 14, (1995).
[112] Yasuda H., Hirotsu T., J. Polym. Sci.: Polym. Chem. Ed., 16, 743-759, (1978).
[113] Plasma surface modification and plasma polymerization, Inagaki N., Wiley, New York,
(1996).
[114] Yasuda H., J. Polym. Sci.: Macromol. Rev., 16, 199-293, (1981).
[115] Staring E.G.J., Braun D., Rikken G.L.J.A., Demandt R.J.C.E., Kessener Y.A.R.R.,
Bouwmans M., Broer D., Synth. Met., 67, 71-75, (1994).
[116] Erjavec J., Sikita J., Beaudoin S.P., Raupp G.B., Mater. Lett., 39, 339-342, (1999).
[117] Hanyaloglu B., Aydinli A., Oye M., Aydi E.S., Appl. Phys. Lett., 74, 606-608, (1999).
[118] Senkevich J.J., Desu S.B., Chem. Mater., 11, 1814-1821, (1999).
[119] Hu S.C., Chuah Y.K., Yen M.C., Building and Environment, 37, 201-208, (2002).
[120] Fujimoto T., Okuyama K., Yamada S., Adachi M., J. Appl. Phys., 85, 4196-4206,
(1999).
[121] Whitby E., Hoshino M., J. Electrochem. Soc., 143, 3397-3404, (1996).
[122] Bai W., Choy K.L., Stelzer N.H.J., Schoonman J. Solid State Ionics, 116, 225-228,
(1999).
[123] Botti S., Celeste A., Coppola R., Appl. Organometallic Chem., 12, 361-365, (1998).
[124] Cannon W.R., Danforth S.C., Flint J.H., Haggerty J.S., Marra R.A., J. Am. Ceram. Soc.,
65, 324-340, (1982).
[125] CVD Techniques for the synthesis or modification of porous ceramics, Dekker J.P., PhD
Thesis, Technische Universitiet Eindhoven, (1994).
[126] Skandan G., Nanostruct. Mater., 5, 111-126, (1995).
[127] Yuan F.L., Chen C.H., Kelder E.M., Schoonman J., Solid State Ionics, 109, 119-123,
(1998).
[128] Botti S., Celeste A., Appl. Phys. A., 67, 421-424, (1998).
[129] Liu B., Gu H., Chen Q., Mat. Chem. Phys., 59, 204-209, (1999).
[130] Ying J.Y., J. Aerosol Sci., 24, 315-338, (1993).
[131] Okuyama K., Fujimoto T., Hayashi T., AIChE J.: Ceramics Process., 43, 2688-2697,
(1997).
[132] Tandon P., Rosner D.E., AIChE J., 42, 1673-1684, (1996).
[133] Dusty Plasmas, Bouchoule A., Wiley, New York, (1999).
[134] Stoffels E., Stoffels W.W., Kroesen G.M.W., Hoog F.J.de., J.Vac. Sci. Technol., A 14,
556-561, (1996).
23
Chapter 1
[135] Boufendi L., Hermann J., Bouchoule A., Dubreuil B., Stoffels E., Stoffels W.W., Giorgi
M.L. de., J. Appl. Phys., 76, 148-153, (1994).
[136] Cabarrocas P.R.I.,Gay P., Hadjadj A., J. Vac. Sci. Technol. A., 14, 655-659, (1996).
[137] Dorier J.-L., Hollenstein Ch. Howling A.A., J. Vac. Sci. Technol. A., 13, 918-926,
(1995).
[138] Watanabe Y., Shiratani M., Plasma Sources Sci. Technol., 3, 286-291, (1994).
[139] Hollenstein Ch., Howling A.A., Courteille C., Magni D., Scholz S.M., Kroesen
G.M.W., Simons N., Zeeuw W de., Schwarzenback W., J. Phys. D: Appl. Phys., 31, 7484, (1998).
[140] Courteille C., Magni D., Deschenaux Ch., Fayet P., Howling A.A., Hollenstein Ch.,
Soc. Vac. Coat. 41st Annual Technical Conference proceedings, (1998).
(1999).
24
Chapter 2
Anisotropic plasma etching of SiCxOyHz thin films
Chapter 2: Anisotropic plasma etching of

SiCxOyHz thin films
Abstract
This chapter looks at the characterization of a remote helicon plasma
discharge using a Langmuir probe as part of a feasibility study for
etching silicone like thin films. Etching is done using SF6, CF4, Ar and
O2 or mixtures thereof. The plasma reactor is characterised using a
Langmuir probe with a planar nickel tip and used to measure ion
currents and floating potentials for the various etch gases used. Etch
rates and etch profiles are determined using SEM images and step
profile measurements.*
2.1 Introduction
The potential ability to tailor the chemical and mechanical properties of hybrid
materials makes them very attractive for a large number of applications, many of
which were discussed in chapter 1. Applications such as micro electro mechanical
systems (MEMS) [1], low-k dielectrics [2] and optical waveguides [3], however,
require post deposition processing of the hybrid thin films in order to integrate
them into these applications. One important post deposition process is anisotropic
reactive ion etching (RIE) [4, 5] and is currently the only technique that offers
chemically selective, anisotropic etching at sub-micron dimensions. RIE involves
the ionization of selected etch gases to form reactive ions and radicals such as
fluorine or oxygen. The reactive species react with the substrate surface to form
volatile by-products that leave the surface and are pumped away. For Si and SiO2
this process is well understood and aspect ratios of 10:1 are readily achieved in
industrial processes [6]. The etching gas mixture however, is specific for the
material being processed and etch rate and etch profile are critically dependant on
the plasma conditions and gases used.
In this study, a remote, inductively coupled r.f. helicon discharge is evaluated
as a means of etching high aspect ratio nanometre size holes in organically
modified ceramic thin films of SiCxOyHz composition. The plasma is characterised
using a Langmuir probe and various process and plasma parameters are correlated
to the etch rates. Etching is performed using either one, or combinations of SF6,
CF4 and O2 as reactive gases. Etch rates and profiles are controlled by means of
varying the discharge power in the helicon source and r.f. bias applied to the
substrate in addition to the overall gas composition. Preliminary results of etch
*
This work was performed at the Australian National University in Canberra as part of a
three month traineeship.
25
Chapter 2
profiles and substrate degradation are investigated using scanning electron

microscopy (SEM).
2.1.1 Reactive ion etching
Silicon, like other materials commonly used in the manufacture of micro electronic
devices, forms volatile compounds when reacted with the halide elements. RIE
processes take advantage of this and etching of Si and SiO2 thin films is achieved
by forming the respective silicon fluoride, or chloride. Fluorine is used almost
exclusively over chlorine in RIE processes as it has a higher reactivity and as
chlorine has the tendency to form hydrochloric acid that corrodes metal surfaces in
the reactor system and pumps. Fluorine is introduced into the reactor in the form of
short chain fluorocarbons (CFC) such as CF4 [7, 8], CF3H [9] and/or C2F6 [10] but
gases such as SF6 [11, 12] and NF3 [13] are also widely used. Hydrogen is added to
the etch chemistry (in the form of H2 or as a partially hydrogenated CFC such as
CF3H) for etching SiO2 to facilitate the formation of H2O and enhance the etch rate.
In contrast, the addition of small amounts of oxygen to the Si etch chemistry is
observed to enhance the etch rate as it oxidized the residual carbon layer resulting
from fluorocarbon decomposition. Excess oxygen however, can cause oxidation of
the photo resist mask and react with Si to form SiO2, retarding the etch rate. Figure
2.1 depicts some of the processes involved in the RIE of SiO2 using an CF4/H2 etch
gas mixture. The standard etch model begins with dissociation of the gas via
electron impact and ion bombardment of the substrate surface. The substrate
surface is activated by the ion bombardment and dangling bonds are formed.
These dangling bonds react with fluorine and oxygen species in the gas to form
volatile compounds. Volatile compounds formed at the surface finally desorb and
are pumped away.
CF4 + e CF3 + F + e
Plasma activation
& H2 + e H + H + e
H2O
SiF4
SiF4
Mask
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000H
0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00SiF
00 00 00 00 00 00 00 00200 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00F00 00 00 00 00 00 00 000 000O
0000000000000000000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 O
00 00 00 00 00 00 00 00 00 00 00 00 00O
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SiO
0 0 0 0 0 0 0 0 020 0 0 Substrate
0000000000000000000000000000000000000000000000000000000000000000000000000000000000000
Figure 2.1: Anisotropic reactive ion etching (RIE) of a silica

substrate. Fluorine reacts to form volatile SF4 gas, while H2
reacts to remove oxygen.
26
Chapter 2
The presence of both oxygen and carbon in silicone-based hybrid thin films
complicates the plasma etch chemistry. Hydrocarbon materials, such as photo resist
masks, and oily contaminations can be removed using oxygen plasma [14, 15].
Several authors have investigated the effect of oxygen plasmas on SiOxCyHz
coating prepared using the sol-gel process [16, 17]. The magnitude of the overall
effect of oxygen plasma treatment on these materials varies depending on the
particular material density and structure. Impinging oxygen species react with the
film to remove carbon and hydrogen moieties causing a change in the chemical
composition. Ion bombardment also puts energy into the film and prolonged
exposure causes film densification. Brief exposure however, has also been
observed to give rise to the formation of silanol groups in the film [18].
In RIE processes the anisotropy of the etch profile is predominantly controlled
by the high directionality of the impinging reactive ions. This condition is readily
formed in parallel plate capacitively coupled reactor configurations, where the
substrate is located on the driven electrode. In this configuration a net negative d.c.
bias potential forms between the plasma and the electrode. Positive ions traversing
this sheath are accelerated towards the substrate table and ion bombardment of the
substrate occurs. In addition to increasing anisotropy of the etch profile and etch
rate, r.f. biasing also subjects the substrate to higher heat load. By coupling more
power into the discharge, higher reactive ion concentrations can be generated and
the etch rate increased. Unfortunately, even at relatively low power levels (and
therefore low ion density) high potentials are reached which can cause damage to
sensitive substrates. In order to couple more power into the plasma and increase
etch rate, it is necessary to actively cool the substrate. This removes some of the
heat generated by the ion bombardment but is difficult to implement and does not
compensate implantation and sputtering effects that occur at very high ion energies.
Capacitively coupled plasmas inherently exhibit much lower plasma densities
than a corresponding inductive system. This is due to the high sheath potentials
accelerating ions to higher energy before they collide with the electrode. As energy
is used to accelerate ions through the sheath, less energy is available for ionisation
and therefore the plasma density is lower than the corresponding inductive
discharge. An ideal etch process is therefore governed by two criteria,
1. High plasma density to generate high concentration of reactive species.
2. Control of sheath potential above substrate to control ion energy.
Still higher plasma densities can be achieved by applying an axial magnetic
field to an inductive plasma discharge. Axial magnetic fields increase the plasma
density in two ways. As free electrons in the discharge cannot cross the magnetic
field lines, electron losses to the reactor walls are reduced. The electron
confinement results in a slight increase in the plasma density. However, more
significantly the magnetic confinement enables waves to be launched in the plasma
that provide additional electron heating mechanisms and couple energy into the
discharge.
One successful magnetised plasma system is the helicon plasma [19, 20]. The
helicon plasma source is comprised of a quart tube surrounded by a two loop
27
Chapter 2
antenna diametrically placed on the outside of the source tube surrounded by

confining magnetic coils. R.f. power supplied to the antenna produces a transverse
r.f. magnetic field perpendicular to both the tube axis and a constant axial magnetic
field. This r.f. field excites the m = 1 azimuthal mode of a helicon wave in the
source tube. Energy is transferred from this wave to the plasma electrons via a
mechanism known as Landau damping [21, 22, 23]. The additional electron
heating, provided by the Landau damping mechanism, enables the helicon plasma
sources to achieve plasma densities considerably higher than conventional
inductive discharges. When used in etch applications [24, 25], the helicon source is
typically placed above a diffusion chamber housing a substrate table. In this
configuration the reactive ion concentration is determined by the helicon power,
without generating high electric fields above the substrate. The addition of a
separate r.f. bias to the substrate table is used to increase both the etch rate and
achieve an anisotropic etch profile.
2.2 Experimental
The helicon etch reactor consists of two sections, a source and a diffusion chamber
(Fig. 2.2). The helicon source consists of a glass tube 15 cm in diameter and 30 cm
long surrounded by the helicon antenna and two solenoids. A 13.56 MHz electric
field is applied to the helicon antenna launching a plasma in the source that
expands into a 35 cm diameter, 30 cm long aluminium diffusion chamber that is
also surrounded by two solenoids. The substrate holder is situated at the bottom and
in the centre of the diffusion chamber directly beneath the helicon plasma source.
The water cooled substrate table is designed to accept 4 inch silicon wafers and is
connected to a 13.56 MHz power supply to enable r.f. biasing of the substrate.
Heating of the substrate, caused by ion bombardment, is transferred to the water
cooled substrate holder via a helium gas pocket between the silicon substrate and
cooled substrate holder. The reactor vessel is maintained at a base pressure of 10-6
Torr and substrates are entered and removed via a load lock system.
All gas flows are controlled using mass flow controllers (MKS) and are fed
into the reactor via an inlet in the centre of the diffusion chamber. Experiments are
performed using the gases SF6, CF4, O2 and Ar, or mixtures thereof, which are
exhausted, along with reaction by-products, via the turbomolecular pump situated
above the source. Further descriptions and details of this reactor configuration can
be found elsewhere [26, 27, 28, 29, 30].
28
Chapter 2
antenna (outside vacuum)
400l/s
turbomolecular
pump
Matching network
and source cooling
source
solenoid
RF supply
ENI
OEM 25
Ni Probe
Gas inlet
Chamber solenoid
Phase locked to
Source supply
RF supply
ENI
ACG 5
wafer cartridge
Load lock and cartridge
transfer mechanism
150l/s
turbo
Matching
network
RF bias
Water cooling
to table
He
Figure 2.2: Schematic of helicon etch machine.

Suitable masks for plasma etch processes are usually created using
lithographic techniques, however, due primarily to time constraints a more robust
and quicker masking method was needed. Mask requirements were fulfilled by
drilling 315 m holes into a 125 m thick nickel plate. This provided a mask that
was robust and virtually unaffected by the reactive plasma gases and could
therefore be transferred and used for all samples. The substrate and mask set-up
have been sketched in Fig. 2.3. A silicon wafer coated with a hybrid film is cut into
six sections and each section is sandwiched between the nickel mask and a steel
substrate table using two steel springs.
29
Chapter 2
Steel springs
Ni mask
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
IPG coated Si
wafer
Steel substrate holder
Figure 2.3: Schematic of substrate holder housing the substrate and nickel mask.
The organically modified ceramics used in this study are photo-initiated
silicone polymer materials primarily designed for optical waveguide applications
by Redfern Polymer Optics. A schematic of IPG is drawn in Fig. 2.4.
R
Si O
Si O Si O
R
Si O Si O Si O
R
Where R is a
polymerisable
side chain e.g.
or
Figure 2.4: Schematic of inorganic polymer glass.

2.2.1 Ion saturation current measurements
When an insulated electrostatic probe is inserted into a gas discharge it will be
bombarded with ions and electrons until it picks up a net potential equal to the
plasma floating potential. If the tip of the probe is biased (VB) in relation to the
plasma potential, it will either attract or repel positive ions or electrons depending
if the probe is biased positively or negatively. A typical Langmuir probe
measurement is depicted in Fig. 2.5. Here the current drawn from the plasma by the
probe has been plotted as a function of the probe potential.
Langmuir probe characteristics are most easily described by assuming the
probe is measuring a low pressure argon discharge. At large and negative bias, the
probe will repel all electrons and a positive ion current is drawn. As the bias
approaches a zero potential the high energy electrons are able to overcome the
negative bias and strike the probe tip. This leads to a transition period where the
probe is collecting both ions and electrons. During this transition the electron and
ion fluxes generate opposite currents and results in net current of zero,
corresponding to the floating potential of the plasma.
30
Chapter 2
Electron collection
-70
100 VB
Electron & ion collection

Positive ion collection
Figure 2.5: Typical I-VB characteristic for a
Langmuir probe. Not to scale.
In this study the Langmuir probe was used in two configurations. In one
configuration the probe tip was biased to a potential of 75 V (with respect to the
reactor ground) and used as an ion probe. At this large negative potential the probe
is assumed to repel all electrons and draw current equal to the ion saturation
current. From standard texts [31], assuming a Maxwellian electron energy
distribution and an ion velocity equal to the Bohm velocity (B) [m.s-1], the ion
current density (j+) at the probe and substrate is,
k T
j + = 0.6n e B e
mi
(2.1)
where Te is the electron temperature [K], kB the Boltzman constant [JK-1], and mi
the ion mass [kg]. Equation (2.1) is only applicable to discharges where all of the
negative charge is carried by the electrons.
The presence of negative ions significantly changes the plasma structure to the
point where the standard treatment of probe data is no longer valid. When the
plasma contains negative ions, a decrease in current in the electron collection
region is observed at high positive probe bias. The electron and ion current regions
become distorted and the Bohm criterion is modified as the negative ion density (n)
increases. Variation in the Bohm criterion as a function of the n/n+ ratio [32], alter
the relationship between the positive ion current and n+. In addition to these
complications, a recent publication by Franklin et al. [33] reports further
inconsistency in the standard treatment of probe data for negative ion plasmas and
even suggests such plasmas are structured consisting of a positive/negative ion core
surrounded by an electron/positive ion outer region. A detailed review of probe
measurements in plasmas containing negative ions is not given, as the probe is
31
Chapter 2
setup to measure the floating potential and ion current at high negative bias
potential only. In this configuration the probe provides insufficient information
upon which to start characterising the plasma. Instead, results of the probe
measurements are used in a qualitative capacity where ion flux and potentials
experienced by the probe are assumed to be similar to those experienced by the
substrate during plasma etching.
The Langmuir probe is inserted into the plasma and a potential, VB, of -75 V
applied to the tip (to collect positive ions only). The probe collection area consisted
of a flat nickel disc with a total surface area A = 2.9510-5 m2. Ion currents are
measured by the potential drop across a 10 k resistor placed between the probe tip
and ground (cf. Fig. 2.6).
20 V
-75V
Ni probe tip
10k
Figure 2.6: Langmuir probe set up at -75 V bias supplied by eight 9 V

batteries connected in series.
2.3 Results
Power is coupled into an inductive discharge by the r.f. field heating the electrons
and ions [34]. The hot electrons impart energy to the neutrals via ionising
collisions, which also release photons. Ions are accelerated towards the walls due to
the potential across the sheath and collide with the walls where upon they become
neutralised and impart their kinetic energy to the walls in the form of heat. In a
global scheme, energy is put into the plasma via the electric field heating the
electrons, and is lost via light emission (from ionising collisions) and heat (via ion
bombardment of walls). As neutral gas molecules collide with the walls they pick
up some of the heat deposited by the ions.
Gases injected into the reactor are removed via the turbomolecular pump
located on top of the helicon source. The increased velocity in which the gas
molecules collide with the system walls, in conjunction with a decrease in the
pumping efficiency with increased gas temperature result in an increased system
pressure, as can be seen in Fig. 2.7. The slight pressure increase seen for argon is
attributed solely to gas heating, whereas CF4 and SF6 both show larger pressure
increases at higher helicon power inputs. The higher pressures seen for the
molecular gases are assumed to be caused by an increase in the effective gas flow
32
Chapter 2
Pressure (mTorr)
caused by fragmentation of the gas. This argument is supported by the fact that the
pressure increases are larger as the molecularity of the gas increases.
10
SF6
O2
CF4
Ar
6
4
2
0
0
200
400
600
800
1000
1200
Discharge power (W)
Figure 2.7: Pressure variations in the helicon discharge as a function

of discharge power.
2.3.1 DC bias measurements
The application of an r.f. voltage to the substrate table produces a d.c. field above
the substrate provided the gas in this region is partly ionised. The relationship
between the applied r.f. field and resulting d.c. bias was measured for the helicon
etch machine and is presented in Fig. 2.8.
Different self-bias potentials are observed at the substrate depending on
whether the bias is applied with or without the helicon plasma. At the same bias
power input, the ion current to the biased electrode is increased due to the
additional density generated by the helicon, and consequently the bias potential is
reduced. If the ion current to the biased substrate remains constant, then the
potential will scale linearly with the applied power. However, if the additional bias
creates additional density above the biased substrate, the ion current will increase
and the potential will scales with the P1/2 (Fig 2.8). The bias potential is seen to
scale linearly with the applied biasing power suggesting that the plasma density
does not increase with increasing bias power. The high biasing potentials observed
are believed to be a consequence of the small surface area of the substrate table in
relation to the grounded area of the steel expansion chamber.
33
Chapter 2
With helicon
Without helicon
Bias potential (-V)
10
Vbias ~ P
100
1/2
Vbias ~ P
1000
10
100
Bias power (W)
Figure 2.8: Negative bias as function of power applied to

substrate table.
A Langmuir probe was used to measure ion currents and floating potentials in
the centre of the diffusion chamber 18 cm above the substrate for each of the pure
process gases at different pressures. In this configuration the probe measures the
ion current (Fig 2.9) carried by the positive ions and as the plasma is assumed to be
quasi neutral, n+ = ne. This criterion however, does not hold if appreciable
concentrations of negative ion are present, in which case n+ = ne + n. Negative ions
will shield the probe tip from some of the positive ions resulting in lower values of
n+. In this situation the positive ion saturation current may well be more
representative of ne then of n+. As argon is the most electro-positive of the gases, it
produced the highest plasma density of positive ion with an estimated value of n+ =
1.21017 m-3 (assuming n = 0) The more electro negative gases all result in low
currents to the probe tip suggesting that there are either fewer positive ions
generated, and that a portion of the positive ions are being shielded from the probe
by negative ions. The most electro-negative element present is fluorine, and as SF6
contained the most fluorine atoms it also produces the lowest positive ion current.
Oxygen and CF4 generated comparable positive ion currents. For the argon
discharge, plasma density scaled linearly with respect to discharge power, which is
typical behaviour for a monatomic gas in an inductive discharge. The molecular
gases O2, CF4 and SF6 all exhibited a jump in plasma density at between 100-200
W. This is attributed to a rapid increase in fragmentation as more power is coupled
into the discharge and an associated transition from capacitive to inductive
operation. This transition is more prominent at low pressures and was also observed
in an argon discharge at > 1 mTorr.
34
Chapter 2
Ar
Probe current (mA)
2.5
CF4
O2
2.0
SF6
1.5
1.0
0.5
0.0
200
400
600
800 1000
Helicon discharge power (W)
1200
Figure 2.9: Positive ion current as a function of helicon discharge power.

The floating potential of argon and oxygen discharges as a function of helicon
power is plotted in Fig. 2.10. A negative and constant floating potential was seen to
develop at a value of approximately -4 V in the argon discharge. The floating
potential for the oxygen discharge was seen to be negative, and become more
negative as the helicon power was increased.
The relationship between floating potential and helicon power is more
complicated for the CF4 and SF6 discharges. In the CF4 discharge, the floating
potential initially becomes increasingly negative as the helicon power increases, but
at approximately 350 W begins to become less negative again (Fig. 2.11).
According to OML theory, the floating potential depends on the ratios of Ti/Te and
mi/me (see Eq. (4.4) and Fig. 4.2 in chapter 4). For argon, Te and Ti remain constant
with increasing power and as no fragmentation can occur, mi/me ratio cannot
change. The floating potential therefore remains constant as the discharge power is
increased. The situation is somewhat more complicated for the molecular gases O2,
CF4 and SF6. Fragmentation of oxygen and the other molecular gases at higher
powers results in a slight reduction of mi. However, little in known about the
plasma potential, and in addition to this, the presence of negative ions may
influence Te, and as the total ion density increases ion-ion and ion-neutral processes
may become significant. All of these factors in combination with other transport
may effect Vf at the probe tip located downstream of the helicon source.
35
Chapter 2
Discharge power (W)

0
200
400
600
800
1000
Floating potential (V)
-2
-4
-6
-8
-10
-12
-14
Ar
O2
-16
-18
Figure 2.10: Floating potential as function of helicon power for

Ar and O2 gases.
Fragmentation and negative ion formation are likely to be even more
prominent processes in the CF4 and SF6 plasma discharges. Explanations for the
behaviour of Vf as a function of helicon power (as shown in Fig. 2.11 and 2.12)
therefore lies in an understanding of the molecular processes governing the
production, transport and loss of these species in the plasma discharge.
Discharge Power (W)
0
200
400
600
800
1000
1200
0
-2
-4
-6
-8
-10
-12
CF4
Figure 2.11: Floating potential as function of helicon power for CF4 gas.
36
Chapter 2
15
SF6
10
5
0
-5
-10
200
400
600
800
1000
1200
Discharge power (W)
Figure 2.12: Floating potential as function of helicon power for SF6 gas.
2.3.2 Etch results
Of the gases and gas mixtures used, SF6 produced the highest etch rates.
Surprisingly the SF6/O2 gas mixture resulted in a lower etch rate than the pure SF6
despite the presence of carbon in the SiOxCyHz layers. Under the conditions used,
CF4 and CF4/O2 plasmas both resulted in deposition on the substrate. A summary of
the etch rates achieved for the various gases is presented in Table 2.1.
Table 2.1: Summary of etch results for different gases. Helicon power
and substrate bias were set to 1000 W and 100 W respectively for all
experiments.
Gas
Ratio (%)
SF6
SF6/O2
Pure SF6
62/38
Etch/Deposition
(nm/min)
524 etch
372 etch
CF4
Pure CF4
303 deposition
CF4/O2
10/90
120 deposition
O2
Pure O2
30 etch
The dependence of etch rate on bias potential and helicon power were also
studied (Fig. 2.13). The etch rate was seen to exhibit a linear dependence to
37
Chapter 2
-750
-600
-450
-300
-150
450
400
350
300
250
200
150
100
50
0
Etch rate (nm/min)
substrate bias potential but a relatively weak dependence on power supplied to the
helicon antenna within the range studied.
450
400
350
300
250
200
150
100
50
0
Bias voltage (V)
100W Bias
300
600
900
1200
1500
Helicon power (W)
Figure 2.13: Etch rate as a function of d.c. bias on substrate (left) and
helicon power (right) for samples etched in a 1.8 mTorr SF6 plasma for
10 minutes.
The bias potential contributes to the ion energy at the substrate surface, and
plasma density directly above. The combination of these effects resulted in a strong
dependence of etch rate with substrate bias for the experimental conditions used in
this study. The helicon plasma is responsible for dissociation of the reactant gas
and generation of ion density in the discharge. The relationship between the etch
rate vs. helicon power therefore is similar to that of helicon power vs. ion saturation
current (cf. Fig. 2.9), although neutral fluorine atoms can also participate in the etch
process.
a.
315 m
b.
226 m
Figure 2.14: SEM images taken of the hole etched in the SiOxCyHz film,
a.) shows a plan view b.) profile view.
38
Chapter 2
Cross sectional and plan view SEM images (Fig. 2.14) taken after the coated
substrate is exposed to a 10 minute SF6 etch plasma indicate that the holes made are
considerably smaller than the holes in the nickel mask and exhibit poor anisotropy.
This is caused by the crude mask arrangement used consisting of a nickel plate held
in place with steel springs. The relatively thick nickel mask (125 m) and poor
contact with the substrate caused shadowing effects near the mask edge.
2.4 Conclusions
Highest etch rates were achieved using pure SF6 gas with typical etch rates in the
order of 500 nm/minute using a helicon power of 1000 W with 100 W substrate
bias. SEM images taken of the substrate after etching indicated a relatively poor
aspect ratio (Fig. 2.14). Poor etch profiles are attributed to the temporary mask
design used, necessitated by time constraints. The nickel shim used to make the
mask was 125 m thick, which is more than 100 times thicker than an ideal mask
made by lithographic techniques. The relatively high sidewalls this mask produced
are likely to have shadowed the substrate from the impinging ion flux and hence
reduced the diameter of the resulting holes etched. The samples also showed a
pebble dashed effect around the edges of the etched hole. This was also attributed
to the poor masking method employed, but also indicated the SiOxCyHz film to be
relatively soft. Etch rate exhibited a greater dependence on d.c. potential formed
from r.f. biasing the substrate then of power applied to the helicon source.
References
[1] Chen J., Liu L., Li Z., Tan Z., Xu Y., Ma J., Sens. Actuators A., 103, 42-47, (2003).
[2] Homma T., Mater. Sci. Eng., R23, 243-285, (1998).
[3] Tien P., Smolinsky G., Martin R., J. Appl. Opt., 11, 637, (1972).
[4] Van der Drift E., Cheung R., Zijlstra T., Microelectron. Eng., 32, 241-253, (1996).
[5] Lang W., Mater. Sci. Eng., R17, 1-55, (1996).
[6] Stern M.B., Microelectron. Eng., 34, 299-319, (1997).
[7] Cho B.-O., Hwang S.-W., Lee G.-R., Moon S.H., J. Vac. Sci. Technol. A, 18, 2791-2798,
(2000).
[8] Balachova O.V., Alves M.A.R., Swart J.W., Braga E.S., Cescato L., Microelectron. J., 31,
213-215, (2000).
[9] Tserepi A., Gogolides E., Cardinaud C., Rolland L., Turban G., Microelectron. Eng., 4142, 411-414, (1998).
[10] Midha A., Murad S.K., Weaver J.M.R., Microelectron. Eng., 35, 99-102, (1997).
[11] Camara N., Zekentes K., Solid-State Electronics, 46, 1959-1963, (2002).
[12] Lishan D.G., Johnson D.J., Lee Y.S., Reelfs B.H., Westerman R.J., III-Vs Review, 15,
48-51, (2002).
[13] Wang J.J., Lambers E.S., Pearton S.J., Ostling M., Zetterling C.-M., Grow J.M., Ren F.,
Solid-State Electronics, 42, 743-747, (1998).
[14] Mozetic M., Zalar A., Vacuum, 71 (1-2), 233-236, (2003).
[15] Choi K., Ghosh S., Lim J., Lee C.M., Appl. Surf. Sci., 206, 355-364, (2003).
[16] Kim H.-R., Park H.-H., Hyun S.-H., Yeom G.-Y., Thin Solid Films, 332, 444-448,
(1998).
39
Chapter 2
[17] Vallee C., Granier A., Aumaille K., Cardinaud C., Goullet A., Coulon N., Turban G.,
Appl. Surf. Sci., 138-139, 57-61, (1999).
[19] Boswell R.W., Chen F.F., IEEE T. Plasma Sci., 25, 1229-1244 (1997).
[20] Chen F.F., Boswell R.W., IEEE T. Plasma Sci., 25, 1245-1257, (1997).
[21] Landau L.D., J. Phys., 10, 25, (1946).
[22] Boswell R.W., Ad. in Space Res., 1, 331-345, (1981).
[23] Yun S.-M., Chang H.-Y., Phys. Lett. A, 248, 400-404, (1998).
[24] Boswell R.W., Henry D., Appl. Phys. Lett., 47, 1095-1097, (1985).
[25] Perry A.J., Boswell R.W., Appl. Phys. Lett., 55, 148-150, (1989).
[26] Boswell R.W., Vender D., Plasma Sources Sci. Technol., 4, 534-540, (1995).
[27] Charles C., Boswell R.W., J. Vac. Sci. Technol., A, 13, 2067-2073, (1995).
[28] Perry A.J., Vender D., Boswell R.W., J. Vac. Sci. Technol. B, 9, 310-317, (1991).
[29] Charles C., Boswell R.W., J. Appl. Phys., 78, 766-773, (1995).
[30] Charles C., Boswell R.W., J. Appl. Phys., 84, 350-354, (1998).
[31] Glow discharge Processes, Chapman B., Wiley, New York, (1980).
[32] Amemiya H., J. Phys. D: Appl. Phys., 23, 999-1014, (1990).
[33] Franklin R.N., Plasma Sources Sci. Technol., 10, 162-167, (2001).
40
Chapter 3
Synthesis and characterisation of SiCxOyHz thin films
Chapter 3: Synthesis and characterisation

of SiOxCyHz thin films
Abstract
The poor adhesion between plastic substrates and hard ceramic-like
coatings is one of the major obstacles in the development of scratch
resistant coatings for plastics. A new dual plasma CVD deposition
system for depositing a novel polymer/inorganic hybrid films has been
developed and constructed and is presented here. Mechanical properties
of the ORMOCER coatings are characterised and compare favourably
with wet chemically synthesized coatings of similar composition*.
3.1 Introduction
The synthesis of organic/inorganic hybrid films on plastics substrates has
traditionally been performed using wet chemical techniques in conjunction with dip
coating or spin coating methods [1]. These methods require several processing
steps, long curing times and need for large volumes of solvents in the synthesis of
these coatings. The use of gas phase deposition techniques to synthesize hybrid
films can potentially offer a solution to a number of these problems. Typically, gas
phase deposition processes are solvent free, one step processes that do not require
subsequent curing. Dense and pinhole free films can be deposited as no reaction
by-products or solvents need to be removed after the deposition. The aim of this
chapter is to develop a new process technique to deposit hybrid coatings using gas
phase deposition technology, in particular plasma based synthesis techniques.
3.1.1 Problem description
Many of the problems encountered when attempting to deposit scratch resistant
films onto a plastic substrates are specific to the deposition technique involved, e.g.
porous films are often encountered when using sol-gel deposition techniques [2, 3].
However, a number of issues arise as being inherent to the plastic-SiO2
combination. One issue of particular concern is that of adhesion between a
polycarbonate substrate and SiO2-like layer in the context of plasma deposition
systems. Poor adhesion is the result of one or a combination of reasons including;
* The work contained in this chapter is the subject of two publications.

Alcott G.R, Linden J.L., van de Sanden M.C.M., Mat. Res. Soc. Symp. Proc., 726, 297-302,
(2002).
Alcott G.R., Eijkman D.J., Schrauwen C.P.G., Linden J.L., van de Sanden M.C.M., 10th
Neues Dresdner Vakuumtechnisches Kolloquium Proceedings, 55-60, (2002).
41
Chapter 3
1) High or uncontrolled water content of the plastic [4].

2) Weak boundary layer (WBL) [5], based on the assumption that the plastic
surface contains low molecular weight compounds and diffuse impurities.
3) A poor match of thermal expansion coefficients between the plastic substrate
and SiO2-like film [6].
4) Low surface energy (e.g. polyethylene and polypropylene).
Notable improvements in film adhesion have been reported by using a variety
of pre-treatments [7]. Most pre-treatments involve some sort of cleaning procedure
followed by either a brief exposure to plasma (Ar or O2) or to high temperature for
anything from a few seconds up to several hours. Both physical and chemical
changes in the plastic substrate can occur during these pre-treatments. Thermal pretreatment serves to drive out absorbed water, whereas plasma (usually performed at
reduced pressure) both drives out water and activates the plastic surface through
ion bombardment. While most authors generally acknowledge ion bombardment
improves adhesion because of cross linking of low molecular weight polymer
chains at the surface, Vallon et al. [8], demonstrated that plasma treatment not only
caused a greater extent of cross linking, but also caused photo-fries type
rearrangements (Fig. 3.1). The photo-Fries rearrangements resulted in an increased
number of active bonding sites for the impinging SiO2 precursor molecules to bond.
Klemberg-Sapieha et al. [13] investigated the adhesion of SiNx to polycarbonate
substrates when deposited using a plasma technique and sputtering. They attributed
the improved adhesion of the plasma deposited SiNx to a 60 nm thick interface
layer generated by ablation and re-deposition of the polycarbonate under ion
bombardment in the initial stages of film deposition.
hv
photo-Fries
OH
C O
OH
O
HO
hv photo-Fries
O C O
hv
R1
PC
chain breaking
C O
O
R2
recombination
OH
OH
H
O
OH
O
OH
OH
Figure 3.1: Reaction mechanisms following carbonate bond breaking in PC

during Ar plasma treatment: photo-Fries rearrangements, chain scission
and recombination.
42
Chapter 3
Plasma deposition inevitably subjects the substrate to a heat load, the intensity of
which depends on the type of plasma (high or low density), its proximity to the
substrate (direct or remote) and the power dissipated by the discharge. For
temperature sensitive substrates some of this heat load may be relieved by cooling
the substrate table, however, the substrate surface in contact with the plasma will
still experience elevated temperatures due to ion bombardment. The substrate cools
when the plasma is switched off, and differences in thermal expansion coefficients
between the inorganic coating and plastic substrate cause a stress build up in the
coating. The coating relieves the stress by cracking and accelerated water and
oxygen diffusion through the cracks leads to de-lamination. Stress induced
adhesion failure occurs on a relatively short time scale from immediately after
deposition up to a few days after depending on the stress.
The difference in expansion coefficients between plastics and SiO2 is a more
difficult problem to resolve. Menichella et al. [9] attribute the poor adhesion of
SiO2 (although they speak generally of any intrinsically inorganic coating) to CR39* lenses to a difference in the thermal expansion. Their solution was to deposit
what they termed a gradient layer (see chapter 1). By starting with a pure TEOS
gas mixture they were able to deposit a silicone plasma polymer like film onto the
CR-39 and by gradually increasing the oxygen concentration the top part of the
film resembled SiO2. The interface layer provides a smooth transition in thermal
expansion coefficients between the CR-39 and SiO2, while the top SiO2 provides
the scratch resistance and hardness required. Silva Sobrinho et al. [10] attempted to
show that through ablation and re-deposition processes, plasma treatment of
polymers leads to a similar interface layer. They believe that the generation of this
mixed interface region (interphase) is why plasma processes give superior
adhesion over other deposition techniques. The nature of this interface region and
why it improves adhesion has been the subject of a number of investigations, using
techniques such as x-ray photoelectron spectroscopy (XPS) [8, 11, 12, 10, 13],
attenuated total reflection infrared (ATR-IR) [8, 10], elastic recoil detection (ERD)
[8], electron microprobe analysis (EMA) [10] and ellipsometry [11, 12, 14]. Results
from compositional profile measurements across such interface regions should
however be interpreted with some caution. Techniques such as EMA and ERD
have beam apertures in the order of 30 nm and would therefore indicate a 60 nm
thick interface region even if used to scan an infinitely narrow interface region
(where polymer and SiO2 phases remain physically separate). In addition, ion
induced mixing can occur for soft materials (plastics) further undermining the
reliability of the results. However, these techniques are useful in a comparative
capacity, and in this manner Silva Sobrinho et al. [10] demonstrated that PECVD
films exhibit interface regions twice as thick as that of comparable samples
prepared via PVD techniques.
Delamination can also be caused by diffusion of water and oxygen through the
polymer from the uncoated side and then react with the polymer-film interface [15].
Adhesion failure caused by these effects usually occurs on much longer time scales
Columbia Resin #39, allyl diglycol carbonate.

43
Chapter 3
of up to several months, depending on the permeability and thickness of the

polymer substrate and rate of reaction at the interface region.
Hydrocarbon rich polymer surfaces are often hydrophobic and exhibit low
surface Gibbs energy. These surface conditions favour an island growth mechanism
whereby growth precursors physisorb and diffuse on the substrate surface until they
find an active site to chemically bond. This results in a film, which although
displays homogenous surface coverage, might only be bonded to the substrate at a
limited number of sites.
The deposition of soft polymeric films invariably gives better film adhesion to
plastic substrates then hard inorganic layers and the problem therefore lies in
combining the superior adhesion of polymeric layers with the hardness and barrier
properties of inorganic films. An alternative to the deposition of gradient layers (as
described above) is to deposit nanocomposite thin films consisting of ceramic
particles embedded in a polymer layer. In this concept, the polymer layer provides
good adhesion to the plastic substrate with added toughness and durability, while
the ceramic particles (in this instance SiO2) provide hardness and an improved
diffusion resistance for gases such as oxygen and water. The dense SiO2 particles
improve the barrier properties by effectively increasing the diffusion length of the
diffusing gas, as depicted pictorially in Fig. 3.2.
O2, H2O
Ceramic
Particles
O2, H2O
0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
Polymer
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Layer
000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 00
000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
Plastic 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
Substrate 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Figure 3.2: Improved diffusion resistance of nanocomposite thin films.

3.2 Process design and reactor concept
A reactor consisting of two separate reaction zones for generating the different
components of the final hybrid material was designed and built (Fig. 3.3).
Precursors for the polymer layer enter the lower half of the reactor (inlet 3) where
they are activated by a capacitively coupled plasma formed between the r.f.
powered substrate table and the grounded reactor walls. A low density (~300 W)
plasma is used for the polymer layer formation to prevent substrate damage and
maintain the functionality of the monomer used. Precursors used for the synthesis
of the particles/inorganic fragments are directed into the upper reaction zone (inlet
44
Chapter 3
1), where they are activated by the inductively coupled plasma (ICP). The intension
here is for the high density ICP to fully oxidize the precursor and generate both
active inorganic fragments and particulates which will then go on to be
incorporated into the growing polymer layer. In addition to the siloxane precursors,
oxygen can also be dosed into either Zone I or II by a pneumatically controlled gas
dosing system.
Particulates are transported from Zone I to Zone II by a combination of
physical transport effects but mainly gas flow and gravity. To avoid particles
charging and becoming suspended in the electric field a pulsed plasma supply is
used. Both plasmas were generated using Huttinger PFG 1000 RF generators, all
gases were delivered using Bronkhorst HiTec mass flow controllers. Pressure
regulation of the main reactor vessel is achieved using a MKS throttle valve and
controller located in the exhaust line between the reactor and vacuum pump.
Precursor
inlet 1 (TEOS)
Zone I
Pulsed
r.f. 1
O
O
O
O
O
O
Precursor
inlet 2 (O2)
Ellipsometer
windows
Zone II
FTIR
FTIR
Substrate
table
Precursor
inlet 3
To pumps
Pulsed
r.f. 2
Figure 3.3: Schematic of dual plasma reactor showing inductive top

plasma (Zone I) and capacitive bottom plasma (Zone II) with in situ
FTIR facilities.
45
Chapter 3
Inlet 1
Inlet 2
O2
Reactor
CEM
00 00
Gas System
Exhaust
Mass Flow Controllers
Inlet 3
To
exhaust
Reactor
Exhaust
Cold Trap
Ar
Pumps
Bubbler 1
Bubbler 2
Figure 3.4: Gas dosing system showing bubbler and CEM systems for dosing of
liquid precursors. All gas lines are heated.
The precursors were vaporised and delivered to the reactor using the gas
delivery system depicted in Fig. 3.4. Liquid precursors were dosed using bubbler
and Bronkhorst High-Tech CEM [16] (Controlled Evaporation & Mixing) systems
with argon as a carrier gas. CEM systems utilises a liquid flow metre to determine
precursor injection rate. In the bubbler system, the volumetric precursor flow is
dependent on the precursor vapour pressure and flow rate of the carrier gas passed
through the liquid. The precursor flow carried from the bubbler is calculated by,
F precursor =
x
Fcarrier
1 x
(3.1)
where Fprecursor is the precursor flow in sccm, Fcarrier the carrier gas flow in sccm
and x the fraction of the precursor vapour pressure divided by the system pressure.
To calculate the precursor flow from the bubblers and to ensure sufficient carrier
gas is dosed into the CEM system, the precursor vapour pressure needs to be know
and have been measured for the precursors used (cf. Appendix A). An overview of
the experiments performed in this chapter is presented in Table 3.1.
46
Chapter 3
Table 3.1: Overview of experiments performed for this chapter.

Experiment
Precursor
Oxygen
Plasma
Pressure
Polymer
synthesis
TPS, TMSE,
TEOS, Glymol
0-60 %
CCP 100-400 W
0.5-2.0
Torr
SiO2 synthesis
TEOS
30-70 %
ICP 300-1000W
CCP-bias 0-170V
0.5-2.0
Torr
Adhesion tests
TMSE
CCP 100-500W
1 Torr
3.2.1 Diagnostics and substrate preparation

Deposition is performed on a number of different substrate materials, including
crystalline silicon <100>, glass, polycarbonate (PC), steel and paper. Post
deposition mechanical characterisation of the films is done using Taber abrasion
(for scratch resistance) and Methyl Ethyl Ketone (MEK) solvent (for detection of
pinholes). Visible transmittance is determined using a Perkin Elmer Lambda 900
photo spectrometer and infrared spectra are recorded using a Midac M2500-C FTIR
spectrometer for films deposited on Si wafers. In situ diagnostics are made using
the same FTIR spectrometer fitted with a mercury cadmium telluride (MCT)
detector and interferometer mounted either side of the reactor. Post deposition
particle size distribution was determined using a Beckman Coulter LS230 particle
size analyser. Film thicknesses are measured using a Tencor P-10 step profiler.
The plasma pre-treatments are performed using the capacitively coupled r.f.
discharge (Zone II), with Ar and O2 as active gases, while deposition was achieved
using 1,2-bis(trimethyl)siloxyethane and Ar mixtures and sometimes O2 to produce
SiO2 layers. Polycarbonate substrates (obtained from General Electric) are first
rinsed with ethanol and dried in air. Adhesion performance was graded 0-5 using
the cross cut tape test [17], where 5 represents no delamination of the coating and 0
indicated more than 65 % removal of the coating in accordance with the standard as
represented in Appendix B.
3.3 Polymer synthesis in the capacitive plasma (Zone II)
To characterise the deposition characteristics of the system, plasma polymerisation
of TPS, TMSE and TEOS were optimised with respect to discharge power and
precursor concentration using argon as carrier gas. Film deposition rate is
dependant on the precursor concentration and the power dissipated by the
discharge. The peak growth rate at 300 W was seen for TEOS precursor, with a
partial pressure of 0.1 Torr corresponding to a peak deposition rate of 2 nm/s. At
higher partial pressures particles were seen in the reactor and the film quality
suffered, this was especially prominent for the TPS precursor. The addition of
oxygen to the siloxane deposition accelerated the growth rate and lowered the
threshold at which powder formation occurred. The increased growth rate is likely
to be caused by atomic oxygen radicals in the discharge. Film hardness also
47
Chapter 3
Transmission(0-1)
(0-1)
Transmission
improved with addition of oxygen but adhesion to the polycarbonate substrates

deteriorated as is usually observed with more SiO2-like properties.
IR measurements of the plasma polymerised TMSE layers (in the absence of
O2) showed them to retain much of the functionality of the original precursor (Fig.
3.5) This is indicated by the large C-H stretching region seen between 2850-3000
cm-1 and the Si-CHx peaks found between 750-1400 cm-1. The CO2 band seen at
approximately 2350 cm-1 is due to atmospheric CO2 in the ambient of the
spectrometer. At high deposition powers denser more SiO2 type films were
deposited as indicated by a reduction and broadening of the C-H stretching
vibrations (2850-3000 cm-1 and at 1100 cm-1).
1
Flim
0.9 Film
0.8
0.7 Liquid
Liquid
0.6
0.5
0.4
0.3
0.2
3800 3500
CO2
C-H
Si-CHx
3200
2900
2600
2300
2000
1700
1400
1100
Si-C
800
501
-1
Wavenumber [cm
Wavenumber
(cm-1] )
Figure 3.5: Plasma polymerised TMSE (film) retains much of the

functionality of the liquid TMSE precursor (liquid).
3.4 Inorganic layer synthesis in the inductive plasma (Zone I)
TEOS and O2 were activated using the ICP and the active precursor fragments
flowed into Zone II where they formed a SiO2 layer. Increasing the O2
concentration resulted in more SiO2-like layers being deposited. Although
stoichiometric SiO2 can be produced with the ICP, ion bombardment achieved by
applying an r.f. bias to the substrate table, was necessary in order to achieve dense
SiO2 layers with relatively low levels of SiOH. The broad absorption feature
appearing between 2900-3800 cm-1 in the infrared spectrum shown in Fig. 3.6
indicates incorporation of SiOH moieties in the plasma deposited TEOS layers.
As the d.c. bias potential is decreased from 170 to 0 V the peak associated to
SiOH groups increases indicating a reduction in density of the layer. The peaks
centred at 1070 cm-1 represent Si-O stretching, while the rather broad absorption
between 2700 and 3600 cm-1 indicates Si-OH character. Silanol bending peaks also
appear at 925 cm-1. The higher spectral definition of these regions supports this
assumption, as these peaks would be much broader if the H2O and CO2 were
incorporated into the films themselves.
48
Chapter 3
140
IR transmission (%)
120
b
100
60
40
20
80
a
Si-OH(s)
CO2
Si-O
Si-O
Si-OH(b)
0
3500
3000
2500
2000
1500
1000
500
-1
Wavenumber (cm )
Figure 3.6: FTIR spectra of SiO2 deposited with 1kW ICP and various
d.c. bias voltages, a. = 170 V, b =100, V c. = 80 V and d = 0 V.
Total pressure, total gas flow and precursor concentrations were all varied in
an attempt to synthesize nanometre size particulates using the inductive plasma.
However a number of issues prevented the synthesis of particles using the ICP.
Firstly, the lack of any confining horizontal electric fields in the ICP required
operating the discharge under high pressure/low flow conditions to maximise the
residence time and encourage homogeneous nucleation. Due to the long mean free
path, activated precursor fragments undergo numerous collisions with the ICP tube
walls before reacting reactor Zone II. This resulted in rapid contamination of the
ICP tube walls with black silicon/carbon deposits. These deposits delaminated as
the tube cooled between plasma operations and contaminated the layers growing on
the substrate beneath. In addition to this, the ICP was unstable at high precursor
concentrations and a very small bright plasma in the centre of the tube (causing the
quartz tube to melt in a matter of seconds) suggested a switched to a capacitive
power coupling as the pressure was increased.
3.5 Film adhesion
The water content of polycarbonate varies depending on the storage conditions.
The typical water content for polycarbonate stored at room temperature and
humidity is approximately 0.1-0.2 wt%. Some of the adsorbed water can be
removed from the polycarbonate by heating in an oven for two hours at 120C. The
structural changes caused by the oven pre-treatment can be seen in the infrared
spectra shown in Fig. 3.7.
49
Chapter 3
0.04
C-H
stretching
IR absorption
0.02
0.00
-0.02
-0.04
-0.06
5000
O-H (H2O)
4000
Aromatic C-H
stretching
3000
2000
1000
-1
Wavenumber (cm )
Figure 3.7: FTIR spectrum recorded during a thermal pre-treatment of
PC, negative absorption bands at 3600 cm-1 and 3100 cm-1 indicate a
loss of H2O and saturated C-H respectively. Positive stretches at 2900
cm-1 indicate an increase in unsaturated C-H content.
The negative absorption bands between 3500 and 3750 cm-1 indicate loss of
water from polycarbonate, as does the re-distribution of modes in the C-H
stretching regions between 2800 and 3100 cm-1. Dehydration of the polycarbonate
from oven treatment can also be seen by changes in the surface energy. Before
oven pre-treatment polycarbonate has a contact angle with water droplets of 75,
samples exposed to oven temperature of 120C for anything from two to twenty
four hours exhibited more hydrophobic surfaces with contact angles ranging from
84 to 90.
Although many authors report improved adhesion with mild plasma pretreatment, in this study plasma pre-treatment over two minutes appeared to lead to
polymer degradation and poor adhesion. In a series of experiments in which a
silicone layers were deposited from 1,2-bis(trimethyl)siloxyethane onto PC
substrates after a series of plasma pre-treatments, film adhesion was seen to
decrease with increased pre-treatment. Fig. 3.8 shows how different exposure to an
argon plasma pre-treatment affects the adhesion of a pp-TMSE layer (0 hours) and
the longer term stability (336 hours).
50
Chapter 3
Adhesion (rated 0-5)
4
3
2
0 Hours after deposition
1
0
2
4
6
8
Treatment time (minutes)
10
Figure 3.8: Adhesion of plasma deposited silicone layer after various

plasma pre-treatment times. Pre-treatment conditions were 300 W r.f.
capacitive Ar discharge at 1 Torr.
Adhesion (rated 0-5)
4
3
2
1
0

0
10
15
20
25
30
Oxygen (vol%)
Figure 3.9: Decrease in adhesion as film composition moves from

silicone like to silica-like with increasing O2 concentration in feed gases.
Deposition performed using a 300 W capacitive r.f. plasma with 1,2bis(trimethyl)siloxyethane, Ar and O2 as precursors.
Exposure times below two minutes do not appear to improve or degrade the
immediate adhesion properties. Above two minutes exposure to the plasma may be
51
Chapter 3
causing damage to the polycarbonate surface either in the form of high-energy ion
bombardment or UV emission from the plasma. As it is common for commercial
polycarbonate to contain UV stabilizers, degradation due to UV radiation from the
plasma is unlikely to be responsible for the failing adhesion. Ion bombardment
does, however, subjects the PC substrate to a heat load and longer pre-treatment
times will therefore result in higher substrate temperatures. As the TMSE film is
deposited immediately after the pre-treatment, longer pre-treatment times will
result in a high substrate temperature during film growth, and consequently higher
intrinsic stress in the final coating. As the deposited layer moves from a silicone
structure towards that of SiO2, compressive stress in the layer causes delamination
(Fig. 3.9).
3.6 Mechanical/optical properties of hybrid layers
The optimum mechanical properties were measured for plasma polymer layers
synthesized in Zone II using TEOS and oxygen as precursors. The films were
pinhole free (good MEK test results) and showed good abrasion and optical
properties. However, despite pre-treating the polycarbonate substrate at 120C in
an oven for 2 hours the films exhibited poor long term adhesion. After 2 weeks the
films showed signs of cracking and delamination and were unsuitable for barrier
measurements. Results of tests administered before film degradation can be seen in
Table 3.2 together with a comparison with wet chemically in-house deposited
hybrid film*.
Table 3.2: Comparison of physical properties of hybrid coatings deposited via gas
phase and wet chemical techniques.
Characterisation
Gas phase synthesis
Wet chemical synthesis
Taber abrasion (% loss of

transmission after 100 cycles)
1.4
5.0
Solvent resistance (MEK)
>100 double rubs
100 double rubs
Adhesion (cross cut tape test)
Good
Good
Impact
Good
Good
Surface free energy
43.22 mJ/m2
17 39.1 mJ/m2
Contact angle water
69
76 - 92
The in-house hybrid film was prepared from a sol-gel solution containing 50 wt% Glymol,
30 wt% Methytrimethylsilane and 20 wt% aluminium isobutoxide reacted with 1 equivalent
water with isopropyl alcohol as a solvent. The coating was oven cured at 130C for 3 hours
and had a thickness of approximately 1 m.
52
Chapter 3
3.7 Discussion
Relating film properties to deposition parameters in PECVD processes is more
difficult to achieve than in thermally activated CVD, and any relationships
determined are usually only valid for a specific deposition apparatus. The
interdependency of many plasma parameters (Te, ni, ne along with various electric
fields) in conjunction with a low pressure system, make PECVD processes difficult
to fully characterise to the extent where no one system parameter can be
independently varied to determine its influence on the overall process. For
example, here we attempted to determine the influence of discharge power on the
film adhesion. Pre-treatment times exceeding two minutes adversely effected film
adhesion and this was primarily attributed to increased heating of the substrate due
to extended exposure to the plasma. However, it is possible that film adhesion
suffered because of damage to the polycarbonate substrate caused by ion
bombardment or exposure to UV radiation. Similar problems were encountered
when correlating growth rate to plasma parameters. Growth rate is observed to
increase as a function of the power applied to the discharge, but are films deposited
at low power comparable to films deposited at higher power? At low power levels,
and hence low density, the precursor will be only partly fragmented and much of
the precursor structure will be present in the final layer. As the power is increased,
the precursor may undergo several dissociations and the resulting film may differ
from the film produced at lower power. However, in addition to ne, Te may also
vary with power and increased dissociation will result in a reduction in the average
ion mass, both of which affect the ion sheath and hence ion energy at the substrate
(as can be seen from Eq (4.4) in chapter 4). Variations in the ion energy will also
affect the film structure and composition. The substrate temperature may also
change as a consequence of increased ion bombardment from the plasma, which
may also influence growth rate and film properties. Does the introduction of
oxygen, for example, increase the growth rate by affecting the discharge parameters
or by accelerating the discharge chemistry? Some of these issues are the subjects of
later work. The chemistry of the TEOS/O2 plasma system is investigated in chapter
5.
3.8 Conclusions
This chapter set out to characterise the deposition system designed for the gas
phase synthesis of hybrid thin films. Although silicone-like polymer films and SiO2
films have been deposited and characterised, a number of questions have arisen
regarding the interaction of plasma on the substrate surface and chemical reactions.
Some general conclusions however can be drawn.
The synthesis of inorganic SiO2 layers was successfully demonstrated using
TEOS/O2 injected into an ICP. Additional r.f. biasing of the substrate was
necessary to achieve reasonable deposition rates and to remove SiOH and densify
the films. Infrared transmission spectra showed that by using high discharge power
and relatively high oxygen concentrations, C and H free SiO2 could be generated in
the inductive discharge, as demonstrated in Fig. 3.6. Although these layers
exhibited excellent mechanical properties they also exhibited poor adhesion to
polycarbonate substrates. All attempts at synthesizing particulates using the ICP
53
Chapter 3
discharge were unsuccessful, probably due to the absence of confining electric

fields and thus short residence times for negative ions (cf. chapter 4).
The synthesis of silicone polymer layers from various siloxane precursors was
successfully demonstrated. Adhesion was observed to deteriorate as the siloxane
layers became more inorganic (increase O2 content in plasma) and was therefore
attributed to intrinsic stress caused by temperature increases during the plasma
deposition process. Of the precursors evaluated, TMSE exhibited the best long term
stability and adhesion to the polycarbonate substrate. This was attributed to the
higher carbon content of this precursor that produced a softer SiOxCyHz material.
The addition of oxygen to the siloxane discharge increased the film deposition rate
and hardness, but reduced layer adhesion to the polycarbonate substrates. Plasma
pre-treatments in excess of two minutes adversely affected layer adhesion to
polycarbonate substrates.
Acknowledgement
I would like to thank M. Creatore for assistance with the deposition and
characterisation of the inorganic SiO2 layers.
References
[1] Sol-Gel Science: The physics and chemistry of sol-gel processing, Brinker J., Scherer
G.W., Academic Press, (1990).
[2] Conde A., Duran A., Damborenea J.J. de, Prog. Org. Coat., 46, 288-296, (2003).
[3] Que W., Zhang Q.Y., Chan Y.C., Kam C.H., Comp. Sci. Technol., 63, 347-351, (2003).
[4] Schafer M.M., Seidel C., Fuchs H., Voetz M., Appl. Surf. Sci., 173, 1-7, (2001).
[5] Vallon S., Hofrichter A., Drevillon B., Klemberg-Sapieha J.E., Martinu L., PoncinEpaillard F., Thin Solid Films, 290-291, 68-73, (1996).
[6] Fu Y., Du H., Sun C.Q., Thin Solid Films, 424, 107-114, (2003).
[7] Baumgartner K.-M., Schneider J., Schulz A., Feichtinger J., Walker M., Surf. Coat.
Technol., 142-144, 501-506, (2001).
[8] Vallon S., Hofrichter A., Guyot L., Drevillon B., Klemberg-Sapieha J.E., Martinu L.,
Poncin-Epaillard F., J. Adhes. Sci. Technol., 10, 1287-1311, (1996).
[9] Menichella S., Misiano C., Simonetti E., De Carlo L., Carrabino M., Soc. Vac. Coat., 37th
Annual Technical Conference Proceedings, 37-40, (1994).
[10] Da Silva Sobrinho A.S., Schuhler N., Klemberg-Sapieha J.E., Wertheimer M.R., J. Vac.
Sci. Technol., .A 16, 2021-2029, (1998).
[11] Vallon S., Brenot R., Hofrichter A., Drevillon B., Gheorghiu A., Senemaud C., Sapieha
J.E.K., Martinu L., Poncin-Epaillard F., J. Adhes. Sci. Technol., 10, 1313-1332, (1996).
[12] Bergeron A., Klemberg-Sapieha J.E., Martinu L., J. Vac. Sci. Technol., A 16, 32273234, (1998).
[13] Klemberg-Sapieha J.E., Poitras D., Martinu L., Yamasaki N.L.S., Lantman C.W., J. Vac.
Sci. Technol., A 15, 985-991, (1997).
[14] Bergeron A., Klemberg-Sapieha J.E., Martinu L., J. Vac. Sci. Technol. A 16, 3227-3234,
(1998).
[15] Roualdes S., Sanchez J., Durand J., J. Membr. Sci., 198, 299-310, (2002).
[16] Boer H.J., J. de Physicque IV, 5, C5-961, (1995).
[17] Annual Book of ASTM standards, 06.01, D-1 on Paint and Related Coatings, Materials
and Applications E-12 on Appearance, PCN: 01-060196-14, (1996).
54
Appendix A
Vapour pressures of precursors for the CVD of silicon based films
Appendix A: Vapour pressures of

precursors for the CVD of
silicon-based films
Abstract
Vapour pressure measurements of the precursors tetraethoxysilane
(TEOS), tripropylsilane (TPS), (3-glycidoxypropyl) trimethoxysilane
(Glymol),
perfluoroctyltriethoxysilane
(PTES)
and
1,2bis(trimethyl)siloxyethane (TMSE) in the temperature range 20 to 200C
are presented. Measured data was fit using the integrated ClausiusClapeyron equation, pvap=p0 e-H/RT, where p0 is the integration constant
and H is the enthalpy of vaporization*.
Introduction
Precursors suitable for CVD experiments need to be easily vaporized to form
vapours stable during transport but readily reactor when activated. Assuming these
criteria are satisfied a precursor is then selected based on the specific properties of
the layer it produces [1], costs (including implementation), health and safety
precautions, and environment impact. In practice, metal halides (usually F or Cl
due to the poor volatility and stability of the heavier halide compounds), alkoxides,
acetyl acetonates and organometallics (M-(CxHy)z, where M = central metal) and
compounds with various combinations of these ligands generally satisfy the
criteria. Since there has been considerable interest in the deposition of silica and
silicone type materials (see chapter 1) a large number of suitable precursors have
been found and tested. Silicon organometallic compounds (Si(CH3)4 and higher)
and SiH4 although used are either extremely flammable or pyrophoric and
considerable safety precautions need to be taken when using them. The silicon
halides although stable, are very reactive and readily form corrosive by products
such as HF or HCl on contact with water. In addition to the corrosive nature of the
halides, the tendency to form environmentally harmful halogen-hydrocarbons (e.g.
Freon) necessitates the implementation of expensive waste gas processing.
Consequently, the most attractive compounds for silica/silicone deposition are the
silicon
alkoxides
and
derivates
thereof.
Tetraethoxysilane,
1,2bis(trimethyl)siloxyethane were chosen as low cost silicon alkoxides. (3glycidoxypropyl) trimethoxysilane is a compound used in the wet chemical
synthesis of silica/silicone based hybrid thin films and was chosen to make
comparisons between the gas phase and wet chemical processing techniques.
Tripropylsilane was selected to deposit hydrophobic SiOxCyHz films as the presence
of Si-O moieties is known to encourage H-bonding and absorption of water.
*
Alcott G. R., van de Sanden M.C.M., Kondig S., Linden J.L., Chem. Vap. Dep., 10, 20-22,
(2004).
55
Appendix A
The controlled vaporization and subsequent transport of liquid and solid

precursors is an essential part of gas phase process technology. To accurately
control precursor feed rate when using conventional bubbler systems, it is
imperative to know the precursor vapour pressure, thermal stability and in many
cases the purity of the sample [2, 3, 4]. Here we report further measurements
performed using a procedure and system developed in-house [1] designed to
quantify vapour pressure, stability and the presence of volatile contaminants. The
precursors investigated in this study are used for the deposition of silicon based
inorganic/organic hybrid thin films.
Experimental
A detailed description of the vapour pressure apparatus can be found elsewhere [1],
however, for convenience a brief description is given here. A sample is loaded into
the sample vial and attached to the vacuum system situated inside an oven at valve
V3 (Fig. A.1). Valves V1 and V2 are opened and the chambers C1 and C2, situated
between the differential pressure gauge and V1 and V2 respectively, are pumped
down to a base pressure of 3.010-2 Torr. After the system has been pumped down,
V1 is closed and V3 opened for a sufficient time to remove all air from the sample.
Solid samples are pumped for longer to allow desorption of water and other
physisorbed contaminants from the sample surface. After removing contaminants
from the sample, V3 is closed and remains closed for a predetermined time interval
(t), while V1 and V2 are opened for a short time interval (10 seconds) and then
closed. The measurement commences at the start of t when V3 is closed. At the end
of the time interval t, V3 is opened and the vapour pressure of the sample expands
into C2, and allowed to equilibrate for 10 minutes. The measurement is concluded
by closing V3 and opening V2.
C1
V1
C2
Cold trap
V2
V3
T
= Rotary vane pump
= Differential pressure
= Oil Diffusion pump
P = Pressure gauge
= Pneumatic valve
T = Quartz thermocouple
Oven
= Sample vial
Figure A.1: Schematic of the vapour pressure measuring system.
56
Appendix A
Measurements at different temperatures are performed by setting the oven

temperature (maximum 200C) and repeating the aforementioned procedure after
allowing the system and sample to equilibrate at a new oven temperature. A
schematic of the apparatus is depicted in Fig. A.1. The number of measurement
points was 29, 26, 19, 29 and 28 for TEOS, TPS, TMSE, Glymol and PTES
respectively. Systematic errors are avoided by performing the measurements in a
random order with respect to temperature. Some degassing of the samples was
observed during initial sample measurements, but these results were discarded.
Experimental error is based on the tolerance of the differential pressure gauge and
taken to be approximately 2 %. Statistical error in the recorded data is calculated
using least squares regression with a 95 % confidence interval.
Precursor stability measurements
The system is designed and operated in such a way as to differentiate effect such as
precursor decomposition, degassing of lines and leakages from the vapour pressure
data itself. This is achieved by varying the time interval t before each measurement
and recording the pressure changes as a function of time both before and after
opening the sample valve. A hypothetical measurement curve can be seen in Fig.
A.2. The gradient of the curve measured during t, i.e. when sample is isolated from
the system (curve A), provides a measure of the system leak rate plus any
degassing of the walls. By subtracting curve A from curve B the pressure increase
due to sample decomposition can be estimated (assuming no degassing of the
sample). If the derivative at A and B are equal then no sample decomposition
occurs. The pressure exerted by the sample is then measured by extrapolating curve
B to the point where the sample is opened to the system (C). The true sample
vapour pressure is then determined by plotting p (C) values as a function of t and
extrapolating back to t = 0.
14
12
p (Torr)
10
6
4
A
2
0
-2
t
0
10
20
30
40
50
60
70
Measurment time (minutes)
Figure A.2: Hypothetical measurement result for determining leak rate

and sample decomposition.
57
Appendix A
(3-glycidoxypropyl) trimethoxysilane [CAS 2530-83-8] (98 %) and tripropylsilane

[CAS 998-29-8] (99 %) were obtained from Aldrich, 1H, 1H, 2H, 2Hperfluorooctyltriethoxysilane [CAS 51851-37-7] was obtained from ABCR
(Brunschwig), 1,2-bis(trimethyl)siloxyethane [CAS 7381-30-8] (99 %) and
tetraethoxysilane [CAS 78-10-4] were obtained from (Merck). Samples were used
as delivered (no special sample preparation methods) and none of the chemicals
were purified before use.
Results & Discussion
The vapour pressure curves for each of the precursors can be seen in Fig. A.1. The
vapour pressures for TEOS, TMSE and TPS are virtually the same within the
temperature range measured in this study. The enthalpy of vaporization can be
calculated from the gradient of the vapour pressure curve using the ClausiusClapeyron equation (cf. Eq. (A.1)) and are presented in Table A.1 along with the
calculated values for the enthalpy of vaporization.
H
p vap = p * exp
R
1 1
*
T T
(A.1)
where p* is the vapour pressure at some temperature T*. H is the enthalpy change
of vaporization in kJ/mol, R is the gas constant and p is the vapour pressure in Torr.
Calculated values H are summarised in Table A.1.
(a)
T ( C)
200 175 150 125 100
75
50
25
Vapour pressure (Torr)
1000
TEOS [22]
TEOS [23]
TEOS
100
10
1
2.2
2.4
2.6
2.8
3.0
-1
1000 / T (K )
58
3.2
3.4
3.6
Appendix A

o
T ( C)
(b)
120
110
100
90
80
70
60
100
TPS
10
2.55
2.70
2.85
3.00
3.15
3.30
-1
1000 / T (K )
(c)
T ( C)
200 175
100
150
125
100
75
50
10
Glymol.
PTES
TMSE
2.2
2.4
2.6
2.8
3.0
3.2
-1
1000 / T (K )
Figure A.3: Vapour pressure plots (a) comparison between Handbook

[5], I & E Chemistry [6] and measured values for the vapour pressure
curve of tetraethoxysilane, (b) tripropylsilane, (c) (3-glycidoxypropyl)
trimethoxysilane, 1H, 1H, 2H, 2H-perfluoroctyltriethoxysilane and 1,2bis(trimethyl)siloxyethane.
The TEOS vapour pressure curve measured in this study shows good
agreement with data previous published in the literature [5, 6] (see Fig. A.3a).
59
Appendix A
Table A.1: Summary of evaporation enthalpies for various silicon based

precursors Glymol, PTES, TPS, TMSE and TEOS. TEOS [6] and TEOS [5]
refer to data reported in the literature.
Precursor
H
(kJ/mol)
Glymol
PTES
TPS
TMSE
TEOS
TEOS [6]
TEOS [5]
69.4 1.7
66.1 .7
48.2 0.5
48.5 0.7
50.8 0.6
45.8 1.1
42.8 0.0
RMM* (g/mol)
236.33
510.24
158.35
206.42
208.32
-
Measuring range
(C)
80-180
100-160
50-110
50-100
20-100
16-168.5
25-110
None of the precursors showed any significant signs of decomposition within

the temperature range they were measured, although signs of de-gassing and the
presence of volatile impurities could be seen for all of the samples during initial
measurements.
References
[1] van Mol A.M.B, Driessen J.P.A.M., Linden J.L., de Croon M.H.J.M., Spee C.I.M.A.,
Schouten J.C., Chem. Vap. Dep., 7, 3, 101-104, (2001).
[2] Spee C.I.M.A., Verbeek F., Kraaijkamp J.G., Linden J.L., Rutten T., Delhaye H., van de
Zouwen E.A., Meinema H.A., Mat. Sci. Eng., B17, 108, (1993).
[3] Spee C.I.M.A., Linden J.L., Mackor A., Timmer K., Meinema H.A., Mat. Res. Soc. Symp.
Pro., 93, 425, (1996).
[4] Waffenschmidt E., Musolf J., Heuken M., Heime K., J. Supercond., 119, 5, (1992).
[5] CRC Handbook of Chemistry and Physics 1st student edition, R.C. Weast, (1987).
[6] Industrial & Engineering Chemistry (Vapor Pressure of pure substances), D.R. Stull,
(1947).
RMM refers to Relative Molecular Mass
60
Appendix B
Cross cut tape test
Appendix B: Cross cut tape test

The ASTM cross cut tape test [17] is designed for evaluating the adhesion of thin
films with a thickness of less than 125 m. The test is administered by making two
sets of six parallel cuts perpendicular to one another on a selected area of the
substrate free of blemishes and minor surface imperfections. The cuts are made
using a sharp razor blade, scalpel, knife or other cutting device in conjunction with
a straight edge and ensuring that the coating film has been penetrated. Tape is
applied across the cuts and pressed into place by rubbing with the eraser end of a
pencil. Within 90 30 seconds of application the tape is removed and the coating
rated according to a visual inspection of the coating. In this study, films subjected
to the cross cut tape test were rated according to Table B.1.
5 0% area removed
4 <5% area removed
3 5-15% area removed
2B 15-35% area removed
1B 35-65% area removed
0 > 65% area removed
Figure B.1: Key to cross cut tape test results.

61
62
Chapter 4
Synthesis, monitoring and characterisation of nanoparticles...
Chapter 4: Synthesis, monitoring and

characterisation of
nanoparticles for incorporation
into nanocomposite layers
Abstract
This chapter deals with the synthesis and characterisation of nanometre
sized SiO2 particles in a PECVD system for incorporation into polymer
layers. Attempts to synthesize particles using the ICP discharge
described in chapter 3 were unsuccessful due to the rather low residence
times and absence of confining fields. Particle formation in capacitive
plasma systems, notably in semiconductor processing, has however been
extensively reported in the literature. Literature concerning the
formation and behaviour of particles in low pressure gas discharges is
reviewed with the goal of determining typical process conditions and
formation mechanisms. Relevant theories and equations are taken and
used in rudimentary calculations for the process conditions and reactor
geometry used in this study. FTIR absorption spectroscopy coupled with
XPS and ESEM measurements are used to determine the composition,
size and number density of particles trapped in the discharge.
4.1 Introduction
Recent trends, aided by the growing hype surrounding nanotechnology, have made
the incorporation of nano-particles into growing layers desirable. Numerous
examples where the controlled incorporation of nanometre-sized moieties into
growing films produces materials with unique properties that can be tailored to suit
particular applications, exist in the literature [1, 2]. The incorporation of nanometre
sized silica particles into polymeric films has been shown to densify the material
and consequently improve mechanical and physical properties. Improvements in
such properties as tensile strength [3], thermal stability [4], scratch and wear
resistance [5], diffusion barrier and low-k have been demonstrated. In addition to
the synthesis of new and novel materials, the incorporation of particles into
growing layers has also been shown to improve growth rates and precursor
conversion yields in precipitation CVD processes [6, 7].
In deposition systems used in the semiconductor and micro electronic
industries, particle production is an undesirable by-product of some plasma
processing steps. Device failure as a result of dust contamination is a familiar issue
and has prompted research into particle formation and transport mechanisms in the
hope that it might be avoided [8]. A prominent area of literature regarding particle
63
Chapter 4
contamination deals with silane chemistries [9, 10, 11, 12, 13, 14] due to its
widespread application in industrial processes. Although particles have been
observed in TEOS and HMDSO [15] plasma discharges, very little has been done
in elucidating the mechanism by which they are formed. This is perhaps partly due
to the added complexity of these gas mixtures brought on with the addition of the
carbon chemistry.
Fujimoto et al. [16] investigated particulate formation from four siloxane
precursors (triethoxysilane, tetramethylorthosilicate, octamethyl-cyclotetrasiloxane
and TEOS) in an atmospheric pressure, thermally activated CVD process and
determined Arrhenius type activation energies (Ea) for the overall conversion of
precursor into particle. As no consideration for thermal or fluid dynamical effects
were taken the precise values for the activation energies reported should be taken
with care. Their work did however suggest that TEOS, with the lowest overall
reaction activation energy, would most readily form particles. Before particles can
be synthesized and incorporated into growing silicone layers, an understanding of
particle formation and transport processes in CVD reactors is required.
4.1.1 Particle formation in r.f. discharges
Two prominent mechanisms have been proposed to explain the manifestation of
particles in low pressure discharges. At pressures typical for plasma processing the
cross sections for chemical reactions between neutral species are too low to explain
the plethora of particles found. Reaction cross sections are several orders of
magnitude higher for ions, but positive ions are easily lost to the vessel walls where
they either react to form a coating or pick up an electron and neutralise. Research
into SiH4 plasmas [17, 18, 19, 20 ,21], has revealed that negatively charged ions
can be formed in some gases and, being repelled by the opposing potential in the
sheath at the reactor walls, become trapped in the discharge. Negative ions are
predominantly the product dissociative attachment processes where electron impact
dissociates a neutral species to form a neutral species and a negative ion. Trapped
negative ions act as nucleation centres for homogeneous particle growth. The high
electro-negativity of fluorine and oxygen, causing high negative ion concentrations,
are believed to be responsible for particle production in etch processes [22, 23, 24,
25]. The presence of negative ions also affects the plasma properties and causes
them to differ from those of electropositive plasmas as formed from noble gases
such as argon. The loss of electron density through electron attachment processes
drives the electron temperature up in order to sustain the ionization rate. At high
negative ion concentration (sometimes a factor of ten higher then ne [26]) most of
the negative charge is carried by the negative ions and the sheath potential drops,
and as the current is still carried by the electrons, the plasma becomes more
resistive.
Particles have also been observed in plasma systems where no reactive or
electronegative gases have been introduced. In these systems, particles are thought
to be ejected from the reactor wall as a result of ion bombardment. Once ejected
particles rapidly attain negative charge from the plasma glow and become trapped.
Suh et al. [27] proposed a more extensive mechanism and devised a model for
cluster formation up to (SiO2)10 from SiH4 and O2 gas mixtures using quantum
64
Chapter 4
chemical theory. Assuming clusters containing more than ten silicon atoms (SiO2)10
were irreversibly formed and that the nucleation and initial growth of these species
was very rapid, they modelled the nucleation and coagulation of particles. The
model predicted a fall in particle concentration immediately after the initial
nucleation period due to coagulation processes. Immediately after nucleation two
mechanisms govern the particle growth rate, coalescence, where particle size
increase at the expense of particle number density, and the arrival of growth
precursors at the particle surface. Coalescence dominates particle growth
immediately after nucleation as growth precursors are depleted. However, as
particle numbers decrease and surface charge increases with particle size causing
particle to repel on another, the arrival of growth precursors begins to dominate.
At a certain point growing clusters of molecules will reach a size where they
begin to experience additional forces to those they experienced as a gas. This point
has been proposed to be once a cluster consisted of ten Si atoms or more (for SiO2
particles) [27] or when a cluster reaches a radius of about 10 nm [8].
In order to achieve controlled synthesis and incorporation of these
macroscopic particles into growing layers, it is therefore necessary to understand
the forces acting on particles in a plasma discharge and how these forces vary
depending on the particle dimensions and plasma parameters.
4.2 Forces on particles
In the electrode configuration used in this study, the electrostatic force and
thermophoretic force act to push the particles away from the r.f. electrode (cf. Fig.
4.1). These forces are counteracted by the ion and neutral drag (Fid, Fnd) and
gravitational force (Fg) (see Eqs. (4.8) to (4.12)) that act to push the particle
downwards towards the electrode (Fig. 4.1). The relative magnitude of these forces
determines the position of the particles since all forces are a function of the position
in the reactor.
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 F
00 00 00 00 00 00g00 00 00 00 00 00+
00 00 00 00 00 00 00 F
00 00 00 00 00 00Nd
00 00 00 00 00 00 00 00 00 00+
00 00 00 00 00 00 00 F
00 00 00 00 00 00id
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0000000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 e
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00+00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00+00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 +
00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00+00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000000000000000000000000000000000000+00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000e00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0+0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000 000 000 000 000 000 000 000 000 000 000+000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000F
0 0 0 0 0 00 00 00 00 00 00 00 00 00+
0 0 0 0 0 0 0F
0 0 0 0 0 0 00 00 00 00 00 00 00 00 00+00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00+00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 000 000 000 000 000es
00 00 00 00 00 00 00 00 00 000 000+000 000 000 000 000 000 000 000 000 000 000th
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00+00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
Figure 4.1: Pictorial representation of forces acting on

trapped particles. Fg is the gravitational force, Fnd and Fid
the neutral and ion drag forces, Fes the electrostatic force
and Fth the thermophoretic force.
65
Chapter 4
Child-Langmuir sheath theory states that the steady state positive ion and
electron flux to the surface of an insolated body suspended in a plasma has to be
equal. To maintain equal electron and ion fluxes, the body gains a negative charge
and becomes surrounded by a sheath of positive space charge (see Fig. 4.2).
a.
b.
Isolated wall
Vp
Thin
Sheath
Vf
Vp
Thick
Sheath
-Vf - -
Figure 4.2: a. Thin sheath model where object is larger than sheath
dimensions and therefore assumed to be planar and b. Thick sheath
model for small objects in the orbit motion limited (OML) regime, note
that some of the electrons are not collected.
For a particle with radius (a) greater than the Debye length (D) the thin or
planar sheath model is used (Fig. 4.2a). In the model, ions entering the sheath are
assumed to have been accelerated to the Bohm velocity in a pre-sheath region and,
assuming a Maxwellian energy distribution function, the potential at the body
surface (Vf) with respect to the bulk plasma (Vp) is,
V f Vp =
k B Te mi
+ 1
ln
2e 2me
(4.1)
where mi and me are the ion and electron mass and ions are first accelerated by
kBTe/e in the pre-sheath region to the Bohm velocity. For argon this gives Vf - Vp
= -5.2 (kBTe/e), and hence for Te = 1 to 3 eV gives Vf - Vp = -5.2 to -15.6 V. A full
derivation of the planar sheath model is treated in standard texts [28, 29]. For very
small particles however (D >> a) not all incident ions are collected by the particle
and Orbit Motion Limited Theory (OML) is needed. This is pictorially represented
in Fig. 4.2b.
Using OML theory the ion current Ii (assuming Maxwellian ion distribution)
and electron current Ie (assuming Boltzmann distribution) to the particle surface
are,
66
Chapter 4
1
2
8k T e(V f (a ) V p )
I i = a 2 ni e B i 1
k B Ti
m i
(4.2)
e(V f (a ) V p )
8k T 2
I e = a 2 ni e B e exp
k B Te
m e
(4.3)
where ni is the ion density in the plasma bulk, Ti, the ion temperature, kB the
Boltzman constant and Vf(a) the potential of the particle. At steady state Ie = Ii
hence,
1
e(V f (a ) V p ) Ti me 2 e(V f (a ) V p )
exp
= T m 1
k B Te
k B Ti
e i
(4.4)
The normalized floating potential [e(Vf(a)-Vp) /kBTe] depends only on the

electron to ion mass and temperature ratios. The numerical solution for e(Vf(a)-Vp)
/kBTe (Equation 4.4) as a function of Ti/Te for Ar ions is plotted in Fig. 4.3.
-eVf(a)-Vp / kBTe
4
3
2
1
0.01
0.1
(Ti / Te)
Figure 4.3: Particle potential as a function of the ion/electron

temperature ratio calculated for an ion mass of 40 (Ar).
From Fig. 4.3 we observe that the potential Vf-Vp at the particle surface (for a
specific ion mass) depends only on the relative temperature of the ions and
electrons and the particle radius.
Using the fit parameters form Fig. 4.3, the potential of the dust particle can be
calculated (Eq. (4.4)), assuming the particle is isolated and has spherical geometry.
The number of charge carriers is then calculated from the capacitance,
67
Chapter 4
C=
Z De
Q
=
= 4 0 a
V f Vp V f Vp
(4.5)
where 0 is the permittivity of free space, Q the value of the particle charge and ZD
the number of charge carriers (electrons) on the surface of the particle. Applying
OML theory to a typical capacitively coupled plasma with an electron temperature
of 3 eV, a density of 1015 m-3 and neutral gas temperature of 398 K, gives a particle
potential of approximately -7.4 V, which is lower than that derived from the planar
sheath model under the same conditions. The electrostatic force (Fes) acting on a
particle is then,
Fes = QE = Z D eE
(4.6)
with E is the electric field vector that points from the bulk plasma towards the
powered electrode. The magnitude of E can be very large in a parallel plate reactor
configuration as the electrode typically reaches a self bias potential of several tens
of volts and is separated from the bulk plasma by a sheath which is typically a few
millimetres thick. The resulting electric field is then several tens of thousands of
volts per meter as was shown in a recent study of the electric field in an r.f.
discharge using fluorescence dip spectroscopy where fields as high as 105 V/m
were measured [30].
In addition to the electrostatic force, positively charged ions travelling from
the plasma bulk towards the electrode can also collide with the dust particles (see
Fig. 4.1). This induces an ion drag (Fid) force that acts to push the particles towards
the electrode. The ions pick up energy as they are accelerated through the sheath
and some of this energy is converted into heat when the ions collide with the
electrode. Heating of the electrode via ion bombardment produces a temperature
gradient and a consequent thermophoretic force (Fth) that acts to push particles
away from the hot electrode. This assumes that gas heating occurs via collisions
with the hot electrode and that heating of neutral species through interaction with
charged particles is negligible. The thermophoretic and ion drag forces are
calculated using Eqs. (4.7) and (4.8) respectively.
32 a 2
Fth =
15 vth , n
Fid =
FidO
FidC
1 + 32 (1 ) T Tn
(4.7)
eV f (a ) V p
eV f (a ) V p
61.32
2
+
a
+
n
k
T
a
n
k
T
ln
1
B e 1 +
i B e
k B Te
k B Te
2
(4.8)
2
where T is the translational part of the thermal conductivity of the gas, ni is the
density of the unperturbed plasma, vth,n the thermal velocity and Tn the temperature
68
Chapter 4
of the neutral gas. The parameter is calculated assuming the ion velocity i is
equal to the Bohm velocity using Eq. (4.9),
Z De2
=
1
4 0 ( m i2 ) D
2
(4.9)
In addition to the electrostatic, ion drag and thermophoretic forces, the gas
flow also produces a neutral drag force (Fnd), and as the particles have mass they
will experience a gravitational force Fg (cf. Fig 4.1). These forces are expressed as,
Fg = 43 a 3 p g
(4.10)
(4.11)
Fnd =
a 2 mn nn th , n H (s )( D n )
where H(s) is a function of s only and is defined by,
1
1
H (s ) = 1s s + exp s 2 + s 2 + 1 2 erf (s )
2s
4s
( )
(4.12)
and s is the ratio,
s=
D n 2
th ,n
(4.13
p is the particle density, g the gravitational constant, mn the mass, nn the number
density of the neutrals and erf(s) the error function of s. D and n represent the
mean velocity of the dust particle and gas flow respectively. Fnd is calculated
assuming the kinetic or long mean free path regime. This regime is established
by a high value of the Knudsen number Kn which is defined as the ratio of the mean
free path of the neutral gas molecules to the dust particle radius (Kn = mfp/a). For
low values of Kn (high pressure hydrodynamic regime) the drag force can be
obtained from the Stokes law and is proportional to the dust particle radius and
velocity. For low pressure processes mfp is in the order of a few hundred
micrometers and a is a few m, Kn is much larger than unity and the kinetic regime
is chosen where the principles for the derivation of the neutral drag force are the
same as those for the ion drag (see Eq. (4.8)). The gas velocity used for
determining the magnitude of Fnd was taken from a fluid dynamic model of the
reactor (see Fig. 4.8.) As the particle radius decreases the electrostatic force
dominates pushing smaller particles to regions of low electric field (high plasma
potential), usually situated in the centre of the plasma glow.
69
Chapter 4
All of the forces depend on either the particle size/mass (as the density is
assumed to be constant). The electrostatic and thermophoretic forces depend on Te
and the temperature gradients inside the reactor respectively. The downwards
forces Fg and Fid depend on the particle mass and plasma density (ne) while the Fnd
depends on the gas pressure and velocity and acts in the direction of the flow.
-9
10
-10
10
Fes
Force (N)
Fth
Fg
-11
10
-12
10
Fid
-13
10
-14
10
Fnd
-15
10
-16
10
-17
10
10
15
20
Particle size (m)
25
30
Figure 4.4: Calculated values for forces acting on dust particles trapped in an
r.f. discharge. Fg is the gravitational force, Fnd and Fid the neutral and ion
drag forces, Fes the electrostatic force and Fth the thermophoretic force.
The relative magnitude of these forces is plotted as a function of particle
radius (a) in Fig. 4.4. The electrostatic force has a linear dependence on the electric
field, the magnitude of which varies greatly in the plasma [30]. Bulk plasma is
considered quasi-neutral and therefore the electric field in this region is relatively
small. In the sheath regions however, large electric fields exist to confine the highly
mobile electrons [30]. In this study, self biasing of the driven electrode plate was
observed to reach 150 V during particle synthesis experiments. Assuming a neutral
gas temperature of 400 K, Te = 3 eV and ne = 1015 m-3 the sheath above the driven
electrode is calculated to be 8.3 mm, which is within the experimentally observed
estimation of between 5 to 10 mm (from the electrode surface to the beginning of
the glow). This gives an electric field above the electrode of approximately 18000
V/m* and consequently the electrostatic force, Fes is calculated to be 7.410-13 N for
100 nm particles. Under the same conditions, and assuming a bulk density of SiO2
*
This is comparable to measurements performed by Takizawa [30] in an r.f. discharge.
70
Chapter 4
to be 2000 kg/m3, the total force contribution from Fg and Fid, is equal to 6.5110-16
N. The enormous electric field in the plasma sheath therefore represent an
insurmountable barrier for the particles and so they become trapped between the
plasma glow and the sheath edge.
The electrostatic force is the most dominant for small particles. As the particle
radius increases the gravitational force begins to dominate and it is not until the
particle diameter reaches 27.2 m that the particle mass becomes large enough for
the particle falls through the electrode sheath (Fig. 4.4). An overall balance of the
up and down forces can be seen in Fig. 4.5.
-9
10
27.2m
Total up force
-10
Force (N)
10
Total down force
-11
10
-12
10
-13
10
-14
10
10
15
20
25
30
35
40
Particle Diameter (m)
Figure 4.5: Balance of forces show that particle with diameter greater
than 27.2 m will be expelled from the plasma.
However, particles considerably smaller than 27 m are observed buried in
heaps of particles observed to form on the electrode during plasma operation.
During preliminary particle forming experiments several unusual structures were
observed on the driven substrate table under certain flow conditions. As the total
gas flow rate entering the reactor decreased, the distribution of particles inside the
reactor varied. At relatively high flow rates (> 200 sccm), particles were seen to
heap in the centre of the electrode and form around the edges of the cage and
conical inlet. At low flow rates (< 56 sccm) particles formed mushroom-like
structures in the centre of the electrode (Fig. 4.6).
71
Chapter 4
Figure 4.6: Mushroom structures deposited onto a 10 cm silicon wafer

during a particle synthesis experiment. Experimental conditions were,
TEOS = 11.2 sccm, O2 = 28 sccm, Ar = 16.8 sccm, Power = 300 W,
Pressure = 1 Torr.
A possible explanation for the observed unusual structures might be as
follows. In order for the particles smaller than 27.2 m to traverse this sheath the
repulsive electrostatic force must be reduced. For Fes to be reduced either the field
in the sheath must be reduced for a time long enough for the particles to cross, or
the particle charge must be reduced. The influence of the various plasma
parameters alone is also not sufficient to explain this phenomenon. The plasma
density only affects Fid, and even if the calculation is performed taking Te = 1 eV,
the forces still balance at a particle diameter of 17.6 m. Figure 4.1 already holds
some suggestions as to how the particles trapped on the sheath edge might
overcome the sheath potential. In the centre of the plasma discharge the particle is
completely surrounded by a positive space charge as described by Eqs. (4.2) to
(4.5). As the particle approaches the electrode sheath, the particle sheath and
electrode sheath can combine and part of the particle surface is shielded from
electrons in the plasma glow. This causes the net electron flux to the particle
surface to drop and consequently the net particle charge to drop. As the particle
charge drops, so does Fes and the particle descends still further towards the
electrode loosing more charge until eventually the gravitational and neutral drag
forces cause it to fall onto the substrate table. In addition to shielding from the
electrode, there are also shielding effects from other particles. As particles
accumulate in the electrode sheath, the OML sheaths around each particle begin to
overlap providing additional shielding from the plasma glow. This process is
represented schematically in Fig. 4.7.
72
Chapter 4
00 0 0 0 0Particle
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0Equal
00000000000000000000000
a):
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00grows.
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00&
00 00 00 00 00
0
ion
to
particle
in
plasma.
0 00 00 00 00 00flux
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
00 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 000 000 000 000 000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0 0 0 0 0 0 0 e0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0e0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000e000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0e0 0 0 0 0 0 0 0 0
00 0 0 0 0 0 0 0Reduced
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0as
00000000000000000000
b):
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00flux
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00particle
00 00 00 00 00 00 00 00 00 00 00 00 00 00
0nears
0 0 0 0 0 0 0 0 0 0 0 0 0electrode
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 sheath.
0000000000000000000000000000
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000e000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000e000 000 000 000 000 000 000 000 000 000 000
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00
0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000e000 000 000 000 000 000 000 000 000 000000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000000 000 000 000 000 000 000 000 000 000 000 000 000 000000 000 000 000 000 000 000 000 000 000 000 000
00 0 0 0 0 0 0 e0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0e0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0Electrode
00000000000000000000000000000000000000000
00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0Electrode
00000000000000000000000000000000000000000
00 00 00 00 00 00 00 Reduced
0 0 0 0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00F00 00 00es00 00 00 00 allows
0 0 0 0 0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
c):
0particle
00 0 00 00 00 00 00 000 000 000 000 000 000 000 000 000 000 000to
00 00 00 00 00 00penetrates
00 00 00 00 00 00 00 00 00 00 00 000 000 000 000 000 000 000 000 000 000 000 000sheath.
000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 e0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0e0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 0000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00e00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0e0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 e0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00Electrode
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Figure 4.7: Particle loss mechanism.

The loss of particle charge as they enter the electrode sheath, and hence
reduction of Fes, implies that the large electric field in this region is not responsible
for confining the particles in the plasma. Assuming the electric field acting on the
charged particles is just above the electrode sheath (where the particles are still
immersed in the plasma), and that particles stop growing once the particle and
electrode sheath start to combine. Balancing the up and down forces gives,
Fdown = Fup = Fg + Fid + Fnd = Fes + Fth = QE + Fth
(4.14)
therefore,
Fg + Fid + Fnd Fth = Fes = QE
(4.15)
re-arranging to give
E=
( Fg + Fid + Fnd Fth )

Q
(4.16)
Using Eqs (4.2) to (4.5), and a particle diameter of between 100 and 200 nm
(taken from ESEM images of particles collected after experiments) the confining
electric field, E, becomes between 3.72 and 1.65 V/m. This field strength is
associated more with the plasma bulk than the sheath region, suggesting that the
particles are trapped just above the electrode sheath.
The particles or particle clusters are experimentally observed to breach the
sheath in the centre of the electrode to form heaps of particles on the substrate.
An understanding of the forces on the particles, combined with a fluid model of the
reactor geometry can explain the different particle deposits at different flow
conditions. From Fig. 4.8 it is evident that the position at which the particle heaps
73
Chapter 4
and mushroom structures form coincides with a stagnation point in the flow profile
where Fnd is perpendicular to the electrode surface. The very low flow, and hence
long residence time, allows the particles to gradually loose their charge and traverse
the sheath as described in Fig. 4.7. At higher flows the stagnation region is large
and particles fall through a larger area of the sheath to form particle heaps. The
stagnation point is smaller at low flows and a narrow column forms and grows
vertically through the sheath as particles continue to breach the sheath-plasma
boundary. Once the tip of the column reaches the sheath edge where the particles
are suspended, particles begin to collide with the sides of the column tip and the
structure begins to grow horizontally in all directions forming a mushroom shape.
The importance of the stagnation point is evident from the fluid models shown in
Figs. 4.8 and 4.9. In Fig. 4.9 the precursor and carrier gas mixture is injected into
the side of the plasma and the stagnation point in the centre of the electrode no
longer forms. When this configuration was tried experimentally, no particle heaps
or mushroom structures were observed.
1.010-3
Inlet flow
8.010-4
6.010-4
4.010-4
2.010-4
Inlet flow from

above
0.010-5
1.810-5
Figure 4.8: Fluid model

of reactor showing
contours in the velocity
magnitude (m/s). Note,
stagnation point in
centre of the electrode.
1.4510-5
1.110-5
7.510-6
3.810-6
1.710-7
74
Chapter 4
1.010-3
8.010-4
6.010-4
4.010-4
Inlet flow
2.010-4
0.010-5
2.610-4
Figure 4.9: Fluid model of

reactor showing contours in
the velocity magnitude
(m/s). Note, stagnation
point no longer exists in
centre of the electrode.
2.110-4
1.610-4
1.110-4
5.410-5
5.510-7
To summarize, we have looked at the influence and relative magnitude of the

various forces acting on particles trapped in a low pressure r.f. discharge. The
calculations suggest that the large electric field generated by the electrode sheath
represents an almost insurmountable barrier for particles trapped in the plasma
glow. Particles however are observed to penetrate this sheath at sized much smaller
than predicted by the calculations. This observation is accounted for by shielding
effects that occur when the particles approach the electrode sheath, and at high
particle number densities. The formation of mushroom structures observed on the
electrode under low gas flows is therefore a consequent of the stagnation point in
the gas flow (vertical and positional contribution) combined with the confining
electric field above the electrode (responsible for horizontal growth at the
mushroom top).
4.3 Experiment
Particles were synthesized, trapped and monitored using a capacitively coupled r.f.
discharge as described in chapter 3 and Fig. 4.10. The steel substrate table acts as
the driven electrode and is surrounded by a grounded steel cage to contain the
plasma. No cooling systems were implemented and the substrate typically reached
a temperature of 100C during film growth or particle synthesis which typically
lasted about 10 minutes.
Infrared spectroscopic measurements were done using a Midac M2500-C
Fourier transform infrared spectrometer (FTIR) that consisted of two units. The
75
Chapter 4
detector unit housed the Mercury Cadmium Telluride (MCT) detector, while the
source unit housed the Michelson interferometer and glow bar source. The two
instrument units were positioned either side of the reactor and the infrared beam (
= 2 cm) passed over the driven electrode clearing the substrate table by 3 mm.
Spectra were recorded using a spectral resolution of 1 cm-1 and averaging 10
consecutive scans to improve the signal-to-noise ratio. Infrared spectra were deconvoluted using the Peak Fit (Systat Software) and heights of absorbance features
were used to monitor variation in species concentration.
Precursor inlet
Trapped particles
FTIR Detec tor
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Powered electrode
FTIR Source
Grounded cage
Figure 4.10: Schematic of particle synthesis and trapping reactor with

in-situ FTIR spectrometer.
The tetraethoxysilane precursor [CAS 78-10-4] 98 % purity was evaporated
and dosed using a Bronkhorst HiTec CEM system while argon and oxygen flows
were controlled using Bronkhorst HiTec mass flow controllers. All gas lines to the
reactor were heated to 80C to prevent condensation of the TEOS precursor. A
MKS throttle valve located in the reactor exhaust system regulated the reactor
pressure, which was constant during each experiment.
The key experiment to deposit hybrid films in which SiO2-like particles are
embedded in an organic matrix is as follows. Conditions from particle synthesis
and trapping experiments were combined with conditions optimized for growing aC:H layer. This was done by first synthesizing the particles, purging the precursor
gases while holding the SiO2 particles in the plasma sheath above the electrode and
then injecting methane. The methane would deposit on the substrate and particle
surfaces to form an amorphous carbon layer, in which the particles would hopefully
become embedded. A summary of the experiments and process conditions are
presented in Table 4.1.
76
Chapter 4
Table 4.1: Experimental condition for particle synthesis and trapping

experiments.
Experiment
Particle
synthesis
Mushroom
formation
Hybrid
simultaneous
Hybrid
sequential
Power
[Watt]
Fraction Fraction Fraction

TEOS
O2
CH4
Flow
[sccm]
Pressure
[Torr]
100-500 0.05-0.3 0-0.7
100-300 1
200-300 0.1-0.2
0.5
56
0.5-1
300
0.2
0.5
0.2
200
300
0.1
0.3
0.4
100
4.4 Infrared analysis

Particle formation in the capacitive plasma could be observed in-situ with FTIR
spectroscopy [31]. The decomposition products of TEOS in the predominantly
oxygen plasma were CO2 (2300 cm-1), CO (2200 cm-1), HCOOH (1776 cm-1), CH4
(2900 cm-1) and H2O or OH (3500 - 4000 cm-1 and 1500 - 2000 cm-1), and residual
TEOS (1000 - 1250 cm-1). The transversal optical (T.O.) vibrational mode
corresponding to solid SiO2 could also be identified at (1066 cm-1). By switching
the precursor gas flow off and keeping the plasma discharge on, particles could be
trapped a region close to the sheath above the powered electrode. It was found that
particles could be trapped for an indefinite time period in the argon discharge. The
solid SiO2 IR peak could be verified by comparison with an IR absorption spectrum
taken of particles collected after the deposition experiment (Fig. 4.11).
The calculated extinction of the infrared beam resulting from either bulk or
surface absorptions was simulated for SiO2 and SiO, assuming spherical particles
and optical constants obtained from the literature [32], using Eqs. C.1 to C.6 in
Appendix C, and plotted as a function of wavelength in Fig. 4.12.
It can be seen that the measured particle peak shows most similarity to the
bulk and surface SiO2 synthetic peaks. XPS measurements taken of particles
collected from after the experiment showed the particle composition to be almost
that of bulk SiO2 at 32.5 % Si, 67.5 % O and 0.4 % C, which gives O/Si of 2.1 and
0.4 % carbon impurity.
77
Chapter 4
1.0
Ex situ: FTIR of particles collected.
In situ: FTIR of particles trapped in plasma.
Absorption (arb. units)
0.8
SiO2
0.6
0.4
residual gas phase products in plasma
0.2
0.0
1000
1500
2000
2500
3000
3500
4000
4500
-1
Wavenumber (cm )
Figure 4.11: Comparison between FTIR spectra of particles trapped in a

TEOS free plasma discharge and particles collected on a c-Si wafer
afterwards.
Extinction (Arb. units)
1.0
As deposited
Surface SiO2
0.8
Bulk SiO2
Surface SiO
Bulk SiO
0.6
0.4
0.2
0.0
600
700
800
900 1000 1100 1200 1300 1400

-1
Wavenumber (cm )
Figure 4.12: Comparison between calculated bulk and surface

vibrations for SiO and SiO2 with measured IR spectra of particles
synthesized.
78
Chapter 4
particle trapping
Ar/TEOS/O2 plasma
Ar plasma
particle formation
particle trapping
Ar/TEOS/O2 plasma
Ar plasma
particle expulsion
no plasma
particle
expulsion
no plasma
1.1
1
0.9
0.8
loss loss
of particles
of particles
0.7
3
[TEOS]u c [mole%]
particle formation
0.6
0.5
trapped
trapped
particles
particles
0.4
0.3
film
0.2
film
deposition 0.1
deposition
0
0
-1
TEOS (mole %)
Area (arb. Units)
area A [a.u.]
Particles trapped due to the high electric field in the sheath region above the
powered electrode were monitored with the FTIR spectrometer. Figure 4.13 shows
the growth, trapping and loss of particles from the plasma as a function of time as
monitored by the solid-state SiO2 peak as shown in Fig. 4.11. The plot is divided
into three sections representing particle nucleation, trapping and expulsion from
plasma stages of the experiment. During particle nucleation the SiO2 peak increases
sharply and reaches a maximum two minutes after the precursor flows are directed
into the reactor. The TEOS signal increase less rapidly during particle nucleation as
the TEOS is being consumed in particle formation and growth reactions. After 2
minutes the first particles grow large enough to leave the plasma glow and an
equilibrium is reached between particle formation/growth and particles dropping
out of the infrared beam and onto the substrate. At four minutes the precursors
flows are switched off and only argon is dosed into the discharge. A drop in the
SiO2 absorption peak at this point indicates loss of some particles from the plasma
sheath. On switching the precursors flows off, the plasma conditions change from
that of an electro-negative discharge caused by the O2 with low ne caused by the
TEOS to an electro positive with higher ne. This particle loss is attributed to a
change in the plasma conditions as argon gradually replaces the precursor gases
(possibly a result of higher ion drag).
Time (min.)
10
time [min]
-0.1
-0.2
Figure 4.13: Time dependent particle measurements of particle trapping

using FTIR spectroscopy in a TEOS/O2 plasma.
A bulk of the particles remains trapped above the powered electrode until the
r.f. field is switched off after nine minutes. At this point all remaining particles
drop out of the infrared beam and onto the electrode. The residual SiO2 signal
79
Chapter 4
observed between nine and eleven minutes is due to deposition of material on the
KBr windows inside the reactor.
4.4.1 Rayleigh scatting
The particle number density for a typical plasma experiment, as shown in Fig 4.13,
can be estimated from the absorption data using Eqs. (C.1) to (C.12) presented in
Appendix C. ESEM images of particles collected after an experiment (Fig 4.14)
give a particle radius of approximately 100 nm. Combining this estimation of the
particle size with the SiO2 absorption peak centred at 1066 cm-1 provides an
estimation of the particle density np (Eq. (C.2) to (C.6) Appendix C), np of
approximately 1014 m-3. This also explains why no scattering signal is observed for
the particles synthesized in the pure TEOS/O2 plasma. Extinction of the infrared
radiation when passing through a particle cloud of np = 1014 m-3 particles of a = 100
nm gives A(1066 cm-1) = 0.003. This value is smaller than the signal-to-noise ratio of
the spectra recorded using a sample rate of 1 scan/s and averaging 10 consecutive
scans, and hence no scattering signal could be detected. In addition to this, drift of
the baseline signal occurred within the 12 minute time scale of the experiment,
preventing the possibility to increase the signal acquisition time and improve the
sensitivity.
Rayleigh scattering was observed however, in an experiment originally
designed to deposit SiO2 particles in an amorphous carbon layer (as described in
the experimental section). Particles synthesized in a TEOS/O2 plasma and
suspended in a methane plasma grew to a size where significant light scattering
could be observed. Fig. 4.15 shows the FTIR spectra recorded of the particles
trapped in the methane plasma and two simulated lines calculated using the method
described in Appendix C. The broad absorption feature centred at 2950 cm-1 is
attributed to an amorphous carbon layer deposited on the particles.
Figure 4.14: Particle size distribution measured by

ESEM image taken of particle clusters.
80
Chapter 4
There is some uncertainty as in the literature as to exactly how the Rayleigh

scattering model should be applied to infrared data exhibiting such baseline tilts.
Courteille et al. [15] used a wide spectral range from 1000 to 4000 cm-1 to fit the
baseline and derive values for the particle size and number density despite apparent
deviation from the observed spectrum at 3000 cm-1. Stoffels et al. [21] however,
applied the Rayleigh equations to a narrower spectral region, between 1000 and
3000 cm-1 to achieve a better correlation with the data. Deviations at higher
wavenumbers were attributed to a breakdown of the model assumptions (namely
2a/ << 1 [33]) and at higher wavenumbers a Mie scattering model should be
applied. Both of these approaches have been were investigated in this study and a
comparison of the results is presented in Table C.1 in Appendix C.
The calculations presented in Appendix C assume that the particles are all of
equal size and spherical. The optical constants for amorphous carbon are not
exactly known in the literature, due to the strong dependence of the optical
properties of these layers on the deposition conditions and the large variety of
deposition conditions reported for this material. The particle size calculations
presented here use values for a-C:H reported by Savvides et al. [34] and Pawlak et
al. [35] of n = 6.1 + 3.4i. ESEM image taken of purely SiO2 particles collected after
a light scattering experiment indicate the SiO2 particle radius (a1) to be between 50150 nm (Fig. 4.14).
IR Absorption (arb. units)
0.020
Recorded spectra
Simulated spectra
0.016
0.012
0.008
0.004
0.000
1000
2000
3000
4000
5000
6000
-1
Wavelength (cm )
Figure 4.15: Infrared spectrum recorded during the deposition of a

nanocomposite material, along with simulated curves for Rayleigh
scattering.
4.5 Physical characterisation of particles
Thermal gravimetric analysis (TGA) is a useful technique that provides information
regarding the thermal stability and to some extent structure of a material. The
measurement is performed by placing a sample of known mass in an oven and
81
Chapter 4
heating under a controlled atmosphere. The thermal stability is determined by

monitoring variations in the sample mass under either inert (Ar or N2) or oxidizing
(O2 or O2/N2) atmospheres. In addition to monitoring variation in sample mass,
monitoring the energy absorbed by the sample at different temperatures can also
indicate variation in the material structure. Results from TGA can help predict how
a material will react during post deposition plasma or thermal processing as is
typically encountered when integrating a new material into semiconductor process.
TGA was performed on particles collected after film deposition in both pure N2 and
20 % O2 in N2 atmospheres, using a temperature ramp rate set to 5C/minute. Both
measurements indicate the particles to be thermally stable up until temperatures in
excess of 800C. A 14 % weight loss was observed for both measurements
indicating no further oxidation of the particles took place in the oxidising
atmosphere. A positive gradient in the temperature differential is indicative of
exothermic reactions within the sample, and a negative gradient indicates
endothermic chemical changes. The initial positive gradient is rather confusing as it
suggests an exothermic reaction is occurring during the initial heat up. A possible
explanation is that during this period any Si-OH groups in the sample might react to
liberate water and form SiO2. The evaporation of water from the sample would be
an endothermic process, but the overall formation of SiO2 and H2O would be
exothermic. The sample showed no signs of blackening or decomposition after
TGA measurements, again indicating that the plasma produced particles are very
stable.
% Weight loss O2
% Weight loss N2
100
80
1
0
60
Temp. Diff. ( C/mg)
Weight change (%)
-1
40
-2
Temp. Diff. O2
20
Temp. Diff. N2
0
200
400
600
800
1000
-3
-4
Temperature C
Figure 4.16: TGA analysis taken in pure N2 and Air of powder collected
from Zone I after nanocomposite layer synthesis. Note the initial positive
slope which might indicate hydroxyl loss.
82
Chapter 4
4.6 Discussion and conclusions

The synthesis, trapping of particles from a TEOS/O2 gas mixture was successfully
demonstrated using a capacitively coupled discharge. The high thermal stability
suggested by TGA, coupled with the dominance of Si-O groups in the infrared
suggests the particles are silica-like. However, it is likely that the particles are
highly porous. Porosity is known to exist in plasma deposited films deposited from
siloxane precursors such as TEOS [36] and HMDSO [37] and high bias potential
are used to increase film density (as demonstrated in Fig. 3.6). The particles
however, are formed from negative ions trapped in the plasma glow and the sheath
surrounding them, and hence the energy of ions impinging on them is much lower
than in the sheath region above the electrode (cf. section 4.2). As the formation of
dense fully oxidized SiO2 networks requires high energy input, (as seen with SiO2
deposition in chapter 3) the energy is not available to form dense particles. In
addition to this, particles collected post-deposition coagulate into extremely light
structures that are easily affected by electrostatic forces. Small angle x-ray
diffraction (SAXS) measurements also show no long range order within the particle
structure. All these observations suggest the particles consist of an extremely
porous SiO2 network.
Particle number density and size could be estimated from in situ absorption
FTIR measurements. The formation of particle heaps, and mushroom structures in
the centre of the powered electrode was discussed in the context of OML theory
combined with a fluid dynamic model of the reactor geometry used.
Although incorporating porous particles into silicone layers is unlikely to offer
any improvement in mechanical or barrier properties, the controlled incorporation
of porosity is interesting for low dielectric constant materials. Two limitations in
the current reactor design have been identified from this work. Firstly the
stagnation point in the current reactor geometry prevents the homogeneous
distribution of particles throughout the growing layer. Secondly, the use of a single
plasma source prohibits the simultaneous synthesis of a SiO2 and silicone layer as
the two chemistries cannot be separately controlled. Using a modified reactor
design, chapter 6 investigates the deposition and characterisation of nano-porous
layers for low dielectric constant applications.
Acknowledgement
I would like to thank A. Klinkenberg for his CFD modelling work.
83
Chapter 4
References
[1] Terranova M.L., Piccirillo S., Sessa V., Rossi M., Botti S., J. Phys. IV France, 9, 365371, (1999).
[2] Senkevich J.J., Seshu D.B., Chem Mater. 11, 1814-1821, (1999).
[3] Suzuki F., Onozato K., J. Appl. Polym. Sci., 39, 371-381, (1990).
[4] Sohoni G.B., Mark J.E., J. Appl. Polym. Sci., 45, 1763-1775, (1992).
[5] Sawyer W.G., Freudenberg K.D., Bhimaraj P., Schadler L.S., Wear, 254, 573-580,
(2003).
[6] Hurt R.H., Allendorf M.D., AIChE. J., 37, 1485-1495, (1991).
[7] Tandon P., Rosner D.E., AIChE J., 42, 1673-1684, (1996).
[8] Dusty Plasmas, Bouchoule A., Wiley, New York, (1999).
[9] Stoffels E., Stoffels W.W., Kroesen G.M.W., Hoog F.J.de., J. Vac. Sci. Technol. A,
14, 556-561, (1996).
[10] Boufendi L., Hermann J., Bouchoule A., Dubreuil B., Stoffels E., Stoffels W.W.,
Giorgi M.L. de., J. Appl. Phys., 76, 148-153, (1994).
[11] Cabarrocas P.R.I.,Gay P., Hadjadj A., J. Vac. Sci. Technol. A, 14, 655-659, (1996).
[12] Dorier J.-L., Hollenstein Ch. Howling A.A., J. Vac. Sci. Technol. A, 13, 918-926,
(1995).
[13] Watanabe Y., Shiratani M., Plasma Sources Sci. Technol., 3, 286-291, (1994).
[14] Hollenstein Ch., Howling A.A., Courteille C., Magni D., Scholz S.M., Kroesen
G.M.W., Simons N., Zeeuw W de., Schwarzenback W., J. Phys. D: Appl. Phys., 31,
74-84, (1998).
[15] Courteille C., Magni D., Deschenaux Ch., Fayet P., Howling A.A., Hollenstein Ch.,
Soc. Vac. Coat. 41st Annual Technical Conference Proc., (1998).
(1999).
[17] Optical Studies of micron-sized particles immersed in a plasma, Swinkels G.H.P.M,
PhD. Thesis, Technische Universitiet Eindhoven, (1999).
[18] Boufendi L., Bouchoule A., Plasma Sources Sci. Technol., 3, 262, (1994).
[19] Hollenstein Ch., Dorier J.L., Dutta J., Sansonnens L., Howling A.A., Plasma Sources
Sci. Technol., 3, 278, (1994).
[20] Dorier J.L., Hollenstein Ch., Howling A.A., J. Vac. Sci. Technol. A, 13, 918, (1995).
[21] Stoffels E., Stoffels W.W., Kroesen G.M.W., Hoog F.J. de, Electron Technology, 32,
255, (1998).
[22] Buss R.J., Hareland W.A., Plasma Sources Sci. Technol., 3, 268-272, (1994).
[23] Steinbruchel C., Yoo W.J., Plasma Sources Sci. Technol., 3, 273-277, (1994).
[24] Stoffels W.W., Stoffels E., Kroesen G.M.W., Haverlag M., den Boer J.H.W.G., de
Hoog F.J., Plasma Sources Sci. Technol., 3, 320-324, (1994).
[25] Kamata T., Kakuta S., Yamaguchi Y., Makabe T., Plasma Sources Sci. Technol., 3,
310-313, (1994).
[26] Amemiya H., J. Phys. D: Appl. Phys., 23, 999-1014, (1990).
[27] Suh S.-M., Zachariah M.R., Girshick S.L., J. Vac. Sci. Technol. A, 19, 940-951,
(2001).
[28] Introduction to Plasma Physics and Controlled Fusion, Vol., I Chen F.F., Plenum
Press, New York, (1984).
[29] Fundamentals of Plasma Physics, Bittencourt J.A., Pergamon Press, Oxford, (1986).
[30] Takizawa K., Sasaki K., Kono A., Appl. Phys. Lett., 84, 185-187, (2004).
84
Chapter 4
[31] Courteille C., Magni D., Howling A.A., Nosenko V., Hollenstein C., Soc. Vac. Coat.
40th Annual Technical Conference Proc., (1997).
[32] Handbook of optical constants of solids, Palik E.D., Academic Press, London,
(1998).
[33] Light scattering by small particles, van de Hulst H.C., Chapman & Hall, New York
& Dover, (1981).
[34] Savvides N., J. Appl. Phys., 59, 4133, (1986).
[35] Pawlak F., Balanzat E., Dufour Ch., Laurent A., Paumier E., Perriere J., Stoquert J.P.,
Toulemonde M., Nucl. Instr. Meth. In Phys. Res. B, 122, 579-582, (1997).
[36] Jeong S.H., Nishii J., Park H.-R., Kim J.-K., Lee B.-T., Surf. Coat. Technol., 168, 5156, (2003).
Process., 5, 297-284, (2002).
85
Appendix C
Infrared analysis of particles
Appendix C Infrared analysis of particles

A molecule can absorb radiation if the energy of the photon matches the transition
between energy states in the molecule. For vibrational energy transitions, the
energy difference corresponds to wavelengths in the infrared region of the
electromagnetic spectrum. The selection rules for the transition state that there must
be a change in the molecules dipole moment during the transition. Therefore,
almost all except homonuclear diatomic molecules absorb infrared radiation,
making IR spectroscopy ideal for identifying chemical systems. For a detailed
description of infrared spectroscopic methods see Griffiths [1] or Pelikan [2]. The
application of the FTIR absorption spectroscopy technique for monitoring particles
trapped in a plasma sheath is presented here.
Infrared spectroscopy can be used to quantify constituent concentrations in a
sample by application of the Beer-Lambert law (Eq C.1) in which the absorbance A
of a species is related to the ratio of the intensity of the incident (I0) and transmitted
(I) light, sample path length (l) and concentration (c),
A( ) = log
I ( )
= ( )lc
I 0 ( )
(C.1)
where is the coefficient of absorption. If scattering can be neglected then, the

absorption of infrared radiation by SiO2 particles can be used to determine particle
number density. The shape and intensity of Si-O absorption features centred at
1066cm-1 differs depending on whether the sample consists of a single bulk of SiO2
or as fine powder dispersion. Using bulk values for the optical constants n and k,
the extinction of the IR beam due to absorption can be calculated assuming
spherical particles using Eq. (C.2),
A( ) = log
I ( )
= e 4k d
I 0 ( )
(C.2)
where is the wavenumber (cm-1), k the extinction coefficient of bulk SiO2 and d is
a parameter relating to the effective combined thickness of material (particles)
traversed by the light beam. In this particular case where we consider a cloud of
particles suspended in the plasma, the material is represented by the thin cloud of
particles, with number density nP. The parameter d can be related to the particle
radius a and number density by the following derivation. The volume of the cloud
of particles within the infrared beam is given by,
Vcloud = rb2 d
(C.3)
where rb is the and diameter of the infrared beam in the plasma. The SiO2 particles
are assumed to be spherical and hence Vparticle is given by,
86
Appendix C
4
3
V Particles = a T
3
(C.4)
where aT is the particle radius (the reason for the subscript T will become evident
later in Fig C.1). The number density nP can now be calculated based on the
volumes calculated above:
nP =
Vcloud
3d
=
Vbeam V particle 4lb aT 3
(C.5)
where,
Vbeam = rb2 l b
(C.6)
and lb is the length of the infrared beam. Consequently, nPaT3 can be given as a
function of the parameter d, which is a constant for one simulation:
3
n P aT =
3d
C1
4lb
(C.7)
where C1 is a constant. d is determined from fitted simulated spectra using SiO2

bulk values for n and k (Eq. C.2).
In addition to absorption, if the particles trapped in the infrared beam are large
enough (2aT >> ) they will also lead to Rayleigh and Mie scattering. This can be
seen as a tilting of the baseline in the spectrum that becomes stronger at higher
wavenumbers as both Rayleigh and Mie scattering are wavelength dependant. The
extinction of light due to scattering can be calculated by defining the scattering
cross section Csca. From Eq. C.2,
A( ) =
C sca n P l b
ln 10
(C.8)
where,
C sca =
128
3
5 aT 6
4
n~ 2 1
n~ 2 + 2
(C.9)
where aT is the particle radius, the wavelength of incident light and n~ the
complex refractive index of the particle. This theory assumes particles are spherical
and of sufficiently low number density that scattering is independent, i.e. distance
between particles is larger then three times the particle radius. Equation (C.9)
reflects the well known 4 dependence of Rayleigh scattering.
87
Appendix C
At this point we have two expressions (C.5 and C.8) and two unknowns (aT
and nP). It is therefore possible to determine both the particle diameter and density
provided we have both scattering and absorption data. Unfortunately no scattering
was observed during the synthesis of pure SiO2 particles. In order to solve for nP, aT
was estimated from ESEM images taken of particles collected after the experiment
(cf. Fig 4.14). Taking aT to be between 100-150 nm, nP is calculated to be
approximately 1014 m-3.
For cases where the particle is composed of more than one material (e.g. a
coated particle), the expression (Eq. C.8) is modified so that aT becomes the total
particle radius (Fig. C.1).
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
a-C:H coating
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
SiO2 core
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 a1 a
T
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 a
2
Figure C.1: Structure of SiO2 particle coated

with a-C:H layer.
n~ is replaced with the effective refractive index approximation n~eff as determined
by Eq. (C.10), the so called Bruggeman effective medium approximation [3].
V
V
n~eff2 = 1 n~12 + 2 n~22
Vtot
Vtot
(C.10)
where V1 and V2 are the volumes of materials with refractive indices n~1 and n~2 .
The total extinction due to Rayleigh scattering is therefore,
A( ) =
128
3
2
5 l b n~eff 1
n P aT6 4 = C 2 4 ,
2
~
ln 10 n eff + 2
(C.11)
where C2 is a constant and aT = a1 + a2. Because C2 can be determined by

simulating the baseline tilt, n P aT6 can be determined from this:
n P a26 =
88
C2
~2
5 neff 1
128 lb
3
ln 10 n~eff2 + 2
= C3 ,
(C.12)
Appendix C
where C3 is a constant.
The two equations derived (Eqs. C.9 and C.12) contain three unknown
parameters a1, a2 and n. However, with these equations, a2 can be determined as a
function of a1:
1
C 6 1
a2 = 3 a1 2 .
C1
(C.13)
and a2 is determined as a function of a1.

SiO2 particle coated with a-C:H did cause significant scattering and absorption
of the infrared radiation, but it is still not possible to solve due to the addition of a
new variable a2. Assuming a1 and nP from the pure SiO2 case we can estimate a2 to
be between 100 - 120 nm, giving a total particle diameter, aT, of to be between 200270 nm (see Fig. 4.14).
References
[1] Chemical Infrared Fourier Transform Spectroscopy, Griffiths P.R., Wiley & Sons, New
York,(1975).
[2] Applications of Numerical Methods in Molecular Spectroscopy, Pelican P., Ceppan M.,
Liska M., CRC Press, London, (1993).
[3] Petrik P., Lehnert W., Schneider C., Lohner T., Fried M., Gyulai J., Ryssel H., Thin Solid
Films, 383, 235-240. (2001).
89
90
Chapter 5
Reaction kinetics and mechanisms during PECVD from a TEOS/O2 gas mixture
Chapter 5: Reaction kinetics and

mechanisms during PECVD
from a TEOS/O2 gas mixture
Abstract
The electron impact dissociation of TEOS and reaction with atomic
oxygen radicals are investigated and compared with a view to elucidate
the TEOS decomposition mechanism in an r.f. capacitively coupled
discharge. Rate constants for a wide variety of electron impact and
neutral chemistry processes are calculated using data taken from the
literature. Dominant reactions for the decomposition of TEOS and
formation of hydrocarbons and CHOOH are identified and compared
with FTIR absorption and optical emission spectroscopy measurements
made during a TEOS/O2 PECVD process.
5.1 Introduction
The thermal and plasma chemistry of the siloxane/O2 deposition process is
complicated by the wide variety of compounds both silicon and carbon readily
react to form. However, these reactions are of considerable interest in the field of
organometallic chemistry and are vital to the understanding of the chemical vapour
deposition (CVD) and plasma enhanced chemical vapour deposition (PECVD) of
silicon, silica, silicon nitride and not least, silicone based hybrid thin films [1].
Determining the intrinsic kinetics and mechanisms of plasma driven CVD
processes using organosilicon precursors is not only complicated by the chemical
nature of the elements present in the precursor, but also due to the nature of the
plasma process itself. Plasmas are non-equilibrium systems that are typically
characterised by three temperatures, the electron temperature (Te) that governs the
electron impact kinetics, the ion temperature and the neutral gas temperature upon
which the neutral chemistry depends. For plasmas with a relatively low degree of
ionization, like the capacitively coupled plasma used in this study, the ion and
background gas temperature are usually very close and are often assumed to be
equal. The non-equilibrium nature of plasma processes results in a complex
chemistry where the global conversion of precursors into products and by-products
is the result of a complex interplay between these processes.
Intrinsic reaction kinetics and mechanisms are also effected by the fluid
dynamics and diffusion processes specific to any given reactor configuration. A
preliminary insight into these processes can be derived by correlating experimental
observation with accepted chemical or physical ideology. An example of this
would be to deriving a global activation energy by correlating growth rate with
91
Chapter 5
substrate or reactor temperature. Rudimentary reaction kinetics describing such

processes, such as precursor conversion to film with one rate constant (and hence
one step mechanism), are derived from such measurements. However, the
oversimplification of the chemistry in these models often limits their applicability
to the reactor system in which the measurements were made. A number of authors
have used this approach to investigate the tetraethoxysilane (TEOS) and oxygen
plasma system with sometimes contradicting findings [2, 3]. In an XPS and mass
spectroscopic study Fracassi et al. [4] concluded that the TEOS decomposition was
primarily the result of electron impact dissociation and that oxygen only slightly
influenced the deposition rate even when the O2/TEOS ratio was over 2.5. Other
authors report a more dramatic increase in deposition rate as the oxygen
concentration is increased, suggesting oxygen plays a more dominant role in the
TEOS decomposition [5, 6, 7]. In reality, the relative influence of oxygen on the
film deposition rate and process chemistry is a consequence of the electron energy
and energy distribution function, which in itself depends on the specific plasma
system.
More sophisticated models based on Monte Carlo simulations offer a greater
insight into the intrinsic processes and offer greater applicability at the cost of
computational expense. This approach was employed by Stout et al. [8] in a Monte
Carlo model of the TEOS/O2 plasma. The results of this model corroborated the
surface atomic oxygen mechanism proposed by Raupp et al. [5]. In addition to the
computation requirements, these models also rely heavily on thermochemical data
derived from either experimental or theoretical studies and is often difficult to find.
In this work, rate constants for a wide variety of reactions are calculated from data
taken from literature sources. This data is evaluated with the aim of deriving
mechanisms for the decomposition of TEOS and formation of several of the
observed by-products. Fourier transform infrared (FTIR) absorption spectroscopy
and optical emission spectroscopy (OES) are used to monitor precursor and byproduct concentrations and substantiate the mechanisms proposed. Data used for
the rate calculations presented here was obtained from work published by Allendorf
et al. [9, 10], Meeks [11], Ho [12] for the oxygen and argon chemistry, Holtgrave
et al. [13] Coltrin et al. [14], Morgan et al. [31] and Ho [15] for the electron impact
TEOS rates and various authors for the neutral hydrocarbon chemistry [16, 17, 18,
19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30].
5.2 Molecular processes in TEOS/O2 discharges
The relative importance of the following two possible initiation processes for
subsequent decomposition of TEOS is considered, namely, the electron impact
dissociation and the reaction with atomic oxygen as outlined in Eqs. (5.1) and (5.2).
e + Si(OC2H5)4 Products
(5.1)
Si(OC2H5)4 + O Products
(5.2)
Very little has been published regarding the collision cross section for electron
impact with TEOS. In an experimental study Holtgrave et al. [13] reported electron
92
Chapter 5
impact cross section of TEOS ionization reactions derived from Fourier-transform

ion mass spectrometry measurements. Their measurements suggested that TEOS
fragments preferentially to form ions of mass 149, 193, and 179 corresponding to
Si(OC2H5)2OCH3+, Si(OC2H5)3OCH2+ and Si(OC2H5)3O+ respectively. These
fragments are formed from the loss of ethyl, ethoxy and methyl radicals. They did
not observe the formation of negative ions and as neutral products cannot be
trapped, the Fourier-transform ion mass spectrometry technique cannot measure
them.
In a more recent study Morgan et al. [31] calculated the cross section for
electron impact excitation and dissociation for TEOS to form neutral products.
Results from both these studies suggest that the threshold electron energy for
dissociation or ionization is in excess of 10 eV and that electrons with lower energy
would only cause excitation. Electron impact dissociations mechanisms suggested
by Morgan, Holtgrave and Coltrin [14] are listed in Table 5.1.
Table 5.1: Electron impact dissociation mechanisms for TEOS.
Si(OC2H5)4 + e Si(OC2H5)3OCH2 + CH3 + e
Energy
Ref.
(eV)
3.6
[31]
Si(OC2H5)4 + e Si(OC2H5)3O + C2H5 + e
4.0
[31]
Si(OC2H5)4 + e Si(OC2H5)3 + OC2H5 + e
5.2
[31]
Reaction
Si(OC2H5)4 + e Si(OC2H5)2OCH3+ + OC2H5+ CH2 + 2e -
[13]
Si(OC2H5)4 + e Si(OC2H5)3OCH2+ + CH3 + 2e
[13]
Si(OC2H5)4 + e Si(OC2H5)3O+ + C2H5 + 2e
[13]
Si(OC2H5)4 + e Si(OC2H5)3(OH) + C2H4 + e
[14]
Morgan et al. reported only one rate constant for the electron impact
dissociation of TEOS and did not specify which of the possible decomposition
mechanisms listed in Table 5.1 this was attributed to. The lower activation energy
associated with the first mechanism (3.6 eV) suggests that would be the preferred
mechanism, although here all reactions are assumed to have an equal probability.
The second reaction for the initiation of TEOS (Eq. (5.2)) requires atomic
oxygen that is formed from the electron impact dissociation of O2 as described in
Eq. (5.3).
O2 + e 2 O + e
(5.3)
The rate constant for this process has been reported by Meeks et al. [11] in a
comprehensive study of the O2/Ar plasma system. The reaction scheme considers
over 35 separate excitation and dissociation processes including electron impact
93
Chapter 5
reactions with Ar and O2, metastable and ion reactions and neutral radical reactions
with O2. A comparison between the rate constant for the electron impact
dissociation of oxygen and TEOS at different electron energy distributions is
plotted in Fig. 5.1.
Morgan [31]: TEOS + e J ions + e

Holtgrave [13]: TEOS + e J ions + e
Morgan [31]: TEOS + e J neutrals + e
*
Meeks [11]: O2 + e J O + O + e
-5
10
Meeks [11]: O2 + e J O + O
-6
10
k (cm /s)
-7
10
-8
10
-9
10
-10
10
-11
10
6
8
Te (eV)
10
12
14
Figure 5.1: Electron impact dissociation rates for oxygen and TEOS as a
function of electron temperature. Rate coefficients for the electron
impact reactions are calculated by multiplying the electron impact cross
sections with the electron density for each electron energy assuming a
Maxwellian electron energy distribution function.
The total rate constant for electron impact ionisation calculated from Morgan
et al. [31] is slightly lower, but agrees well with the total ionisation rate calculated
from the experimental data published by Holtgrave et al. [13]. The datasets suggest
that the CH3, CH2, OC2H5 and C2H5 species are the most dominant by-products of
electron impact dissociation and electron impact ionisation reactions, with a slight
preferential decomposition into ions as opposed to neutral species.
5.3 Experimental
Experiments were performed in a capacitively coupled plasma driven by a 1 kW r.f.
power supply operated at 13.56 MHz (Fig. 5.2). The steel substrate table acts as the
driven electrode and is surrounded by a grounded steel cage to contain the plasma.
No cooling systems were implemented and the substrate typically reached a
temperature of 100C during film growth or particle synthesis experiments that
typically lasted 10 minutes (cf. chapters 4 and 6). Precursors are evaporated and
dosed using a Bronkhorst HiTec CEM system while argon and oxygen flows are
94
Chapter 5
controlled using Bronkhorst HiTec mass flow controllers. All gas lines to the
reactor were heated to 80C to prevent condensation of the TEOS precursor. A
MKS throttle valve located in the reactor exhaust system regulated the reactor
pressure, which is maintained at 1 Torr during all experiments.
Precursor inlet
Plasma glow
0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
FTIR detector 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
FTIR source
S
OE
~
Powered electrode
Grounded cage
Figure 5.2: Reactor configuration indicating in situ FTIR absorption

and optical emission spectroscopy. The substrate table is also the
driven electrode and is surrounded by a grounded cage to confine the
plasma glow.
Tetraethoxysilane [CAS 78-10-4] 98 % purity was supplied by Merck and
used as delivered. Experimental flows and conditions are listed in Table 5.2.
Table 5.2: Experimental condition for particle synthesis of nanocomposite films.
Objective
Plasma
Power [W]
TEOS Flow
[sccm]
O2 Flow
[sccm]
Actinometry
0 - 400
0 90
100
0.12 1
Chemistry
0 400
0 30
0 50
100
Total Flow Pressure

[sccm]
[Torr]
Infrared absorption spectroscopic measurements were done using a Midac

M2500-C Fourier transform infrared spectrometer (FTIR) consisting of two units.
The detector unit housed the Mercury Cadmium Telluride (MCT) detector, while
the source unit housed the Michelson interferometer and glow bar source. The two
instrument units were positioned either side of the reactor and the infrared beam (
95
Chapter 5
= 2 cm) passed over the driven electrode clearing the substrate table by
approximately 3 mm. Spectra were recorded using a spectral resolution of 1 cm-1
and averaging ten consecutive scans to improve the signal-to-noise ratio. Infrared
spectra were de-convoluted using Peak Fit (Systat Software) and heights of
absorbance features were used to monitor variation in species concentration.
Optical emission spectroscopy was done using an Ocean Optics USB-2000
spectrometer with a spectral range of between 280 to 800 nm with a resolution of
approximately 1 nm. Details of the OES measurements and treatments of the data
are presented in Appendix E.
5.4 Mechanisms in the TEOS/oxygen plasma system
FTIR absorption spectroscopy was used to identify the dominant decomposition
products of the TEOS/O2/Ar chemistry. If the oxygen concentration is in excess,
and the chemistry allowed to go to completion, the expected by products would be
SiO2, CO2 and H2O as these are the most thermodynamically stable compounds
formed from the available elements. As the chemistry is kinetically limited, a range
of intermediates in the oxidation process are formed. A typical spectrum recorded
during particle synthesis is shown in Fig. 5.3 and the various molecular vibrations
listed in Table D.1 in Appendix F.
IR absorption (arb. units)
0.06
0.05
CHOOH
C2H4
0.04
0.03
0.02
TEOS
SiOx
CHOOH
C2H4
CO2
TEOS
CH4
0.01
H2O
CH4
CO
H2O
CO2
0.00
500
1000
1500
2000
2500
-1
Wavenumber (cm )
3000
3500
4000
Figure 5.3: FTIR absorption spectrum recorded during SiO2 particle

synthesis using 30 % TEOS, 50 % O2, 30 % Ar at 1 Torr pressure and
300 W discharge.
TEOS decomposes via electron impact dissociation (Eq. (5.4)) and reaction
with atomic oxygen radicals (Eq. (5.5)). Evaluating which of these mechanisms
dominates the TEOS decomposition is complicated by the uncertainty in
96
Chapter 5
determining the concentration of the other reactants, namely the plasma density and
atomic oxygen concentration when TEOS is injected.
k TEOS
e
e + Si(OC 2 H 5 ) 4 SiOCH 2 (OC 2 H 5 ) 3 + CH 3 + e

k TEOS
O
Si(OC2 H 5 ) 4 + O C2 H 4OSi(OC 2 H 5 )3 + OH
(5.4)
(5.5)
Reaction between TEOS and atomic oxygen exhibits a rather weak

temperature dependence and within the temperature range 0 to 200C, k OTEOS varies
between 2.8710-13 and 2.1610-12 cm3/s. The corresponding electron impact
induced decomposition of TEOS is a thousand times higher with an associated rate
constant of 10-8 to 10-9 cm3/s. However, Langmuir probe measurements for pure
argon and argon/oxygen gas mixtures give ne to be approximately 1015 m-3
(assuming Te = 3 eV) whereas actinometry (see Appendix E, chapter 6) give nO to
be approximately 1020 m-3 in an O2/Ar plasma. The overall rate for these two
processes ( ReTEOS , ROTEOS ) is the product of the rate constant and concentration. The
initial TEOS decomposition rate resulting from these two processes is described by
Eq. (5.6), for electron impact dissociation and Eq (5.7) for reaction with atomic
oxygen.
ReTEOS = k eTEOS ne nTEOS
(5.6)
ROTEOS = k OTEOS nO nTEOS
(5.7)
The decomposition rate for electron impact (Eq. (5.6)) and reaction with
atomic oxygen (Eq. (5.7)) under the typical conditions considered here both rates
are estimated to be 102 mol s-1. These calculations suggest that both electron impact
decomposition and reaction with atomic oxygen are equally important in the TEOS
decomposition process. However, in these calculations ne and nO are measured
from pure Ar and Ar/O2 plasmas and the addition of TEOS to the plasma will affect
both ne and nO. However, in the actinometric study, nO is the balance of atomic
oxygen production (via electron impact on O2) and loss. The addition of TEOS to
this system will provide an additional loss mechanism and consequently shift the
production/loss equilibrium towards loss and nO will be lower than that observed in
the absence of TEOS. Based solely on the electron impact cross section data, the
dominant initiation reaction for TEOS is likely to be via electron impact resulting
in the formation of both radical or ion fragments of similar composition. Once
activated, the TEOS fragments will go onto react with O2 (as it is the most
abundant molecule) and other unreacted TEOS molecules. The following reaction
mechanisms and rate constants presented in Table 5.3 are either taken directly or
estimated by comparing work published by Holtgrave et al. [13], Morgan et al.
97
Chapter 5
[31], and Ho et al. [15]. The mechanisms and rate constants for the hydrocarbon
chemistry was taken from a number of literature sources [16-30].
Table 5.3: Reaction of TEOS in a TEOS/O2 PECVD process. (r/i) indicates the
species can be either a radical or ion, x represents the number of ethoxy groups
(0-4) and n the number of electrons for ionizing collisions.
Initiation via electron impact TEOS (k 10-8 10-9 cm3/s)
Si(OC2H5)x + e Si(OC2H5) x-2OCH3(r/i) + (r/i)CH2 + (r/i)OC2H5 + (n)e
Ref.
[13]
Si(OC2H5)x + e Si(OC2H5) x-1OCH2(r/i) + (r/i)CH3 + (n)e
[31]
Si(OC2H5) x + e Si(OC2H5) x-1O(r/i) + (r/i)C2H5 + (n)e
[31]
Initiation via oxygen attack (k 10-12 cm3/s)

Si(OC2H5) x + O Si(OC2H5)x-1OC2H4 + OH
[15]
Si(OC2H5)x + O3 Si(OC2H5) x-1OC2H4 + OH + O2
[15]
Si(OC2H5) x + OH Si(OC2H5) x-1OC2H4+ H2O
[15]
Oxidation reactions (k 510-11 cm3/s)

Si(OC2H5)xOCH2 + O2 Si(OC2H5)xO(r/i) + CHOOH
Est.
Si(OC2H5)xOC2H4 + O Si(OC2H5)xO(r/i) + CH3HCO
[15]
Si(OC2H5)xOC2H4(r/i) + O3 Si(OC2H5)xO(r/i) + CH3HCO + O2
[15]
Si(OC2H5)xOC2H4(r/i) + OH Si(OC2H5)xOH + CH3HCO
[15]
Proton transfer (k 10-11 cm3/s)

Si(OC2H5)xO(r/i) +Si(OC2H5)x Si(OC2H5)xOH + Si(OC2H5)x
Si(OC2H5)xOCH2(r/i) +Si(OC2H5)x Si(OC2H5)xOCH3 + Si(OC2H5)x
[15]
Est.
Starting with an O2/TEOS ratio of 50/20 the TEOS initiation is predominantly

via electron impact to form a variety of TEOS ion and radical fragments. These
fragments then go on to react with O2 and other oxygen species or again with
unreacted TEOS (proton transfer reactions). Side reactions between
C2H4OSi(OC2H5)3 or OSi(OC2H5)3 radicals with TEOS can also occur to form
dimer and trimer species. These species are of similar chemical and structural
composition to TEOS and are therefore likely to participate in the same electron
impact and atomic oxygen reactions as TEOS. If the clusters become large enough
they can pick up a net negative charge in the electronegative plasma and become
trapped (see chapter 4). This mechanism is believed to be responsible for the
formation of the particles observed in the plasma glow. TEOS can also react with
Si(OC2H5)OH to form a dimer, however, the associated rate constant and lack of
ethanol observed in the absorption FTIR suggests that this route does not
significantly contribute to the production of clusters. TEOS oxidation reactions
with O, O2 and OH produce CHOOH and CH3HCO (acetaldehyde). CH3HCO can
98
Chapter 5
also go on to react further with O to form CHOOH and CH3 as depicted in Eq.
(5.8).
CH3HCO + OH CH3 + CHOOH
(5.8)
Evidence for CH3HCO in the infrared is difficult to identify due to extensive

overlap with peaks from CHOOH (around 1600 - 1900 cm-1), with CH4 and TEOS
(2850 to 3150 cm-1) and with Si-O bulk and surface vibrations (1000 - 1300 cm-1).
CH3HCO is identified by a strong CO stretch at 1743 cm-1 and the absence of this
peak is believed to be due to the high reactivity of this compound with atomic
oxygen (at 300 K k 8.510-12 cm3/s) and therefore short lifetime under deposition
conditions.
The CH3 radical is also a product of the electron impact dissociation of TEOS
and other TEOS fragments. It is central to the subsequent hydrocarbon chemistry
along with CH2 formed from electron impact on TEOS. CH2 can go onto react with
H to form CH. Unsaturated hydrocarbons such as C2H4 and C2H2 are also byproducts of the TEOS chemistry, despite the relatively high oxygen concentration
(50 vol%). The formation of C2H2 is a direct product of the CH2 and C2H3 radicals,
as depicted by the mechanism presented in Fig. 5.4.
CH3 from CH3CHO + OH
k = 1.610-11
CH3 + OH CH2(s) + H2O

k = 2.410-11
CH2 formation
CH3 + OH CH2 + H2O

k = 6.610-11
CH2 + CH2 C2H2 + H + H

k=
1.610-10
CH2 + H CH + H2
k = 5.010-10
CH2(s) + H CH + H2
CH formation
k = 1.710-10
C2H3 + O C2H2 + CO
k = 1.510-10
C2H3 + H C2H2 + H2
k = 5.010-11
CH + CH3 C2H3 + H
k = 3.310-11
C2H3 + OH C2H2 + H2O

k = 8.310-11
C2H3 + CH C2H2 + CH2

k = 3.310-11
C2H3 + CH3 C2H2 + CH4
Figure 5.4: Mechanism for formation of acetylene (rate coefficients are in cm3/s).
99
Chapter 5

Mechanisms for the formation of compounds observed in the infrared have been
presented. These mechanisms can also be corroborated with trends in the
concentration of by-products as a function of the process parameters.
Figure 5.5 shows the concentrations of several decomposition products as a
function of the TEOS inlet concentration. At low TEOS concentrations, TEOS
fragments formed from initial electron impact processes and reactions with atomic
oxygen produced from electron impact dissociation of O2 initiate decomposition of
the TEOS. When O2 is in excess, oxygen is available to react with the
hydrocarbons formed from the electron impact reactions on TEOS to form
secondary by-products such as CO2 and H2O. As the TEOS concentration
increases, more oxygen is consumed with TEOS initiation and less is available to
form secondary reaction by-products. Consequently the CO2 concentration
decreases as the TEOS concentration increases. Formic acid, however, is formed
early on during the TEOS decomposition cycle (from reactions with primary TEOS
fragments and CH3CHO), along with CH3. As more TEOS decomposes, the
CHOOH concentration increases along with products from the hydrocarbon
chemistry resulting from the additional CH3 production, namely CH4 and C2H2.
IR peak height (arb. units)
0.035
CHOOH
CH4
0.030
C2H2
0.025
CO2
0.020
0.015
0.010
0.005
0.000
5
10
15
20
25
30
TEOS concentration (%)
Figure 5.5: Infrared peak height for CHOOH (1105 cm-1), CH4 (3017
cm-1), C2H2 (730 cm-1) and CO2 (668 cm-1) as function of TEOS
concentration. The other plasma conditions are discharge power = 300
W, O2 concentration = 50 %, total pressure = 1 Torr and total flow rate
= 100 sccm.
Similar trends are seen if the TEOS concentration is fixed and the oxygen
varied as shown in Fig. 5.6. When no oxygen is added the chemistry is completely
governed by electron impact dissociation of TEOS to predominantly produce
100
Chapter 5
mostly CH3 that goes on to form CH4 and C2H2 similar to that observed in Ar/CH4
plasmas [32]. As the oxygen concentration increases atomic oxygen starts to
initiate the decomposition of TEOS forming CHOOH and CH3 react with the initial
TEOS fragments. At 30 vol% oxygen, excess oxygen begins to participate with the
hydrocarbon chemistry to form CO2 and the hydrocarbons are consumed. Finally,
the oxygen concentration is high enough that the initial stable products (CHOOH
and CH4) begin to be consumed.
0.030
CHOOH
CH4
0.025
C2H2
CO2
0.020
0.015
0.010
0.005
0.000
0
10
20
30
40
50
60
70
80
Oxygen concentration (%)
cm-1), C2H2 (730 cm-1) and CO2 (668 cm-1) as function of oxygen
concentration. The other plasma conditions are discharge power = 300
W, TEOS concentration = 20 %, total pressure = 1 Torr and total flow
rate = 100 sccm.
The influence of discharge power was also investigated but was not seen to
have a large influence on the TEOS/O2 chemistry. Increasing the discharge power
affects the process in two ways. Firstly, it increases the plasma density, which
drives the electron impact reactions and favours the production of by-products
related to electron impact processes. Secondly, increasing the discharge power
increases the potential across the electrode sheath above the substrate. The effect of
this is to increase the energy picked up by positive ions as they cross the electrode
sheath and collide with the substrate table. The increased ion energy available at
the substrate surface favours heterogeneous film forming processes (or at high
energy, sputtering at the substrate). Figure 5.7 shows the variation in CH4, CO2,
C2H2 and CHOOH concentration as a function of discharge power. CHOOH is
readily polymerised and the decrease in CHOOH with increasing discharge power
is likely to be the result of increased ion energy available at the substrate surface to
101
Chapter 5
drive the polymerisation reactions. CH4, CO2 and C2H2 are less polymerisable and
the slight increase in plasma density does not significantly influence the production
or consumption of these compounds. The weak dependency of these product
concentrations on the discharge power also suggests that their production is limited
by reactions between neutral species and less dependant on the electron impact
processes.
CHOOH
CH4
0.035
C2H2
0.030
CO2
0.025
0.020
0.015
0.010
0.005
100
200
300
400
500
Discharge power (W)
cm-1), C2H2 (730 cm-1) and CO2 (668 cm-1) as function of discharge
power. The other plasma conditions are, O2 concentration = 70 %,
TEOS concentration = 20 %, total pressure = 1 Torr and total flow rate
= 100 sccm.
5.6 Conclusions
In the TEOS/O2 plasma process the two dominant initiation steps for TEOS
decomposition are electron impact dissociation and reaction with atomic oxygen.
Calculated rate constants for electron impact with O2 and TEOS suggest that the
process is initiated by electron impact of TEOS to form Si(OC2H5)2OCH3,
Si(OC2H5)3OCH2, Si(OC2H5)3O ions and neutral species. These fragments then go
on to react with O, O2, O3 or OH to form CHOOH and CH3CHO. TEOS
decomposition can also be initiated by attack of oxygen radicals and the relative
magnitude of these two processes depends mainly on the relative concentrations of
TEOS and oxygen, but also on the plasma conditions (ne and Te). Despite the
apparent complex chemistry the presence of C2H2, H2O and CHOOH appear to be
the most kinetically favourable products found by following reactions with the
highest rate constants presented in the literature.
The electron impact data on TEOS suggests an almost 1:1 mixture of ions and
neutrals are formed in the initiation processes. However, a majority of the TEOS
and hydrocarbon chemistry reviewed here deals with reactions between neutral
102
Chapter 5
species. The ability to account for many of the species observed in the infrared, by
reviewing data for primarily neutral reaction suggests that the ions undergo similar
reactions to the radicals reactions reviewed.
Acknowledgement
I would like to thank Pauline Ho and Ellen Meeks for providing rate data for the
neutral chemistry and Lowell Morgan for providing electron impact cross section
data for TEOS. Special thanks goes to Ton van Mol for his assistance in finding
and help in evaluating the data.
References
[1] Alcott G. R., Linden J. L., van de Sanden M.C.M., Mat. Res. Symp. Proc., 726, 297302, (2002).
[2] Buchta C., Wagner H. G., Wittchow W., Zeitschrift fur Physikalische Chemie, 190,
167-181, (1995).
[3] Janca J., Talsky A., Zvonicek V., Plasma Chem. Plasma Proc., 16, 187-194, (1996).
[4] Fracassi F., dAgostino R., Favia P., J. Electrochem. Soc., 139, 2636-2643, (1992)
[5] Raupp G. B., Cale T.S., Peter H., Hey W., J. Vac. Sci. Technol. B., 10, 37-45, (1992).
[6] Dobkin D.M., Mokhtari S., Schmidt M., Pant A., Robinson L., Sherman A., J.
Electrochem. Soc., 142, 2332-2340, (1995).
[7] Kim M. T., Thin Solid Films, 360, 60-68, (2000).
[8] Stout P. J., Kushner M. J., J. Vac. Sci. Technol. A., 11, 2562-2571, (1993).
[9] Allendorf M. D., Melius C. F., J. Phys. Chem., 96, 428-437, (1992).
[10] Allendorf M. D., Melius C. F., Ho P., Zachariah M.R., J. Phys. Chem., 99, 1528515293, (1995).
[11] Meeks E., Larson R. S., Ho P., Apblett C., Han S. M., Edelberg E., Aydil E. S., J.
Vac. Sci. Technol. A, 16, 544-563, (1998).
[12] Ho P., Melius C. F., J. Phys. Chem., 99, 2166-2176, (1995).
[13] Holtgrave J., Riehl K., Abner D., Haaland D., Chem. Phys. Lett., 215, 548-553,
(1993).
[14] Coltrin M. E., Ho P., Moffat H. K., Buss R. J., Thin Solid Films, 365, 251-263,
(2000).
[15] Chemical Reactions in TEOS/Ozone Chemical Vapour Deposition, Ho P., Sandia
National Laboratories Report, SAND2000-0217, (2000).
[16] Wilson C., Balint-Kurti G.C., J. Phys. Chem. A, 102, 1624-1631, (1998).
[17] Bauerle S., Klatt M., Wagner H.Gg., Ber. Bunsenges Phys. Chem., 99, 870-879,
(1995).
[18] Baulch D.L., Cobos C.J., Cox R.A., Esser C., Frank P., Just Th., Kerr J.A., Pilling
M.J., Troe J., Walker R.W., Warnatz J., J. Phys. Chem. Ref. Data, 21, 411-429,
(1992).
[19] Devriendt K., Van Poppel M., Boullart W., Peeters J., J. Phys. Chem, 99, 1695316959, (1995).
[20] Tsang W., Hampson R.F., J. Phys. Chem. Ref. Data, 15, 1087, (1986).
[21] Mebel A.m., Diau E.W.G., Lin M.C., Morokuma K., J. Am. Chem. Soc., 118, 97599771, (1996).
[22] Fahr A., Laufer A.H., Tardy D.C., J. Phys. Chem A, 103, 8433-8439, (1999).
103
Chapter 5
[23] Atkinson R., Baulch D.L., Cox R.A., Hampson R.F. Jr., Kerr J.A., Rossi M.J., Troe
J., J. Phys. Chem. Ref. Data, 26, 521-1011, (1997).
[24] Marinov N.M., Int. J. Chem. Kinetics, 31, 183-220, (1999).
[25] Prada L., Miller J.A., Combust. Sci. Technol., 132, 225-250, (1998).
[26] Opansky B.J., Seakins P.W., Pedersen J.O.P., Leone S.R., J. Phys. Chem., 97, 85838589, (1993).
[27] Pedersen J.O.P., Opansky B.J., Leone S.R., J. Phys. Chem., 97, 6822-6829, (1993).
[28] Knyazev V.D., Slagle I.R., J. Phys. Chem., 1995, 2247-2249, (1995).
[29] Tsang W., J. Phys. Chem. Ref. Data, 17, 887, (1988).
[30] Tsang W., J. Phys. Chem. Ref. Data, 20, 221-273, (1991).
[31] Morgan W.L., Winstead C., McKoy V., J. Appl. Phys., 92, 1663-1667, (2002).
[32] Masi M., Cavallotti C., Carra S., Chem. Eng. Sci., 53, 3875-3886, (1998).
104
Appendix D
Table of infrared absorption frequencies
Appendix D: Table of infrared absorption

frequencies
Table D.1: Products of the TEOS/O2/Ar plasma chemistry.
Molecule
Vibration
Frequency
H2O
CHOOH
CO2
CO
C2H4
C2H2
CH4
SiO2
TEOS
bending
C-O stretch
bending
stretching
bending
bending
bending
Bulk
Si-O sym str.
CH2 rocking
C-O sym str.
CH3 scissor
CH2 wagging
C-H sym str.
1400-1800 cm-1
1105 cm-1
668 cm-1
2025-2265 cm-1
949 cm-1
730 cm-1
1305 cm-1
1000-1250 cm-1
792 cm-1
800 cm-1
1087 cm-1
1170 cm-1
1394 cm-1
2937 cm-1 (CH3)
Vibration
Frequency
stretching
C=O stretch
stretching
3500-4000 cm-1
1776 cm-1
2340 cm-1
stretching
Surface
Si-O asym str.
CH3 rocking
C-O asym str.
CH2 twisting
C-H asym str.
C-H asym str.
3017 cm-1
1100-1250 cm-1
1087 cm-1
964 cm-1
1116 cm-1
1302 cm-1
2895 cm-1 (CH2)
2980 cm-1 (CH3)
105
106
Chapter 6
Deposition of nanocomposite layers for ultra low dielectric applications
Chapter 6: Deposition of nanocomposite

layers for ultra low dielectric
applications
Abstract
The introduction of controlled porosity has been shown to dramatically
reduce the dielectric constant of a material and porous thin films are
currently of great interest in the semiconductor industry. In this chapter
results from chapters 4 and 5 are used to design a new reactor
configuration for the controlled synthesis and incorporation of
nanoparticles into silicone polymer layers. Structure and composition of
the nanocomposite layer produced are characterised using ESEM and
Infrared spectroscopy. Thermal stability and electrical properties are
determined to evaluate the thin films suitability in low-k dielectric
applications. Dielectric constants as low as 1.82 0.02 were achieved at
1 MHz.
6.1 Introduction
As the dimensions of integrated circuit devices scale to smaller feature sizes, the
resistance-capacitance (RC) delay of the metal interconnect will increasingly limit
the performance of high speed logic chips [1, 2]. The integration of ultra low
dielectric materials* (k < 2.5) can reduce this problem and there are a number of
candidates with k in the range of 2-3 [3].
The dielectric constant or relative permittivity of a material is dependant on
the material structure, polarization and polarizability [4]. There are several
microscopic mechanisms of polarization in a dielectric material. Electronic dipole
moments are induced by external electric fields or can be permanent due to the
presence of polar bonding between the elements. Induced dipoles result from the
electron cloud surrounding the atoms being pulled out of place (electronic
polarization), or by a distortion of the material structure (distortion polarization) by
the external electric field. A permanent dipole moment can also exist if the
elements of different electronegativity are bonded within the material and this is
known as orientation polarization. In a d.c. electric field all of these phenomena
contribute to the dielectric constant. When the applied field is alternating the
frequency of the signal comes into play. Electronic polarization follows the electric
field almost instantaneously as only the displacement of electrons is involved. The
distortion polarization involves the motion of atoms and consequently cannot
*
Throughout this thesis the symbol k has been chosen to represent the dielectric constant r,
in line with nomenclature used in the semiconductor industry.
107
Chapter 6
Dielectric constant
respond so rapidly. Both induced polarizations are subject to counter-active

restoring forces, which give rise to a resonant frequency. In contrast, orientation
polarization requires the motion of complete molecules and there is no
counteractive restoring force and hence no characteristic resonance. The orientation
polarization is, however, opposed by thermal disorder.
At low frequencies all three components contribute to the real part of the
dielectric constant. The maximum frequency for orientation polarization is
approximately 109 Hz. Above this frequency only the induced polarization affects
the dielectric constant (distortion and electronic polarization). The threshold for
distortion polarization is approximately 1013 Hz, beyond which only electronic
polarization is defining the dielectric constant. The resonant frequency of the
electronic polarization is typically beyond the frequency of visible light at
approximately 1015 Hz. The overall affect of this is that the dielectric constant
decreases as the frequency increases. The various contributions to the dielectric
constant as a function of frequency are represented schematically in Fig. 6.1.
Orientation
polarization
Distortion
polarization
Electronic
polarization
109
1013 1015
Frequency (Hz)
Figure 6.1: Dielectric constant as function of frequency.

The dielectric constant is therefore high if the molecules are polar and easily
polarizable. Due to the frequency dependence, it has become standard to measure
and report the dielectric constant in an electric field of 1 MHz. Although most
semiconductor devices requiring ultra low-k materials run at much higher
frequency (i.e. the Intel Pentium IV processor running at > 1 GHz), the negative
dependence of k with respect to frequency ensures RC delays will not occur as the
frequency is ramped up.
There are several ways in which the dielectric constant of a material can be
modified to suit a certain application. All aspects of the polarizability can be
affected by changing the elemental and structural composition of the film. A
majority of low-k materials are based on SiO2 or a-C:H [5, 6, 7] or hybrid layer
somewhere between the two. Carbon is less electronegative than oxygen, and hence
the introduction of hydrocarbon moieties into SiO2 reduces the dielectric constant.
108
Chapter 6
In a similar way, Si-OH groups create a large dipole, and although they indicate
porosity, they increase orientational polarization and hence the dielectric constant
increases. The introduction of voids into the material effectively reduces the
material density and has a much stronger affect on the dielectric constant than
trying to alter the polarizability. Porous SiO2 layers synthesized from various POSS
precursors [8, 9, 10] and prepared using sol-gel deposition techniques have
produced thin films with dielectric constants typically between 2-3. Sol-gel
processes with siloxane precursor have also been extensively studied [11, 12, 13,
14], and reports of dielectric constants as low as 1.7 [15] have been reported using
TEOS as a precursor. The ultra low-k properties of these materials are achieved by
a combination of low film density and a lowering of the polarizability by
incorporation of CH moieties. Post deposition curing of these films can reduce the
dielectric [16] constant by driving polar OH groups out of the material, or increase
it as film densification reduces porosity.
Gas phase technologies are an attractive alternative to wet chemical processes
as they utilize existing vacuum reactors and toolsets in the semiconductor
industries. Synthesis of porous SiO2 layers from siloxane precursors using
expanding thermal plasma sources [17], PECVD [18, 19] and reactive evaporation
of SiO [20] have all been investigated. More recently Grill et al. [21, 22, 23, 24]
attempted to synthesize porous low-k films by incorporating volatile organic
hydrocarbon fragments into siloxane layers. By removing the volatile hydrocarbon
fragments in a 400C post deposition annealing procedure, Grill et al. were able to
produce porous layers exhibiting dielectric constants as low as 2.1. Unfortunately,
gas phase technologies offer less control over the deposition chemistry and
therefore less control over film structure and consequently polarizability.
While there are a number of potential ultra low-k dielectrics candidates [25],
they all share some common properties. Firstly, all are porous; the dielectric
constant of silica ksilica = 4, while air has kair 1 and there are currently no dense
low-k dielectric material for semiconductor applications available with a dielectric
constant below 2.5 [3]. The porosity of these materials already introduces technical
challenges when implementing into copper based interconnect ICs. Porous ultra
low-k materials are also mechanically weak compared to SiO2 and are susceptible
to damage during subsequent processing. In addition to this, porous materials are
sensitive to wet and dry cleaning chemicals and hold volatile compounds which
become free to contaminate subsequent processing steps [26]. A number of author
have investigated the mechanical properties and other issues concerning the
implementation of these porous materials in semiconductor processes [27, 28, 29,
30]. An overview of the current state of the art for low-k and ultra low-k
materials and processes is presented in Table 6.1.
109
Chapter 6
Table 6.1: Overview of current state of the art for ultra low-k materials.
Reference
Liu [8]
Baklanov [9]
Xu[10]
Chang [11]
Uchida [12]
Jain [13]
Liu [15]
Lee[30]
Murray [31]
Creatore [17]
Grill [16]
Grill [21-24]
Shaniryan [18]
Si [20]
Frohlich [32]
Technique
Sol-gel
Sol-gel
Sol-gel
Sol-gel
Sol-gel
Sol-gel
Sol-gel
PECVD
Sol-gel
ETP
PECVD
PECVD
PECVD
PVD
Sol-gel
Precursors
HSQ
HSQ/MSQ
MSQ
Not specified
BTM/BTE
TEOS/Glymol
TEOS
BTMSM
Not specified
HMDSO
TMS
TMCTS
TMS
SiO
ORMOCER
Porosity (%)
43-56 %
56 %
30-80 %
62 %
36-55 %
0-29 %
5.5-7.5 %
-
Dielectric
2.7-3.3
2.5 0.1
2.0-2.1
2.65-3.36
1.9-2.7
1.70-3.70
2.1
1.7-2.3
2.9
3.0-4.4
2.05-2.8
1.89-2.11
3.1
Sol-gel processes are a popular choice for fabricating low-k materials because
they inherently produce porous materials and materials with void volume fraction
as high as 95 % have been reported [14]. In addition to this, wet chemical processes
offer greater control over the process chemistry and polarization effects can be
minimised in the final film. Although highly porous materials have dielectric
constants around 1, their susceptibility to copper diffusion makes them unsuitable
for implementation into semiconductor processing. On the other hand, gas phase
deposition techniques inherently deposit dense materials and can easily be
implemented into existing semiconductor processes. The complimentary nature of
these two techniques has been recognised by a number of researchers who have
used PECVD [33] or PVD [34] technology to encapsulate wet chemically
synthesized xerogels.
The necessity for separate reaction conditions for the synthesis and processing
of dual phase hybrid materials has been the central theme to this work. Chapter 3
investigated the feasibility of depositing a dual phase polymer/SiO2 layer using an
ICP/CCP plasma reactor. The failure to synthesize nanometre sized particles in
chapter 3 lead to study of particle synthesis and trapping phenomena presented in
chapter 4. This lead to the discovery that particles formed in low pressure
capacitively coupled discharges are porous, similar to thin films deposited under
such conditions. Consequently incorporating these particles into silicone layers to
improve scratch resistance, as discussed in chapter 1, is somewhat pointless as the
particles themselves are mechanically weak. TGA results however, showed these
particles to be thermally stable (cf. Fig. 4.16) and thermally stable, nano-porous
particles are interesting for low-k applications.
110
Chapter 6
Building on the finding presented in chapters 4 and 5, a new dual capacitively

coupled plasma reactor is designed and used to synthesize nano-porous particles
and incorporate them into a simultaneously synthesized silicone layer. The nanoporosity of these particles is used to introduce controlled porosity into the silicone
layers with the aim of producing ultra low-k layers.
6.2 Reactor design
As in chapter 3, the concept of a dual plasma system for the deposition of a mixed
phase material is deemed necessary, as the two material phases require separate
process conditions. Here, oxygen is required for the synthesis of SiO2 particles, but
would be detrimental to the polymer chemistry causing oxidation of the
hydrocarbon content of the polymer. The first dual plasma concept presented in
chapter 3 relied upon an ICP for particle synthesis. However, plasma stability
issues and the absence of any confining electric fields needed to trap negative ions
and nucleate particle growth, resulted in this concept being abandoned.
Subsequently a modified dual plasma system has been conceived and built to
overcome the limitations of the first.
The new reactor concept consists of two capacitively coupled plasma systems
connected to two 13.56 MHz 1 kW power supplied through matching networks
(Fig. 6.2). Both plasmas can be pulsed. TEOS and oxygen, for particle synthesis,
are fed through the top inlet into Zone I where they are activated by the plasma
formed between the r.f. electrode mesh and the top and bottom plates of the
chamber. The self biasing of the r.f. electrode mesh acts to trap negative ions so
that they may nucleate particle growth in this region. Both plasmas can be
modulated so that particles trapped in Zone I can traverse the r.f. electrode mesh
and flow into Zone II where they are combined with the polymer forming
chemistry. A mesh with a hole diameter of 1 mm was chosen for the top electrode
in order to insure the sheath thickness was larger than the hole diameter and that no
hollow cathode discharges would ignite in the electrode and disrupt the trapping
electric field. The precursors were evaporated and dosed using the gas delivery
system described in chapter 3.
A computation fluid model (CFD) of the particle forming region was made to
identify recirculation or stagnation regions caused by the geometry (Fig. 6.3). The
mesh electrode and grounded plate separating the two plasma regions have a open
area of 38 %. The effect of this is to produce a more laminar flow at the substrate
(Electrode II) and consequently a more homogenous film deposition profile.
111
Chapter 6
TEOS/O2
R.F.
Separating
mesh
Zone I
Electrode I
(wire mesh)
Zone II
OES
Adjustable FTIR
window
Electrode II
TMSE
R.F.
Figure 6.2: Dual CCP deposition system.
3.010-4
2.410-4
1.810-4
Electrode I
1.210-4
6.010-5
1.510-5
0.0
Electrode II
Figure 6.3: CFD model of reactor flows in m/s.
112
Chapter 6
6.2.1 Experiments
Low-k films were synthesized using both the upstream (Zone I) and downstream
(Zone II) plasmas. TEOS was dosed using a Bronkhorst HiTec CEM system and
combined with O2 and Ar flows in the heated gas lines (80C) and fed into the
upstream plasma. TMSE was dosed using a bubbler system (cf. section 3.2) using
Ar as a carrier gas and fed via heated gas lines (80C) in the to downstream plasma.
The injection of oxygen during the synthesis of the plasma polymer layer causes
the layer to become more glass like and consequently the adhesion to decrease (see
chapter 3). It is therefore necessary to ensure that all of the oxygen is consumed in
the particle forming plasma where TEOS and oxygen are injected. Optical emission
spectroscopy was done using an Ocean Optics USB-2000 spectrometer with a
spectral range of between 280 to 800 nm and resolution of approximately 1 nm.
Light from the reactor was fed into the OES spectrometer via an optical fibre. The
concentration of O atoms in an Ar/O2 plasma was monitored under various
conditions using optical emission spectroscopy (OES) using an approach reported
by Han et al. [35] and Lieberman [36]. Details of this study can be found in
Appendix E. This study indicated that 0.05 TEOS mole fraction injected into the
0.5 O2 mole fraction discharge (cf. Fig E.3 Appendix E).
The flow and conditions for the experiments presented in this chapter are
listed in Table 6.2.
Table 6.2: Experimental conditions for particle synthesis of nanocomposite films.
Objective
Oxygen
depletion
study
Low-k
materials
Top
Plasma
[W]
Bottom
Plasma
[W]
TEOS O2
[sccm] [sccm]
TMSE Flow
Pressure
[sccm] [sccm] [Torr]
0-200
0-400
0-30
0-50
100
50
10-20
200
200-400 100-400 10-35
The plasmas are modulated with on and off times of 475 ms and 25 ms
respectively to prevent particle trapping in the upstream plasma and above the
substrate. When operating at 200 W, a self-bias potential of 150 V is observed to
develop on the substrate table. Depositions are performed using a deposition time
of 5 minutes resulting in film thicknesses ranging from 0.88 to 2.04 m depending
on the precursor flows and discharge power. Film thicknesses are determined using
a Tencor P-10 surface profiler with an estimated accuracy of approximately 10
nm.
Boron doped silicon wafers with a resistivity of between 7 and 21 ohm cm
were placed in the centre of the r.f. biased substrate table and used as substrates for
113
Chapter 6
determining the dielectric constants. All substrates were used as delivered and no
attempt was made to clean or remove the native oxide.
The dielectric constants are determined from measurements of the materials
capacitance using two independent methods described in Appendix F.
A variety of deposition conditions were used to deposit the low-k layers with
deposition rates varying from 3.0 to 6.8 nm/s. Increasing the TEOS and oxygen
flows to the top plasma resulted in hazy coatings exhibiting high surface roughness
as shown in the ESEM image in Fig. 6.4. The haze and high surface roughness
observed was attributed to the increased synthesis and incorporation of particles
and particle clusters into the growing TMSE layers.
Figure 6.4: ESEM images of nanocomposite layer.

Some of the samples were cleaved and cross sectional ESEM images taken in
an attempt to identify the individual particles inside the TMSE layer however, no
particles could be seen. The absence of particles in the cross sectional ESEM
images suggests that the particle are mechanically weak and the structure is not that
of dense silicate. If the particles are dense and the intermolecular Si-O bonds within
the particle are strong, then the particles should not break and would therefore be
visible in the cross sectional SEM image. However, the absence of particles in the
cross sectional ESEM image suggests the particles break apart internally instead of
separating from the coating. This process is depicted schematically in Fig. 6.5.
114
Chapter 6
a.
00000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
b.
SiOxCyHz Particles
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000000000000
000000000000000
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 TMSE
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Polymer
00 00 00 00 000 000 000 000 000 000 000 000 000 000 000 000 000 000 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000000000000000000000000000000000000000000000000000000000000000000000000000
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000
Silion substrate
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
000000000000000
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 000 000 000 000 000 000 000 000 000 000 000 000 000 000
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 00 00 00000000000000000000000000000000000000000000 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000000000000000000000000
Figure 6.5: Schematic of particles encapsulated in ppTMSE layer, a. strong particles bonding to pp-TMSE
severed curing cleavage, b. weak particles.
This is consistent with the particle structure discussed in chapter 4 where,
similar to SiO2 films deposited in r.f. discharges, the low electric field surrounding
particles in a plasma results in high porosity.
6.3.1 Dielectric constant
The dielectric constant of a material is affected by the material composition and
porosity as discussed in the introduction to this chapter. The influence of both these
phenomena on the dielectric constant can be seen in Fig. 6.6.
Nanocomposite: Method A
Nanocomposite: Method B
Pure TMSE layer: Method A
4.5
4.5
4.0
4.0
3.5
3.0
2.5
2.0
1.5
100 150 200 250 300 350 400

Bias power (W)
Nanocomposite: Method A
Nanocomposite: Method B
b.
Dielectric constant
Dielectric constant
a.
3.5
3.0
2.5
2.0
1.5
5 10 15 20 25 30 35 40
TEOS concentration (% total flow)
Figure 6.6: Dielectric constant as a function of a. substrate bias and b.

oxygen concentration added to particle synthesis plasma.
115
Chapter 6
Increasing the r.f. bias potential (Fig. 6.6a) is known to enhance film density
by increasing the energy of ions bombarding the film surface. The subsequent drop
in porosity caused the dielectric constant to increase. Films deposited under
identical conditions but without the TEOS/O2 particle plasma (Fig 6.6a) exhibit
much higher k values indicating that the lower k values observed for the
nanocomposite films is indeed due to the incorporation of porous particles.
The effect of film composition on the dielectric constant is illustrated in Fig.
6.6b. At low oxygen concentrations silicone-like particles are produced retaining
some of the organic functionality of the original TEOS precursor. As the oxygen
concentration in the particle forming plasma increases, so does the inorganic nature
of the particles. Inorganic materials are characterised by polar bonding, this
increases the concentration of permanent dipoles present in the film.
Identical trends in k as a function of bias power and TEOS concentration are
observed independent of the technique used to measure k. A critical assessment of
the two techniques used showed Method A to yield the most reliable results (cf.
Appendix F).
6.4 Conclusions
The successful synthesis of nanometre sized silica/silicone particles from a
TEOS/O2 plasma, and subsequent incorporation into plasma polymerised TMSE
layers was demonstrated. Thin films exhibiting ultra low dielectric constants
(values as low as 1.82 0.02 are obtained as measured using Method A) were
deposited. The dielectric constant increased with increasing oxygen added to the
particle plasma, and increasing r.f. bias applied to the substrate table. This is
believed to be due to an increase in the SiOH content of the particles, and the
subsequent increase in the dielectric constant is therefore attributed to increased
polarity of the layer. Increasing the power applied to the substrate table increases
the bias potential above the substrate and the resulting increase in ion energy causes
film densification. The consequent reduction in film porosity leads to an increase in
the dielectric constant at high bias powers.
Acknowledgement
I would like to thank Adriana Creatore, Herman de Jong and Bertus Husken for
their help in the dielectric constant measurement.
References
[1] Jeng S.P., Havemann R.H., Chang M.C., Mater. Res. Soc. Symp. Proc., 35, 337, (1994).
[2] Ryan J.G., Geffken R.M., Poulin N.R., Paraszczak J.R., IBM J. Res., Dev., 39, 371,
(1995).
[3] http://public.itrs.net
[4] Maex K., Baklanov M.R., Shamiryan D., Iacopi F., Brongersma S.H., Yanovitskaya Z.S.
J. Appl. Phys., 93, 8793-8841, (2003).
[5] Grill A., Meyerson B.S., Patel V.V., IBM J. Res. Develp., 34, 849-857, (1990).
[6] Grill A., Thin Solid Films., 398-339, 527-532, (2001).
[7] Grill A., Diamond and Relat. Mater., 10, 234-239, (2001).
116
Chapter 6
[9] Baklanov M.R., Mogilnikov K.P., Microelectron. Eng., 64, 335-349, (2002).
[10] Xu J., Moxom J., Yang S., Suzuki R., Ohdaira T., Appl. Surf. Sci., 194, 189-194, (2002).
[11] Chang T.C., Tsai T.M., Liu P.T., Mor Y.S., Chen C.W., Mei Y.J., Sheu J.T., Tseng T.Y.,
Thin Solid Films, 420-421, 403-407, (2002).
[12] Uchida Y., Katoh T., Oikawa M., Mater. Sci. Semicond. Process., 5, 259-264, (2003).
[13] Jain A., Rogojevic S., Ponoth S., Agarwal N., Matthew I., Gill W.N., Persans P.,
Tomozawa M., Plawsky J.L., Simonyi E., Thin Solid Films, 398-399, 513-522, (2001).
[14] Kim R.-H., Park H.-H., Hyun S.-H., Yeom G.-Y., Thin Solid Films, 332, 444-448,
(1998).
[15] Liu P.-T., Chang T.C., Hsu K.C., Tseng T.Y., Chen L.M., Wang C.J., Sze S.M., Thin
Solid Films, 414, 1-6, (2002).
[16] Grill A., Patel V., J. Appl. Phys., 85, 3314-3318, (1999).
[17] Creatore M., Barrell Y., Kessels W.M.M., van de Sanden M.C.M., Mater. Res. Soc.
Symp. Proc., 766, E6.9.1, (2003).
Process., 5, 279-284, (2003).
[20] Si J.-J., Show Y., Banerjee S., Ono H., Uchida K., Nozaki S., Morisaki H.,
[21] Grill A., Patel V., J. Appl. Phys. 79, 803-805, (2001).
[22] Grill A., J. Appl. Phys., 93, 1785-1790, (2003)
[23] Grill A., Patel V., Rodbell K.P., Huang E., Baklanov M.R., Mogilnikov K.P., Toney M.,
Kim H.-C., J. Appl. Phys., 94, 3427-3435, (2003).
[24] Grill A., Neumayer D.A., J. Appl. Phys., 94, 6697-6707, (2003).
[25] Jin C., Luttmer J.D., Smith D.M., Ramo T.A., MRS Bull., 22, 39, (1997).
[26] Mogsi K., Jacobs T., Brennan K., Rasco M., Wolf J., Augur R., Microelectron. Eng., 64,
11-24, (2002).
[27] Volinsky A.A., Vella J.B., Gerberich W.W., Thin Solid Films, 429, 201-210, (2003).
[28] Travaly Y., Eyckens B., Carbonel L., Rothschild A., Le Q.T., Brongersma S.H., Ciofi I.,
Struyf H., Furukawa Y., Stucchi M., Schaekers M., Bender H., Rosseel E.,
Vanhaelemeersch S., Maex K., Gaillard F., Van Autryve L., Rabinzohn P.,
[29] Lanckmans F., Maex K., Microelectron. Eng., 60, 125-132, (2002).
[30] Lee H.J., Oh K.S., Choi C.K., Surf. Coat. Technol., 171, 296-301, (2003).
[31] Murray C., Flannery C. Streiter I., Schulz S.E., Baklanov M.R., Mogilnikov K.P.,
Himeinschi C., Friedrich M., Zahn D.R.T., Gessner T., Microelectron. Eng., 60, 133-141,
(2002).
[32] Frohlich L., Houbertz R., Jacob S., Popall M., Mueller-Fiedler R., Graf J., Munk M., von
Zychlinski H., Mat. Res. Soc. Symp. Proc., 726, 349-354, (2002).
[33] Schulz S.E., Koerner H., Murray C., Streiter I., Gessner T., Microelectron. Eng., 55, 4552, (2001).
[34] Iacopi F., Tokei Zs., Stucchi M., Brongersma S.H., Vanhaeren D., Maex K.,
Microelectron. Eng. 65, 123-131, (2003).
[35] Han S.M., Aydil E.S., Thin Solid Films, 290-291, 427-434, (1996).
117
Appendix E
Optical emission spectroscopy and actinometry
Appendix E: Optical emission spectroscopy

and actinometry
Optical emission spectroscopy was done using an Ocean Optics USB-2000
spectrometer with a spectral range of between 280 to 800 nm and resolution of
approximately 1 nm. Light from the reactor was fed into the OES spectrometer via
an optical fibre. The concentration of O atoms in an Ar/O2 plasma was monitored
under various conditions using optical emission spectroscopy (OES) using an
approach reported by Han et al. [35] and Lieberman [36]. Argon is added as an
inert tracer gas to the oxygen plasma and the transition = 750.4 nm (2p1 1p0)
was monitored. Changes in the atomic oxygen concentration were followed using
the = 777.4 nm transition arising from the decay, O(3p5P) O(3s5S) +
hv(777.4nm). The O(3p5P) state is populated via both electron impact excitation of O
and as a result of electron impact dissociation of O2,
43
k eO
e + O(2 p P) O(3 p 5P) + e

O
k de
e + O 2 O + O(3 p 5P) + e
(E.1)
(E.2)
O
where k eO and k de
represent the production rate due to electron impact excitation
and electron impact dissociation excitation respectively. The Ar 750.4 nm
transition has a similar excitation cross-section and threshold dependency to
electron temperature as the O emissions at =844.6 nm (3p3P 3s3S transition)
and = 777.4 nm (3p5P 3s5S transition). The relative error associated with these
measurements was estimated to be 55 % and 20 % for the 777.4 nm and 844.6 nm
lines respectively when compared to VUV measurements [1]. Unfortunately due to
the restricted wavelength range of the emission spectrometer (300-800 nm) the =
844.6 nm transition was not available. From work published by Pagnon et al. [1]
O
k eO / k de
is estimated to be approximately 10-3, i.e. the dissociative excitation
process contributes approximately 0.1 % to the total O(3p5P) population and is
therefore ignored for simplicity. The atomic oxygen concentration can be related to
the argon concentration by,
IO
k eO n O
=
I Ar k eAr n Ar
(E.3)
where k eO and k eAr are the excitation rate coefficients. The ratio k eO / k eAr 0.1 [1] is
insensitive to changes in plasma operating conditions, due to the similarity in the
respective process dependencies on the electron energy distribution. The O(3p5P)
excited state can also decay non radiatively by quenching collisions with other gas
atoms and reactor walls. These processes have not been considered as their
118
Appendix E
magnitude depends on the reactor system and therefore would require a

complimentary calibration technique such as laser induced fluorescence (LIF) [2].
As the quenching losses have not been considered, the subsequent results are an
underestimate of the O density and should therefore only be interpreted as order of
magnitude estimations.
Results
As might be expected the atomic oxygen concentration in the plasma increases as
the discharge power is increased (Fig. E.1). The close agreement in atomic oxygen
concentrations for measurements made at 1 Torr and 0.5 Torr indicates that there is
no significant change in the production (i.e. no change in Te or ne) or quenching of
the atomic oxygen excited state within this pressure region. The higher scatter in
the data measured at 1 Torr is attributed to hollow cathode formation at higher
pressures (caused by a slight drop in D with increasing p).
25
1.0 Torr
0.5 Torr
-3
nO x10 (m )
20
20
15
10
5
0
100
200
300
400
500
Discharge power (W)
Figure E.1: Variations in atomic oxygen density as a function of the

discharge power, O2 flow = 10 sccm, Ar flow = 90 sccm.
Under certain operating conditions, hollow cathode discharges appeared in the
grounded containment cage surrounding the plasma and were observed to flicker.
In areas where hollow cathode discharges were seen, higher levels of O emissions
were also observed. The increased O emission at hollow cathode sites is most likely
a result of higher electron density in these regions causing an increased
concentration of excited O atoms. The flickering of these hollow cathode
discharges appeared to be most severe at 1 Torr and 0.1 mole fraction O2 as can be
seen in the increased scatter of the data in Fig. E.2. Hollow cathodes appear when
the dimensions of hollow cathode hole are of the order of a few Debye lengths (D)
and electron confinement occurs. For ne = 1015 m3 and Te = 3 eV the Debye length
is approximately 0.4 mm whereas the grid holes are 2 mm in diameter. For high
119
Appendix E
oxygen concentrations the plasma is electronegative, where electron attachment

and negative ion formation cause ne to drop and, in order to maintain the ionization
rate, Te to increase. Decreasing the oxygen concentration causes the plasma to
become more electropositive and ne increases and Te drops. The Debye length is
proportional to ne1/2 and inversely proportional to Te1/2 and thus D also decreases
until a hollow cathode emission occurs.
Hollow cathode region

16
14
10
20
nO x 10 (m )
12
0.12 Torr
0.50 Torr
1.00 Torr
6
4
2
0.0
0.2
0.4
0.6
0.8
1.0
Ar fraction (0-1)
Figure E.2: Oxygen atom density (nO) as function Argon volume fraction
at different total pressures. Scatter in data at 0.9 fraction Ar is due to
hollow cathode discharge flickering.
0.7
0.5
20
-3
nO x10 (m )
0.6
0.4
0.3
0.2
0.1
0.0
0.00
0.05
0.10
0.15
0.20
TEOS Fraction (0-1)
Figure E.3: Atomic oxygen concentration in bottom plasma as a function

of TEOS fraction in top plasma.
120
Appendix E
The dramatic drop in the atomic oxygen concentration as TEOS is injected

into the top plasma is evidence that all of the oxygen is consumed in the TEOS/O2
plasma (Fig. E.3). The consumption of atomic oxygen is already complete at
relatively low discharge powers (Fig. E.4).
400 Watts
2.0
300 Watts
20
-3
n O x10 (m )
1.5
200 Watts
1.0
100 Watts
0.5
300 Watts + TEOS
0.0
0
50
100
150
200
250
300
Top Plasma Power (W)
Figure E.4: Atomic oxygen concentration in bottom plasma as a function

of discharge intensity of top plasma.
References
[1] Pagnon D., Amorim J., Nahorny J., Touzeau M., Vialle M., J. Phys. D: Appl. Phys., 28,
1856-1868, (1995).
[2] George A., Engemann J., Brockhaus A., J. Phys. D: Appl. Phys., 35, 875-881, (2002).
121
Appendix F
Dielectric constant determination
Appendix F: Dielectric constant

determination
Method A: 1 cm2 samples are prepared from the Si substrate and sputtered with a
60 nm gold front contact layer. Silver paint is applied to the backside of the sample
to ensure a good electrical contact to the silicon wafer. The sample is then mounted
between two contacts and connected to the Hewlett Packard 4284A Precision LCR
meter (20 Hz to 1 MHz) (Fig. F.1) and the capacitance measured at 1 kHz and 1
MHz oscillating potentials of 50 and 100 mV.
Au top contact
SiO2 nanocomposite
HP 4284A
LCR meter
Si wafer
Ag paint
Figure F.1: Circuit setup for determining the

dielectric constant using Method A.
Method B: A contact pattern of aluminium is sputtered onto the top of the sample
as shown in Fig. F.2a and F.2b. The sample is then placed into an oscillating circuit
consisting of two comparators and a flip-flop reset switch (Fig. F.2c). The negative
and positive inputs of comparators one and two (Cp1- and Cp2+) are set at 1/3 and
2/3 the line voltage respectively, by R2. The other two comparator inputs (Cp1+ and
Cp2-) are charge via resistors R4 and R5 until the comparator outputs switch and the
flip-flop and transistor grounds the comparator inputs (Cp1+ and Cp2-) and resets
the circuit.
c.
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
000000
00000
000 000 000 000 000 000 0 0 0 0 0 0 0 0 0 0 0 0 000 000 000 000 000
00 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 0
00 000 000 000 000 000 00 00 00 00 000 000 000 000 00 00 00 00 00 000 000 000 000
Al sputtered 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
electrode
b.
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
R1
R4
n
Low-k layer
Cp1
R2
R5
Si Wafer
C
Cp2
Flip-Flop Reset
a.
R3
Figure F.2: Method B for determining the dielectric constants. a. plan view
of sample b. cross sectional sample view showing Al layer, dielectric
material and Si wafer and c. resonance circuit.
122
Appendix F
(f),
For the idea case, the overall capacitance (CT) relates to the circuit frequency
CT =
1.44
(R 4 + R 5 ) f
(F.1)
The sample preparation technique used in Method B actually generates two

series capacitances within the sample, C1 and C2 [F]. The relative contribution to
the overall capacitance (CT) scales with the respective areas A1 and A2 [m2] as
described in Eq. (F.2).
1
1
C T =
+
C1 C 2
1
1
=
+
A1
A2
0 r
0 r
d
d
0 r 1
1
+
=
d A1 A2
(F.2)
where d is the thickness of the dielectric layer [m] with permittivity r, and 0 is the
permittivity of free space [F.m-1].
The oscillating frequency is determined by R4, R5 and C. The presence of
internal and parasitic capacitance are accounted for by calibrating the circuit
resonant frequencies with a series of fixed capacitors (1 % tolerance).
f [kHz]
1000
100
10
Standard fit
Calibration
100
1000
10000
C [F]
Figure F.3: Calibration of circuit used in Method B for

determining the capacitance of nanocomposite samples.
For both measurement techniques, the dielectric constant k is determined from
the capacitance by,
123
Appendix F
k=
Cd
0 A
(F.3)
where C is the sample capacitance in [F].

Errors in the dielectric constant measurements are predominantly caused by
the uncertainty in the film thickness and uniformity and are estimated to be
approximately 2-6 %. Method B is likely to be less reliable than A for several
reasons namely,
1. The circuitry is not mounted in any form of enclosure and hence not
protected from surrounding environmental influences (electromagnetic
interference, moisture and mechanical vibrations).
2. No subsidiary circuitry is included to compensate temperature variations or
fluctuations in the power supply.
3. Each sample is effectively measured at a different voltage frequency as the
circuit resonance depends on the capacitance of the sample.
124
Chapter 7
General Conclusions
Chapter 7: General Conclusions

The primary aim of this work is to develop a process for the synthesis of hybrid
materials using solely gas phase synthetic methods. Such a broad project objective
required putting into context and the initially objectives became to focus on the
mechanical properties of the layers for barrier and scratch resistant applications. A
dual plasma reactor system utilizing both inductive and capacitive plasma
configurations to generate a high plasma density for synthesizing inorganic
particles and low density for polymeric layers was designed. Limitations in the
initial reactor concept prompted investigations into the plasma chemistry and
physics behind dusty plasmas. A deeper understanding of these processes lead to a
change in the focus of the work and a new reactor configuration was designed for
the deposition of low-k layers. The main conclusions resulting from the various
investigation contained in this work are summarized below.
7.1 Scratch resistant layers
PECVD techniques offer several advantages over wet chemical synthetic methods
for producing single phase hybrid films of the general form SiOxCyHz. However,
adhesion to polymeric substrates degrades with the increasing inorganic nature of
the layer and almost complete de-lamination occurs as the material structure
approaches that of stoichiometric thermal silica. Wet chemical studies show that
mechanical properties can be achieved while maintaining good adhesion particles if
inorganic particles are embedded into the soft organic layers. The conclusion at this
stage is therefore to focus on the synthesis of nanocomposite materials starting with
a study of dusty plasmas.
7.2 Particle monitoring
Particle synthesis was attempted in both inductively coupled and capacitively
coupled plasma discharges. The failure to synthesize particles in the inductive
plasma was attributed to the low residence time in the ICP tube caused by the
absence of confining electric fields, and plasma instability attributed to negative
ions. The successful synthesis of particle in the capacitively coupled plasma was
attributed to the confining effects of the electrode sheath. It was found that the huge
potential present in the ion sheath above a driven electrode represents a
theoretically insurmountable barrier for trapped particles. OML theory predicts that
the electrostatic force dominates for small particles, pushing them to the lowest
potential (centre of the plasma discharge). As the particle mass scales with a3, the
gravitational force dominates as the particle radius increases. Despite this, particles
several orders of magnitude smaller than those predicted by OML theory are
capable of breaching the sheath to arrive at the electrode surface. Particles are able
to overcome this electric field when the electron flux to the particle surface is
reduced by shadowing from the other particles and the electrode itself.
125
Chapter 7
General Conclusions
7.3 Plasma chemistry

Despite the apparent complex chemistry the decomposition of TEOS and formation
of the unexpected by-products acetylene, formic acid and water can be explained
using relatively simple reaction mechanisms (typically involving less then 4 steps)
with relatively high rate constants (> 10-16 cm3/s) involving purely neutral species.
Although electron impact dissociation data suggests that the process is initiated by
electron impact on TEOS, reactions between neutral species are sufficient to
elucidate much of the deposition process chemistry.
7.4 Potential application as low dielectric layers
The deposition of ultra low-k layers using the new dual capacitively coupled
plasma reactor was successfully demonstrated. Thin films exhibiting ultra low
dielectric constants (values as low as 1.82 0.02) were deposited. The dielectric
constant was found to be dependent on the oxygen concentration in the particle
synthesis plasma and bias power applied to the substrate table. This is effects were
attributed to an increase in polarity of bonds within the film (caused by the
incorporation of SiOH groups) and film densification due to ion bombardment
respectively.
126

Chemical abbreviations
ORMOCER
organically modified ceramic
PP
plasma polymer
R
arbitrary hydrocarbon side group
Al(iPA)3
aluminium tri isopropoxide
Glymol
(3-glycidoxypropyl) trimethoxysilane
HPC
hydroxypropyl cellulose
MPS
3-methacryloxypropyltrimethoxysilane
MMA
methyl methacrylate
TEOS
tetraethoxysilane
Ti(iPA)4
titanium tri isopropoxide
TMOS
tetramethyloxysilane
TMTSO
1,3,5,7-tetramethylcyclotetrasiloxane
TVTMTSO
2,4,6,8-tetravinyl 2,4,6,8-tetramethylcyclotetrasiloxane
TMMOS
tetramethylmetoxysilane
HMDS
hexamethyldisilazane
TMSE
1,2- bis(trimethyl)siloxyethane
TMDSO
tetramethyldisiloxane
VpMDSO
vinylpentamethyldisiloxane
DVS
1-3 divinyl tetramethyldisiloxane
MTS
methyl trimethoxysilane
DLC
diamond like carbon
DADBS
diacetoxy-di-tert-butoxysilane
DPXC
dichloro-di-paraxylylene
TRIES
triethoxysilane
TMOS
tetramethyloxysilane
OMCTS
octamethyltrisiloxane
HMDSO
hexamethyldisiloxane
HSQ
hydrogen-silsesquioxane
MSQ
methyl-silsesquioxane
POSS
polyhedral oligomeric silsesquioxane
Abbreviations
UV
CVD
PVD
CCP
ICP
ECR
WBL
PC
XPS
ERD
ultraviolet radiation
chemical vapour deposition
physical vapour deposition
capacitively coupled plasma
inductively coupled plasma
electron cyclotron resonance
weak boundary layer
polycarbonate
x-ray photon spectroscopy
elastic recoil detection
127
ATR-IR
EMA
PECVD
FTIR
XPS
ESEM
OML
RC
TGA
a.c.
d.c.
Symbols
Ti
ne
ni
mfp
Ea
a
Vf
mi
me
kB
0
e
Te
ns
M
I
Ii
Poff
Pon
Vf
Vp
Fes
Fg
Fid
Fth
Fnd
128
attenuated total reflection infrared

electron microprobe analysis
plasma enhanced chemical vapour deposition
fourier transform infrared
x-ray photon spectroscopy
environmental scanning electron microscopy
orbital motion limited
resistance-Capacitance
thermal gravimetric analysis
alternating current
direct current
ion temperature
electron density
ion density
mean free path
activation energy
Debye length
particle radius
potential of a body with respect to plasma
ion mass
electron mass
Boltzman constant
permittivity of free space
unsigned charge on an electron (1.602 x10-19 C)
electron temperature in units of volts (V)
ion density at the sheath/plasma interface
ion mass (Kg)
current (A)
ion current (A)
pressure when plasma is switched off (mTorr)
pressure when plasma is switched on (mTorr)
floating potential (V)
plasma potential (V)
Bohm velocity, speed of sound in plasma (m.s-1)
electrostatic force
gravitational force
ion drag force
themophoretic force
neutral drag force
translational component of the thermal conductivity of a gas
Summary
Summary
Recent developments in materials technology, fuelled by the growing hype
surrounding nanotechnology, have given rise to a new breed of materials known as
nanocomposites. Nanocomposite materials (a subgroup of hybrid materials) are
formed from standard polymers impregnated with nanometre sized ceramic or
inorganic particles. Bulk polymeric properties, such as toughness and low weight
are retained, while the incorporation of ceramic/inorganic nano-particles adds
additional functionality such as improved scratch resistance and reduced
permeability and makes hybrid materials interesting for a wide variety of
applications. In this thesis, the possibility to incorporate controlled porosity into
nanocomposite thin films as low-k material for dielectric interlayers in
semiconductor devices is successfully investigated.
Wet chemical process technologies are currently the only methods used to
synthesize nanocomposite materials on an industrial scale. However, high curing
temperatures, multiple process steps and use of environmentally harmful solvents
make wet chemical techniques unattractive for production on the scale necessary to
fulfil industrial requirements. In contrast, gas phase process technologies are
typically one-step solvent free processes and therefore offer an attractive alternative
to wet chemical processes. In addition to these benefits, the extensive use of
vacuum and gas phase processing equipment in the semiconductor processing
industry make gas phase synthesis of nanocomposite films for semiconductor
applications particularly attractive.
The primary aim of this work has been the conception and realisation of a
process for synthesizing hybrid materials using solely gas phase synthetic methods.
The emphasis is on plasma-based synthesis of nanocomposites since for these
processes the substrate temperature can be kept low. The main theme through the
thesis has been the continued design, testing and adaptation of a reactor system
conceived from reviewing literature regarding a range of materials and deposition
processes. The development of the reactor concept progresses with increased
understanding of the plasma physics and chemistry as obtained from applying in
situ plasma diagnostics to monitor the plasma and particle formation, and from
techniques related to film formation. Alongside the reactor development, the
research has also investigated reactive ion etching and adhesion of hybrid coatings,
chemical kinetics and the physics of dusty plasmas.
The thesis begins with an extensive review of relevant literature regarding a
variety of hybrid film deposition techniques and material descriptions. Results of
this review prompted the design of the initial reactor concept consisting of a dual
inductively coupled/capacitively coupled (ICP/CCP) plasma reactor, i.e. the initial
approach chosen was to separate the production of nanoparticles spatially from the
film formation process on the substrate. Due to its ability to form both hard silica
and flexible silicone based materials, silicon based process chemistry was chosen
as the main chemistry with which to deposit the hybrid films. Several silicon
129
Summary
precursors were evaluated for producing silica and silicone materials using the ICP
and CCP plasmas. Preliminary findings showed the precursor combinations
O2/Tetraethoxysilane (TEOS) and 1,2-bis(trimethyl)siloxyethane (TMSE) were
optimal for synthesizing the inorganic and organic film fractions respectively. Both
inorganic and polymeric thin films were successfully synthesized and compared
well to wet chemically synthesized films.
With the early success in synthesizing hybrid layers, the more challenging
task of synthesizing nanocomposite materials became the objective of the research.
Unfortunately all initial attempts at synthesizing nanoparticles with the ICP were
unsuccessful. An investigation into dusty plasma physics and the plasma chemistry
of the O2/TEOS precursor system identified negative ion formation and trapping as
being the key steps in the formation of particles in low pressure discharges. The
low residence time in the ICP tube caused by the absence of confining electric
fields was the primary reason why particles were not formed in this plasma. On the
other hand, in the CCP, due to the presence of confining electric fields, successful
generation of dust particles was demonstrated. Characterisation of these particles,
both in situ using Fourier transform infrared absorption spectroscopy (FTIR) and of
particles collected after the experiments using a range of physical and chemical
characterization techniques, indicated the particles to be porous and silica-like. It
was demonstrated that these particles could be grown, manipulated and kept for a
long time above the electrode of a CCP.
Combining the finding of the dusty plasma and silicon chemistry studies
resulted in the design of a second dual plasma reactor consisting of two capacitive
plasmas. In addition, the nanoparticle production and the film deposition were
temporally separated by the use of pulsed plasma operation. This reactor was then
used in combination with the O2/TEOS and TMSE precursor systems to deposit
nanocomposite films with controlled porosity for low dielectric constant films in
semiconductor devices. This approach proved to be successful and low-k films
exhibiting dielectric constants as low as 1.82 0.02 were demonstrated. On this
finding a patent application has been based.
130
Samenvatting
Samenvatting
Recente ontwikkelingen in de materiaaltechnologie hebben, onder invloed van de
groeiende aandacht voor de nano-technologie, tot een nieuwe type materialen
geleid die, vanwege hun structuur, bekend staan als nanocomposieten.
Nanocomposieten (een subgroep van de hybride materialen) worden gevormd uit
standaard polymeren die gempregneerd zijn met anorganische nanodeeltjes. In het
geval van de nanocomposiet materialen blijven de polymeren eigenschappen zoals
de taaiheid en het lage gewicht behouden, terwijl door het inbouwen van de
anorganische of keramische nanodeeltjes additionele functionaliteit wordt
toegevoegd zodat bijvoorbeeld de krasvastheid wordt verbeterd en de
gasdoorlaatbaarheid wordt verlaagd. De mogelijkheid om unieke eigenschappen te
combineren maakt hybride materialen erg interessant voor een groot aantal
verschillende toepassingen. In dit proefschrift wordt ondermeer succesvol
aangetoond dat films aangebracht kunnen worden met een controleerbare (nano)porositeit, waardoor dit type hybride materiaal geschikt is als een isolerende
tussenlaag met lage dilectrische constante (zgn. low-k materialen) in
halfgeleidertoepassingen.
Natchemische procestechnologien zijn momenteel de enige methoden die
gebruikt worden voor de synthese van nanocomposieten op een industrile schaal.
Door de hoge droogtemperaturen, de vele processtappen en het gebruik van
schadelijke oplosmiddelen is deze produktiemethode echter niet erg aantrekkelijk
voor de hoge eisen voor de produktie op een industriele schaal.Een aantrekkelijk
alternatief voor natchemische productie zou gasfase procestechnologie zijn. Dit is
een enkelstapsproces, waar geen oplosmiddelen aan te pas komen. Bovendien is het
gebruik van vacum- en gasfase depositie-apparatuur standaard in bijvoorbeeld de
halfgeleiderindustrie.
De primaire doelstelling van het promotie-onderzoek betreft het ontwerp en de
realisatie van een proces dat hybride materialen kan synthetiseren op basis van
gasfase processen. De nadruk ligt met name op plasma geassisteerde processen
omdat deze processen het gebruik van een lage substraattemperatuur mogelijk
maakt. Het belangrijkste thema in dit proefschrift is het ontwerpen, testen en
aanpassen van een reactorsysteem, op basis van een diepgaand onderzoek van het
plasma en de deeltjes gevormd in het plasma middels geavanceerde in situ
plasmadiagnostieken. Het ontwerp van het gasfase depositieproces en de reactor
wordt telkens verbeterd ten gevolge van een beter begrip van de plasmafysica en
chemie en de processen die tot een hybride filmvorming leiden. Parallel aan de
reactorontwikkeling, is er onderzoek verricht naar het plasma etsen van hybride
materialen, en de hechting tussen hybride coatings en een polymeren substraat.
Op basis van uitgebreid literatuuronderzoek op het gebied van hybride
materialen is een eerste reactor concept ontworpen bestaande uit zowel een
inductief gekoppeld plasma (ICP) als een capacitief gekoppeld (CCP) plasma. Deze
eerste benadering was gebaseerd op de gedachte de produktie van nanodeeltjes en
131
Samenvatting
het film depositieproces ruimtelijk te scheiden. Er is gekozen voor een op silicium

gebaseerde chemie, omdat dit de mogelijkheid opent zowel hard silica als flexibel
siliconenachtige films te deponeren. Verschillende precursoren zijn gevalueerd
voor de synthese van silica en siliconen met behulp van de ICP/CCP plasmas.
Voorbereidend onderzoek liet zien dat een combinatie van O2/tetraethoxysilaan
(TEOS) en 1,2-bis(trimethyl)siloxyethaan (TMSE) het meest geschikt waren voor
het synthetiseren van respectievelijk de anorganische en de organische fractie.
Zowel de anorganische als de organische films zijn succesvol gesynthetiseerd en
zijn qua eigenschappen vergelijkbaar met lagen gemaakt met behulp van
natchemische technieken.
De volgende stap in het promotie-onderzoek was de synthese van
nanocomposiet materialen. Helaas bleek het niet mogelijk om nanodeeltjes te
maken met behulp van het ICP plasma. Onderzoek naar de fysica en de chemie van
het plasma met TEOS en O2 als precursor systeem, wees uit dat de vorming en het
opsluiten van negatieve ionen in het plasma de belangrijkste stap is bij de vorming
van deeltjes in een lage druk plasma. In een ICP is de verblijftijd van de
precursoren te kort door de afwezigheid van opsluitende elektrische velden
waardoor deeltjesvorming in het plasma nauwelijks plaatsvindt. In tegenstelling tot
het ICP werd in het CCP, ten gevolge van de aanwezigheid van opsluitende
electrische velden, met succes deeltjesvorming aangetoond. De deeltjes konden
over een langere tijd gemanipuleerd en bewaard worden boven de electrode van het
CCP. Zowel in situ als ex situ karakterisatie van de nanodeeltjes gemaakt in het
CCP met behulp van o.a. Fourier Transform Infrarood (FTIR)
absorptiespectroscopie, wees uit dat de deeltjes silica-achtig en poreus zijn.
De resultaten uit het onderzoek naar poeder-plasmas en de siliciumchemie
hebben tot het ontwerp van een tweede plasmareactor geleid bestaande uit twee
capacitief gekoppelde plasmas (CCP). De scheiding van deeltjesvorming en
filmdepositie werd nu verwezenlijkt in de tijd door het gebruik van gepulste
plasmas. In combinatie met de precursorsystemen O2/ TEOS en TMSE zijn in deze
reactor nanocomposiet films gemaakt met een controleerbare porositeit en met een
laagst bereikte dielektrische constante van 1.82 0.02. De potentie van deze films,
als low-k films in de halfgeleiderindustrie is groot en een patent op deze vinding is
ingediend.
132
Acknowledgement
Acknowledgement
It goes without saying that the research presented in this thesis, performed over the
last four years, is the result of teamwork and not only the efforts of a single person.
I wish then to thank all those who have contributed, in whatever way, to the
realization of this manuscript.
I would first like to thank my supervisor, Richard van de Sanden for his
guidance, inexhaustible energy and enthusiasm. Thanks also for sticking up for me.
Many thanks also to the members of the ETP group, Adriana for your valuable
input, not only during the brief period we shared and office, but throughout the
duration of the project. The two students Sascha Kondic and Dimitri Eijkman,
thank you for your valuable contributions.
A majority of the experimental work presented in this thesis was conducted at
the TNO TPD and I would like to express my thanks to all my colleagues who
helped things run smoothly and provided morel support. In particular I would like
to mention Gerwin and Frank for your help around the laboratory and especially
Hans for your support and unique ability to simply know the right people to get
things done! Special thanks must go to Ton for his invaluable support, friendship
and always finding time no matter how trivial the question. Thanks for introducing
me to the Zeelste kermis and making me feel right at home with all the great people
who inhabit that small town. And not least, thanks for waiting, if it were not for you
Id still be standing at Eindhoven station!
For the three months of this project I spent in Australia I would like to thank
all the members of the RSPhysSE group for making me feel welcome. Rod Boswell
for allowing me to attend his insightful plasma course entitled my friend the
electron that has formed the foundation to my understanding of plasma physics.
Rob and Christine for all weekend surfing trip, Christmas dinners and generally
making my time in Australia so unforgettable. Thanks.
This would all have been much harder without my close friends outside of my
work environment who have offered moral support over the years. In particular
Akif and John. The Bommel crowd Kornel, Richard, Vivianne, Huub (on
occasion), Hans, Frank and Danny for putting up with me talking nonsense. And of
course Ursula for putting up with so much more.
Finally I would like to thank my family who have always stood by me. My
parents, George and Neil for helping me move and make my flat a home, Sally for
your feng shui advice and artistic input and Simon for making so many trips over to
visit and keeping me company.
David Sheel and Karel Spee, thanks for believing in me.
133
Acknowledgement
134
Curriculum vitae
Curriculum vitae
4 December 1974
Basingstoke, UK
September 1993 - July 1997

BSc Hons Chemistry, University of Bath, UK
Twelve months traineeship at Pilkington Technology Centre,
Ormskirk, UK
November 1997 November 1999

Marie Curie fellowship, TNO TPD, Eindhoven, NL
November 1999 February 2004

PhD, Department of Applied Physics, Eindhoven University
of Technology
Three months working visit to the group of Prof. R. Boswell at
the Research School of Physical Sciences and Engineering,
Australian National University, Canberra, AU
February 2004 - Present

Research Scientist, TNO TPD, Eindhoven, NL
135
Curriculum vitae
136

Alcott G.R., van Mol A.M.B., Eijkman D., Linden J.L., van de Sanden M.C.M.,
Reaction kinetics and mechanisms during PECVD from a TEOS/O2 gas mixture,
submitted to Plasma Sources Science and Technology (2004).
Alcott G.R., van de Sanden M.C.M., Kondic S., Linden J.L., Vapour pressures of
precursors for the chemical vapour deposition of silicon based films, Advanced
Materials, Chem Vap. Deposition, 10 (1), 20-22, (2004).
Alcott G.R., Linden J.L., van de Sanden M.C.M., Gas phase deposition of hybrid
coatings, Materials Research Society Symposium Proceedings, Vol 726, Q9.9.1Q9.9.6, (2002).
Alcott G.R., Eijkman D.J., Schrauwen C.P.G., Linden J.L., van de Sanden M.C.M.,
10th Neues Dresdner Vakuumtechnisches Kolloquium Proceedings, 55-60, (2002).
Patents
Alcott G.R., van de Sanden M.C.M., Linden J.L., Hamers E.A.G., CVD van
Hybride materialen, Application Nr PCT/NL03/0037, Priority date 18/01/2002,
International filing date 17/1/2003.
Alcott G.R., Creatore M., Linden J.L., van de Sanden M.C.M., Method of
manufacturing ultra low dielectric layers. Application date 13/2/2004.
137

A Plasma

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

A Plasma

Transféré par

Droits d'auteur :

Formats disponibles

Plasma deposition of nanocomposite thin films

Process concept and realisation

Dit proefschrift is goedgekeurd door de promotoren:

CIP-DATA LIBRARY TECHNISCHE UNIVERSITEIT EINDHOVEN

Chapter 2: Anisotropic plasma etching of SiOxCyHz layers

Chapter 3: Synthesis and characterisation of SiOxCyHz layers

Synthesis, monitoring and characterisation of

4.2 Forces on particles

Chapter 5: Reaction kinetics and mechanisms in the plasma

Deposition of nanocomposite layers for ultra

Chapter 7: General Conclusions

Symbols and Abbreviations

Publications related to this work

Chapter 1: General introduction

Organically modified ceramic: Derived from

Nanocomposites: Particles mixed into a

Figure 1.1: Classification of material forms as described in this thesis.

Figure 1.2: Hydrolysis of TEOS, R represents an arbitrary hydrocarbon

1.1.2 Current state of the art & Applications

Figure 1.3: Theoretical structure of organically modified ceramic

Figure 1.4: Structural representation of Polyhedral

After 500 taber cycles

Urreaga [34] TEOS/MPS/MMA

Long curing time and high

Relatively high investment

Gas phase deposition processes are generally classified as either Chemical

Figure 1.5: Schematic of e-beam evaporation apparatus.

Volatile and stable during transport to the CVD reactor.

Figure 1.6: Typical apparatus for CVD experiments, a. direct

Figure 1.7: Flowchart representing approach to the literature review

Silicon based polymers. Gas phase polymerisation of alkoxy silicon

Figure 1.8: Retention of chemical functionality between thin film and

HMDSO [65, 75, 76, 77, 72, 69, 78,

TMDSO [65, 76, 84]

Methyl silane [91]

Styrene [100, 107, 99]

Ethane [101, 108]

Hexane [103, 104]

Fluoro ethene, Fluoro

In general, plasma polymers find applications as protective coatings on

TEOS [120], TRIES [120],

CeCl3, YCl3 + O2 [122]

Thermal excitation (size 980 nm)

SiH4 [128, 123]

CO2 laser excitation

SiH4 + NH3 [124]

CO2 laser excitation

SiH4 + CH4 [124]

CO2 laser excitation

Joule heating / Magnetron

Joule heating / Magnetron

TiCl4 NH3 [125]

Inert gas condensation /

Figure 1.9: Laser activated particle-forming processes, as used by

formation from four siloxane precursors (triethoxysilane, tetramethylorthosilicate,

In order to maintain a good control of film composition, the inorganic and

Anisotropic plasma etching of SiCxOyHz thin films

Chapter 2: Anisotropic plasma etching of

Anisotropic plasma etching of SiCxOyHz thin films

profiles and substrate degradation are investigated using scanning electron

Figure 2.1: Anisotropic reactive ion etching (RIE) of a silica

Anisotropic plasma etching of SiCxOyHz thin films

Anisotropic plasma etching of SiCxOyHz thin films