$4)',‘.-‘:.€M‘4 g“. _

‘vj

findnm, , A

 

‘ v :1” : I"

‘ :fifié’fi
.:. 1-;3 has.
.1 .

,4 .

A '-

r.‘ ,, 4-1 ~
V« .

a , ,L ,..
p. . r. .
:fxg.“ WA?

4

.. "in“
gvxk. .
WK, 1 9:735; «3‘
u *1.“ 35sz
J?"‘l‘ ' ‘

“K

V
a 1.
vi 7:5: w.
d.
"K‘t
1‘.‘

1'3“]?
A.“ ‘f

Fl
1. v

 

in» w.

lllllllllllllllllHllHUllllllllllllWIIIJIIIHIHIIIIIHIII

31293 01051 7971

This is to certify that the

thesis entitled

CHARACTERIZATION OF THE SOL-GEL AND MOCVD
PROCESSES FOR THE DEPOSITION OF ALUMINUM
OXIDE THIN FILMS

presented by

Mark Joseph Waner

has been accepted towards fulfillment
of the requirements for

M- S - degree in Chemig try

 

 

U y‘é‘professo/

Date @‘/ z/?77y

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE N RETURN BOX to romovo this chockout from your mood.
TO AVOID FINES rotum on or bdoro duo duo.

' DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1| I l

MSU is An Afflrmotivo Action/EM Opporuanity institution
Wanna-m

CHARACTERIZATION OF THE SOL-GEL AND MOCVD PROCESSES FOR
THE DEPOSITION OF ALUMINUM OXIDE THIN FILMS

By

Mark Joseph Waner

A THESIS

Submitted to Michigan State University in partial fulfillment of the requirements for the
degree of

MASTER OF SCIENCE

Department of Chemistry

1994

ABSTRACT
CHARACTERIZATION OF THE SOL-GEL AND MOCVD PROCESSES FOR
THE DEPOSITION OF ALUMINUM OXIDE THIN FILMS
By

Mark Joseph Waner

In this Study the sol-gel solution chemistry of a valerate modified aluminum
alkoxide is characterized with FTIR The hydrolysis and condensation of aluminum
divalerate species in solution leads to the formation of a gel which is then spin cast and dip
coated to form thin films. The films are characterized with FTIR, XPS and SEM, before
and after calcination. The layered valerate films form continuous aluminum oxide films
when calcined at 400°C.

The effect of substrate and carrier gas on the deposition of aluminum oxide thin
films from aluminum acetylacetonate was examined. Films deposited in air have less
carbon contamination than those deposited under nitrogen. When KBr, quartz and SiC
were compared, KBr produced the least contaminated films, while SiC showed the largest

amount of aluminum deposited.

ACKNOWLEDGEMENTS

I would like to thank Dr. Jeffrey Ledford for his guidance, assistance and ideas.

Thanks also goes out to my co-workers: Ed, Greg, Jefi‘, Kathy, Mike, Paul, Per and

Radu. A final thanks to my family who have supported my studies from the beginning.

iii

TABLE OF CONTENTS

Page
List of Tables .................................................................................... vi
List of Figures .................................................................................. vii
Chapter 1: Introduction to the Preparation and Characterization of Metal Oxide Diffusion
Barrier Coatings ............................................................................ l
1.1 Overview of Barrier Coatings for use with Silicon Carbide ......... 1
1.2 Sol-Gel Route to Metal Oxide Thin Films ................................... 2
1.3 Metal Organic Chemical Vapor Deposition (MOCVD) ............... 8
1.4 Materials Characterization .......................................................... 13
1.5 Focus of Research ...................................................................... 26
References ........................................................................................ 28

Chapter 2: Characterization of Aluminum Oxide Coatings Derived from Aluminum

Valerate Thin Films ........................................................................... 31
2.1 Introduction ............................................................................... 31
2.2 Experimental .............................................................................. 33
2.3 Results and Discussion ............................................................... 35
2.4 Conclusions ................................................................................. 49
References ........................................................................................ 53

Chapter 3: The Effect of Surface Composition on the MOCVD of Alumina fi'om

Aluminum Acetylacetonate ................................................................ 55
3.1 Introduction ............................................................................... 55
3 .2 Experimental .............................................................................. 57
3.3 Results and Discussion ............................................................... 61
3.4 Conclusions ................................................................................. 65
References ........................................................................................ 68

iv

TABLE OF CONTENTS

Chapter 4: Future Work ...................................................................

4.1 Aluminum Sol-Gel Studies ........................................................
4.2 Aluminum Oxide MOCVD Studies ............................................
4.3 Structure of Model Metal/SiC Interfaces ...................................
References .......................................................................................

7O

7O
73
75
76

LIST OF TABLES

Table Page
Table 2.1: Atomic Ratios for Aluminum Valerate Derived Films ............ 47
Table 2.2: Atomic Ratios for Al(acac)3 MOCVD .................................. 62

vi

LIST OF FIGURES

Figure Page
Figure 1.1: CVD Mechanism ........................................................................ 10
Figure 1.2: Example of Precursor Design in MOCVD .................................. 12
Figure 1.3: XPS Energy Level Diagram ....................................................... 16
Figure 1.4: Inelastic Mean Free Path for Low Energy Electrons ................... 18
Figure 1.5: XPS Chemical Shift Efl‘ect ......................................................... 21
Figure 1.6: Limitation of XPS for Thin Film Analysis ................................... 23
Figure 1.7: SEM Schematic ......................................................................... 24
Figure 1.8: SEM-Surface Morphology ......................................................... 25
Figure 2.1: FTIR Spectra for Reactants ....................................................... 36
Figure 2.2: FTIR Spectra for Aluminum Valerate Sol-Gel Solutions ............ . 38
Figure 2.3: Aluminum Valerate Sol-Gel Solution and Dried Films ................ 42
Figure 2.4: FTIR Spectra : Aluminum Valerate Derived Films ...................... 43
Figure 2.5: FTIR Spectra for Al;uminum Valerate Film Calcination .............. 45
Figure 2.6: Carbon and Oxygen XPS for Films ............................................. 46
Figure 2.7: SEM of as Cast Films ................................................................. 48
Figure 2.8: SEM of as Calcined Films .......................................................... 50
Figure 2.9: SEM of Calcined Dip Coated Film ............................................. 5]

vii

LIST OF FIGURES

Figure

Figure 3.1: MOCVD Depositor ...................................................................
Figure 3.2: FTIR Spectra for Al(acac)3 Deposition ...................................... .
Figure 3.3: SEM Images of SiC and KBr Substrates .....................................

Figure 3.4: SEM Images of Quartz Substrate ................................................

viii

Page
60
63
66

67

Chapter 1

Introduction to the Preparation and Characterization of Metal Oxide

Diffusion Barrier Coatings
1.1. Overview of Barrier Coatings for use with Silicon Carbide

Silicon carbide is an important component of many advanced materials. Its high
strength, hardness, and 2700°C melting point makes SiC quite usefiil as a structural
material. In addition, SiC has a band gap of 2.2 eV and thus is an ideal candidate for a
new generation of high temperature electronics capable of handling large amounts of
power at high fiequency [1].

In recent years the use of ceramic fibers as a reinforcing material for interrnetallic
matrix composites has received a great deal of attention. The fiber/matrix interface acts as
a route for energy dissipation during crack propagation. Ideally, the interaction between
the fiber and the matrix should be strong enough for load transfer from the matrix to the
fiber to occur, but weak enough that the interface will fail before catastrophic failure of
the composite.

Silicongcarbide _ fibers have been shown to be effective as reinforcements in a

unfinzm..-
,_._.— ..
-... -—;,m, ~4.- , _.
"" "h ‘ u .4-......‘.m--J‘ “I" ""-- ~F-Ir'q»...- “1 _,_ ..

variety of ceramic (e.g. A1203, ZrOz) and glass mitiiccs [2-5]. Silicon which;
received much interest as a reinforcement material for interrnetallic composites. The use
of silicon carbide for this application is generally found to be quite limited, however, due
to the reactivity of the fiber with the matrix. It has been shown that SiC will react with
many metals, such as Ti [6,7], Nb [8,9], Al [10-12], Fe [13-15] and Ni [16-20] at typical

composite fabrication temperatures (> 600°C).

Depending on the metal involved, metal carbides or metal silicides may form as shown in

the following reactions:

SiC + 3N1 —~ Ni3Si + C (1)

SiC+ Ti—‘TIC +Si (2)

In many cases these reactions can consume the fibers, leaving the brittle metal carbides
and silicides which degrade the composite performance [21]. The extent of these
reactions may be limited if a diffusion barrier is placed between the SiC fiber and the metal
matrix. Some of the more effective barrier coatings in use are the refractory metal oxides.
A few examples which have shown some chemical compatibility with both the SiC and

metal alloy matrix materials are zirconia, yttria stabilized zirconia and alumina [22-25].
1.2. Sol-Gel Route to Metal Oxide Thin Films

One of the most studied low temperature routes to the production of glass and
ceramic materials is the solution sol-gel process. Sol-gel chemistry most often involves
the formation of metal oxide polymer networks through themcoflntrolled hydrolysis and
polycondensation of precursor materials, typically aqueous metal cations or metal
alkoxides. Through careful control of network formation one may obtain a wide range of
products, from monodispersed colloids to thin films and monoliths. Monoliths produced
by sol-gel techniques may be used as optical filters and lenses. One of the major
commercial uses of sol-gel derived thin films is in the coating of glass in order to modify
its optical properties. Alternating layers of silica and titania have found application as
coatings to produce automotive rear view mirrors since the 1950's [26]. This same

technology has also been successfully used for the coating of large plate glass windows

with a solar reflecting titania layer for use in modern buildings [26]. The techniques of

spray or dip coating, or spin casting are often used for the formation of sol-gel derived
thin films [26].

Through low temperature sintering processes sol-gel glasses may be converted into
amorphous ceramic materials. Crystalline ceramics may be obtained by high temperature
processing. The low sintering temperature enables the fabrication of novel glasses and
ceramic materials which are not compatible at the high temperatures of conventional
ceramic and glass production. In addition to new combinations of glass and ceramic
materials, low temperature processing also enables the modification of inorganic networks
with organic functional groups. This is currently of interest to the optical chemical sensor
field, since inorganic and organic guest molecules can be incorporated into monoliths and
thin films of glass and ceramic materials [27-29].

A sol is best described as a suspension or dispersion of colloidal particles ( 10-
1000 A) in solution. The gel state is not well defined, but can be thought of as an
amorphous colloidal solid which has some liquid component dispersed throughout. In
practice it is not possible to determine precisely the point at which a sol transforms into a
gel or to measure an activation energy for the process. Hench and West define a gel as' a
material that can support a stress elastically [30]. As the sol becomes a gel, colloidal
particles interconnect. Once discrete particles are connected, the newly formed network
continues to connect to other particles and networks.

Although many early sol-gel processes involved the use of metal nitrates or
halides, the hydrolysis of metal alkoxides has received more attention recently. Besides
the ease of hydrolysis, metal alkoxides also have physical and chemical properties which
make them appropriate for use as sol-gel precursors. First, synthesis of metal alkoxides is,
in general, quite facile. Many alkoxides are synthesized simply by reacting the pure metal

with an alcohol [31]:

I2
Al + 3C2H50H —- Al(OC2H5)3 + %H2 (3)

Another widely used synthesis technique is a ligand exchange reaction utilizing the parent
alcohol of the desired alkoxide as the solvent [48]:

SiCl4 + 4C2H5OH —’ Si(OC2H5)4 + 4HC1 (4)

Secondly, most metal alkoxides are easily purified either by distillation or recrystallization
[31]. In addition, the metal alkoxides are soluble in a range of organic solvents, most
notably in solutions of their parent alcohol. This property is critical to the sol-gel process.
Since most sol-gel syntheses involve the hydrolysis of alkoxides, and are often catalyzed
by acids or bases, it is important that the alkoxide be soluble in water miscible solvents
such as alcohols.

The formation of the inorganic network involves two competing types of reaction:

hydrolysis and condensation. The reaction scheme is most simply illustrated as:

M(OR)n + XOH —. M(OR)n,x(OX)x + ROH

The Lewis acidity of the metal center is the major driving force for this reaction scheme.

The reactions occur by a 8N2 nucleophilic substitution reaction illustrated below:

 

 

5+

H H

\5' 5+ \ 6'

O M OR —-> O—M—O—R
x/ x/
(5)
H
/
—"" X 0 M 0 ——-> XO—M + ROH

R

X represents hydrogen for a hydrolysis reaction or a metal center in the case of a
condensation reaction. In some cases a third type of substitution reaction which may
occur is a complexation reaction, where X represents some nucleophilic ligand.

The condensation reaction may occur by two different mechanisms, olation and
oxolation. The dominant reaction pathway depends both on the degree of hydrolysis and

the stability of the various metal species in found in the solution.

H

+

+ fl -
6 8 8

M—OH2 + Ho—M —. M—O—M + H20 (6)

V

Oxolation may occur along with olation when aquo precursors are available, and
exclusively when only hydroxyl species are available. Oxolation is a two step process:

first a nucleophilic addition of the hydroxyl group followed by a 1,3 proton transfer.

a 5‘
H on
M—OH + HO—M—* M—O—M
(7)
5* 5' 5* 5'
H OH H—OH

M—o—M -—- M—O—M —- M—o—M + H20
5‘ 5*

The reactivity of the alkoxide has a great deal to do both with the nature of the metal
center and the alkoxy group used. The reactivity of an alkoxide will increase with the size
and inversely with the electronegativity of the metal center [31,32]. Thus alkoxide
reactivity should increase as one goes down a group in the periodic table. The alkoxy
ligand can also have a dramatic effect on the hydrolysis and condensation reactions
[31,32]. The inductive effect of the alkyl group influences the efi‘ective electropositivity of
the metal center. In addition the size and the steric bulk of the alkoxy group can be
modified in order to control the reactivity of a metal center [31].

While the mechanistic details of silicon sol-gel chemistry are well established [33-
37], the understanding of other main group and transition metal sol-gel chemistry is quite
limited. Further, whereas silicon alkoxides hydrolyze slowly and gel formation is quite
slow, the rate of hydrolysis for transition metal alkoxides is many orders of magnitude
faster. The two major factors which account for this difference are that most transition
metals can have multiple coordinations and, compared to silicon, they are much more
electropositive [31]. When multiple coordination numbers are possible, a transition metal
complex with a low coordination number may undergo nucleophilic attack through a
coordination expansion. The higher electropositivity of the transition metals makes them

more susceptible to nucleophilic attack.

In general, when dealing with sol-gel processes involving silicon alkoxides the
reaction rate must be increased in order to obtain a gel with desired physical properties
(i.e. particle size, pore volume, etc). This is usually achieved through the use of acid or
base catalysts and small alkoxy ligands such as ethoxide and methoxide. In the case of
transition metal sol-gel synthesis, the hydrolysis and condensation reactions often must be
slowed down in order to obtain a gel, rather than a precipitate. This control is most often
achieved through the use of complexing ligands bound to the metal center. A complexing
ligand can lower the reactivity of the metal by decreasing the electropositivity of the metal
and/or by steric hindrance. The reagents most often used are organic acids and B-
diketones [3 8-47]. Organic acids are more strongly binding ligands than the alkoxides and
therefore, make the metal center more stable with respect to hydrolysis. The role of
organic acids in the control of titanium and zirconium alkoxide hydrolysis has received
some attention in the literature [4143,47-51]. B-diketones are strongly binding chelating
ligands which can decrease the rate of nucleophilic attack at the metal center due to the
chelation effect and steric hindrance.

The hydrolysis of aluminum ions has been investigated for over a century [52].
The classical studies of aluminum sol-gel chemistry involve the hydrolysis of aqueous
aluminum ions. In the 1970's Yoldas found that large monoliths of aluminum oxide could
be produced through the hydrolysis and polycondensation of aluminum alkoxides [53,54].
Although differing in synthesis, the products obtained by the hydrolysis of aqueous
solutions are quite similar to the products obtained by the Yoldas method. Brinker and
Scherer note that this similarity may have much to do with the large excess of water in
each of the synthesis methods [52]. The study of aluminum sol-gel chemistry in solutions
of low water concentration has received very little attention in the literature [52].

The typical sol-gel synthesis involves the hydrolysis of a metal alkoxide in an
excess of water (i.e. ~100:l water : alkoxide molar ratio). In some cases small amounts of

acid or base are used as a catalyst for the hydrolysis and/or condensation reaction,

typically the acid or base to alkoxide ratio is 5 1. The conditions for synthesis are highly
dependent on the nature of the alkoxide used. Recently, however, Berglund et al. [48,51]
have shown that high quality films may be produced using a large excess of organic acids
with titanium and zirconium propoxide systems. In these syntheses low molar ratios of
water to alkoxide were used (i.e. ~1.5). One of these studies carefully examined the effect
of carboxyl chain length and acid to alkoxide molar ratio on the film quality [48]. For
titanium isopropoxide sol-gel it was observed that the best films were obtained when the
relatively long chain (valeric and butyric) acids were used, while shorter (propionic) and
longer (hexanoic) carboxylic acids gave fairly poor quality films [48]. It was proposed
that the shorter chains don't provide sufficient steric or nucleophilic hindrance, while the

very long chain acids hinder the polymerization reactions [48].

1.3. Metal Organic Chemical Vapor Deposition (MOCVD)

Chemical vapor deposition (CVD) techniques have been in use for many years as a
synthetic route to refractory compounds at reduced temperatures [61]. A typical CVD
process involves the reaction of gas phase molecules at a hot substrate surface. A wide
variety of reaction types have been utilized, such as pyrolysis, oxidation-reduction, and
hydrolysis [55]. The use of reactive pathways to film deposition separates CVD from the
physical vapor deposition (PVD) techniques such as evaporation and sputter coating. The
classical CVD processes typically involve the reactions of metal halides or hydrides as
precursors, and require temperatures in excess of 1000°C. More recently metal-organic
chemical vapor deposition (MOCVD) techniques have been widely studied as pathways to
coatings at even lower temperatures. An example of the utility of organometallic
precursors involves the deposition of silicon nitride. The classical CVD process for

deposition of silicon nitride is given by

>1000°C _
8H4 + NH3 _"'" 3131\I4 (8)

while Fix et a]. [56] have reported deposition of silicon nitride at temperatures as low as

600°C

. 600°C .
HXSI(N(CH3)2)4_x—> SI3N4 (9)

There are five basic steps in the general CVD mechanism (Figure 1.1), any of
which may be the rate determining step for a particular deposition [55]. Once the reactant

species are in the gas phase, usually diluted by a carrier gas stream, they must diffuse to

, ~___..—u—~--n.;_~u ”Wm

 

,—

the substrate surface. Reactant gases that have diffused through the boundary layer must
be adsorbed onto the substrate. The adsorption process is followed by surface chemical
interactions which lead directly to the deposition products. Besides various types of
reactions, surface interactions may also involve lattice incorporation or surface diffusion
[55]. Once the solid reaction products have formed on the substrate, the byproducts of
the reaction, which by design should be volatile species, desorb fi'om the surface. The
final step in this general mechanism is the diffusion of the byproducts from the surface
region into the carrier gas stream. The kinetics of these processes will often determine
many of the film characteristics and properties [57,58].

MOCVD processing has become a popular coating technique because of the low
temperatures generally involved in the volatilization and decomposition of metal-organic
precursors. However, it is the ability to tailor the properties of precursors that has
attracted the interest of most chemists. In these methods organo-metallic compounds are
designed to provide a core composition around the metal center which is similar in both
stoichiometry and structure to the desired coating. Ligands are chosen to enhance the

volatility of the precursor and reduce gas phase oligomerization. Figure 1.2 shows an

10

Main flow of reactant gases

 

—-—> -——> —>
Gaseous
"‘3 ,"" ' by-products
V l
; , Boundary
G :1. :¢-—@ layer
............ V. .-
interface (negligible

 

<23 ‘1' f mun...)

Diffusion in of reactants through boundary layer
Adsorption of reactants on substrate

Chemical reaction takes place

Desorption of adsorbed species

Diffusion out of by-products

9‘99”“?

Figure 1.1: CVD Mechanism [55]

ll

interesting example of this approach, studied by Interrante et al. [59], which involves the
use of [(CH3)2AINH2]3 as a precursor to aluminum nitride.

The preparation of useful coatings (i.e., desired stoichiometry and morphology)
using MOCVD relies on the interaction of many complex factors. The partial pressures of
carrier and reactant gases, temperature gradients, fluid flow dynamics of the system and
the composition of the precursor and substrate all influence the structure and composition
of the final product [61,60]. The mean free path of the vapor phase reactants and the
diffusion properties of the substrate surface will play a major role in the coating
morphology [58]. Fischer et al. [61] noted that for deposition involving the pyrolysis of
an organometallic compound such as alkoxides and B-diketonates, the use of a carrier gas
with an oxygen source tends to limit carbon contamination in the deposited films.
Although the effect of various precursors, carrier gases and temperatures for many CVD
processes have been studied, relatively little attention has been given to the effect of the
substrate composition and structure.

There has been much investigation of oxide coating deposition by CVD processes

[55,62]. Since the discovery of the high Tc Y-Ba-Cu oxide superconductors, however,

the MOCVD technique has become the focus of increased effort for the deposition of

 

oxide coatings [63-65]. Larkin et a1. [59] have used the MOCVD process to deposit thin
barrier coatings of yttria onto :silicon carbide fiber at 630°C using yttrium B-diketonate
precursors Although this was found to be an effective coating for preventing interfacial
reactions, the coatings produced were not continuous. Additionally, in many CYD
processes problems arise due to the co-deposition of carbon and Slow—igwr‘owth rates of
films While high quality and high purity coatings are commercially produced in
electronics applications, the equipment and materials used are costly. Through a better

understanding of the effect of all CVD parameters, the use of CVD techniques can be

improved.

12

 

[(CH3)2AINH2] 3

 

 

 

 

 

......

 

 

Figure 1.2 : Example of Precursor Design in MOCVD

13

1.4. Materials Characterization

In order to understand the sol-gel and MOCVD processes we need to obtain
detailed molecular level information. For the MOCVD process we need to understand
better the chemical processes which occur at the surface of the substrate. In the sol-gel
process we would like to understand the reaction process which leads to the thin film
formation and to characterize the structure of the oxide films produced. Fourier transform
infrared spectroscopy (FTIR) will provide information which describes the molecular level
structures involved in sol-gel syntheses and the films produced. X-ray photoelectron
spectroscopy (XPS) will be used to provide both structural information as well as
stoichiometric information relevant to the films produced. We will also be using scanning
electron microscopy (SEM) in order to examine the surface morphology of the coatings

and substrates used.
X-ray Photoelectron Spectroscopy (XPS)

In XPS an X-ray photon transfers its kinetic energy to an atom, causing the
ejection of an electron by the photoelectric effect. Conservation of energy requires that

Iw+13itot = Elm, +I~‘.f0t (10)

where Bio, and EL, are the total energies of the initial and final states, respectively, and

Ekin is the kinetic energy of the photoelectron produced. The binding energy of the

electron, referenced to the vacuum level is then equal to

v _ f i
EB-Etot-Etot (11)

14
When (10) and (1 l) are combined they yield the photoelectric equation

hv=Ekin+EE (12)

When measuring the binding energy of a solid sample the energy scale is referenced to the
Fermi level. When a metal sample is examined, an electrical contact is made between the
sample and the spectrometer. Since the sample and spectrometer are now in
thermodynamic equilibrium, they have the same Fermi level. However, the vacuum level
may be different for the sample and the spectrometer. The work function, (I) is used to
describe the difference between the Fermi level and the vacuum level. Photoelectrons feel
a potential which is equal to the difference between the spectrometer and sample work
functions, which leads to the measured kinetic energy being lower than the kinetic energy

of the ejected electron.

E2338 = Ekin " (q’spec " (psample) (13)

The binding energy, referenced to the Fermi level may then be written as

EE=hv-<E$§”+¢.p.c) (14)

The relationship between these energy levels for XPS of a metal sample is illustrated in
Figure 1.3 [66].

Magnesium and aluminum Ka (1253.6 eV and 1486.6 eV respectively) are the
most commonly used X-ray lines for XPS. These X-rays are of high enough energy to
generate photoelectrons from the core levels, as well as valence shells.

In the analysis of solids, low intensity peaks are observed at higher binding energy

than the photoelectron signals. These features are due to the energy loss which occurs

15

when a photoelectron excites a plasmon oscillation in the sample. A classical model is
often used to predict the plasmon frequency. In the model a solid is pictured as a
collection of positively charged spheres within a sea of electrons. If the electron gas
expands 6r fi'om its equilibrium radius, where n is the number of electrons in the shell, an

electric field, é, will be produced.

§=f2—-8n=4nnq Sr (15)

Where q is the charge on an electron. The force generated by this field is given by
F: —q§=-—4rt q2 n8 r (16)

Using the harmonic oscillator approximation the frequency of the plasmon vibration is

2 %

given by

m

While the interaction region of X-ray photons with solids is on the order of
micrometers, the discrete peaks observed are attributed to electrons which escape fi'om
the top 20-50 A of the sample. The electrons produced deeper in the solid contribute to
the backround observed in XPS. Each inelastic interaction which an electron undergoes
will lower its kinetic energy and alter its trajectory. The surface sensitivity of electron
spectroscopies, such as XPS, derives from the limited mean free path of electrons within a
solid. The plot shown in Figure 1.4 is the mean free path of electrons versus the electron
kinetic energy [69]. At kinetic energies below the plasmon energy, the most likely

scattering processes involve discrete excitation of the valence electrons. At higher kinetic

 

 

 

 

 

 

 

 

7 II
SAMPLE % --e‘ SPECTROMETER
i -—L—- i—
i - I
l I
' l
l I
i i l
j : EIIirI
1
Eu": I i VACUUM LEVEL
I / 1 la... ii“.
VACUUM LEVEL - _____ , ______________
h t; j i ¢sooc
" t I I
FERMI LEVEL 5 . FERMI LEVE/L
A/ / 1 '/ /
' I
egmg :
l ' '
J I
k , i :
I
SAMPLE : E SPECTROMETER

Figure 1.3: XPS Energy Level Diagram [66]

17

energies the most common inelastic collisions involving photoelectrons are plasmon
excitations. When excitation of the plasmon fiequency is the dominant source of

photoelectron energy loss, the mean free path, A, may be written as

 

 

(13)

Where v is the velocity of the electron and e is the charge on the electron. The escape
depth, 37», is then on the order of 15-150A for most samples.

Besides being able to determine the identity of the atoms in the surface region of a
solid, the binding energy of photoelectrons can be used to determine atoms which are in
different chemical environments. A binding energy shift, which is often called a chemical
shift, is observed for atoms which are in different oxidation states, or in different lattice
sites or bonding environments. The binding energy shift has as its physical basis that
electrons which are subjected to different potentials will have different energies. As a first
approximation the charge potential model proves to be effective for predicting the
observed binding energy shift for many systems. This model assumes a hollow sphere of

radius r, whose surface has a charge q, equal to the charge of the valence electron shell.

Ei=E9+kqi+Zfll (19)
rij

where Ei is the binding energy of a core level electron on atom i, Eoi is an energy
reference and qi is the charge on atom i. The final term is a summation of the potential at
atom i due to the surrounding atoms, j, which are treated as point charges. From this
purely Coulombic treatment the chemical shift for the initial state comes from the

displacement of electrostatic potential, as in the case of an electronegative ligand drawing

18

 

 

 

 

 

YTYVYI Y Y T f7 YTYI T T Y TYTYUI Y T rrvvvr] Y Y Y
I- -I
Elements

10m:- -‘

: 1

: 1

~ I

I- . <
‘3 1(1)?-
g t
2 t-
o r
C D-

E
,<'
10:-
I
)-
C
1 ° ..

E- rooo "-1

. q

I ' ‘

LIIIAI I L AlLllLl I I IIIIILI I 1 1111111 1 ALL

1 10 100 1000
Energy, eV

Figure 1.4: Inelastic Mean Free Path for Low Energy Electrons [69]

l9

electron density away from a metal center. There is also a shift for the final state due to
the polarization which the atom feels due to the creation of a hole. The core level binding

energy shift for atom i in two chemically different environments is given by

AEi = k AC] '1' AVi (20)
where Vi is given by
VI = 29% (21)
i: j ’11

This is strictly true only when the work function is the same for each material, which is
true for conductors, but not insulators. This model predicts that the chemical shift for all
the core levels, assuming no core-valence interaction, is the same since each level is within
the same sphere of potential. This model also predicts that as the radius ’ij increases the
binding energy shift for the same change in electron density will be lower. Although a
simple model, it is quite useful for predicting experimentally observed phenomena. Ethyl
trifluoroacetate is a good example of the chemical shift effect. This compound has four
carbon atoms which are all chemically different. It is seen from Figure 1.5 that the binding
energy shifts observed for each type of carbon can be easily resolved by an electron
spectrometer [66]. The binding energy shifi effect will be used in this thesis to examine
the differences between different chemical environments of carbon, oxygen and silicon.

In addition to providing information on chemical identity and environment, XPS is
also a useful tool for quantitative analysis of surface stoichiometry. The energy analyzers
used in XPS are generally operated in a pulse counting mode, so that the spectra are
presented as the number of photoelectrons counted as a firnction of energy. The number

of atoms of a given element can be written as a fiinction of the measured intensity.

20

I
=— 22
n GAD ( )

where I is the intensity, 0 is the photoionization cross section, A is the electron mean free
path, D is the analyzer detection efficiency. Scofield has calculated the photoionization
cross sections, 6, for both Mg and Al Kor radiation, treating the electrons of the final and
initial states as moving in the same Hartree-Slater potential [67]. Both theoretical [68]
and experimental [69] determinations of the mean fi'ee path, A, for various materials have
been made. The detection efficiency is a function of operating mode of the spectrometer
and usually varies inversely with kinetic energy. Due to uncertainties in the, A, o and D
the error associated with absolute quantitation is around i30% for XPS [701. However,

the quantitation of atomic ratios can be calculated by

nA _ 1A GBABDB
n13 18 0A AA DA

 

 

(23)

In practice the photoionization cross section, mean free path and detection efficiency can
be grouped together into one term and called a sensitivity factor. When empirically
derived sensitivity factors, such as those found in Appendix 5 of Practical Surface
Analysis by Auger and X-ray Photoelectron Spectroscopy [69] are used to calculate
relative atomic ratios the error can be :i:5% [70].

One limitation of the XPS technique, as it relates to the current study, is that it
cannot differentiate between very thin continuous films (i.e. s 50 A) and thicker but
discontinuous films. Since the photoelectrons have an escape depth on the order of 100 A
or less, silicon 2p photoelectrons fi'om the silicon carbide substrate would be detected in
both very thin continuous films and also in discontinuous films of 100 A or more. This

phenomenon is illustrated in Figure 1.6.

21

 

 

K I i' I
’ F7C_c—O—i-i-H Cis <
i3 F H H
4
m
I— g H _‘
r:
2
D
NIE) _ 3 .
E

 

1 l 1 k

IO 8 6 4 Z O -.
CHEMICAL SHIFT

 

 

 

 

 

 

 

I

 

b 0 d

 

 

 

l l l l l
800 700 600 500 400 300 200
BINDING ENERGY, eV

Figure 1.5: XPS Chemical Shift Effect [66]

22

Scanning Electron Microscopy (SEM)

Scanning electron microscopy makes use of the interaction of a rastered electron
beam with a solid surface. A schematic diagram of a basic SEM is shown in Figure 1.7.
Electrostatic lenses allow a beam of electrons to be focused down to a spot size of 5-200
nm at the sample surface. The most common SEM technique involves the measurement of
the inelastically scattered secondary electrons. Secondary electrons are produced due to
the collision of a beam electron with a conduction electron in the sample. Because of the
low angle scattering of secondary electrons, the beam which exits the sample is not much
wider than the incoming beam. The limit for the image resolution is then dependent more
on the spot size of the incoming beam. The secondary electron image is primarily
dependent on the surface morphology, since more secondary electrons will be produced
from a protrusion in the surface than fi'om a pore as shown in Figure 1.8 [71].

The secondary electrons are typically detected with an Everhart-Thornley detector,
which uses a scintillator connected to a photomultiplier tube. Although secondary
electrons are emitted in all directions, because of their low kinetic energy (3-5 eV), they
can be deflected toward the detector by application of a small potential field. This is
usefill since the sample/beam interactions also produce elastically backscattered electrons.
Backscattered electrons have a kinetic energy on the order of 1-100 keV, and are thus not
easily deflected by a small potential field. Therefore, only the backscattered electrons
which are emitted in the direction of the detector will be detected with the secondary
electrons. Although the high energy primary electron beam has a relatively large depth of
interaction in the sample, the mean free path for the low energy secondary electrons limits

the depth of detection to about 50 A for metals and 500 A for insulators.

23

 

 

 

 

 

A120;

 

   

 

A129

 

 

 

 

 

 

 

 

 

r— l J l -+
co
A120:
A129
100 so so

Bindinametsy (eV)

Figure 1.6: Limitation of XPS for Thin Film Analysis

 

/ Filament

—— Wehneltcap

 

Condenserlens ElecImnGun

 

Objective lens

 

 

 

 

Figure 1.7: SEM Schematic

25

Electron
beam

    
  

 

 

 

 

 

 

Electron
beam
Many
_secondary
electrons
escape
Few
secondary
electrons
L3 1 I‘escape

   

Figure 1.8: SEM-Surface Morphology [71]

26

1.5. Focus of Research

Sol-Gel Synthesis of Aluminum Oxide Thin Films

The present work will first focus on the formation of aluminum oxide thin fihns
through a new sol-gel synthesis using aluminum isopropoxide, which is first modified
using valeric acid and then hydrolyzed in an isopropanol solvent. As this is a new
synthesis procedure we would not only like to characterize the structure of the films
produced, but also to examine the structure of the intermediates in the reaction sequence.
Fourier transform infiared spectroscopy (F TIR) will be used to determine the structure
and types of ligand binding in the reaction solutions, as well as of the as cast and calcined
films. XPS will be used as a tool for determination of stoichiometry and the oxidation
states of the aluminum, carbon and oxygen atoms in the surface region of the as cast and
calcined films. SEM analysis will allow us to characterize the surface morphology of the
films deposited. Using all of this information we will present a chemical picture of the

aluminum sol-gel process.

MOCVD of Aluminum Oxide Coatings

The role of many parameters in CVD processes have been examined in the
literature. One role which has not been given much mention is the effect of the substrate
composition on the deposition of thin films. The purpose of this work is to examine the
role which substrate structure plays in the MOCVD process. SEM analyses will give
information about the surface morphology of the silicon carbide, quartz and potassium
bromide substrates. Thin films, based on the pyrolysis of aluminum acetylacetonate, will
be deposited onto each substrate. FTIR will be used to characterize the structure of the

films deposited onto the KBr. XPS will be used to probe the composition and structure of

27

the deposited films. We will use these data to determine what effect, if any, the substrate

has on the pyrolysis process.

 

'Pn_r math-r1-

28

 

References
1. Mat. Res. Soc. Symp. Proc. 242, Moustakas, T.D.; Pankove, J.I.; Hamakawa, Y.,
Eds; Materials Research Society, Pittsburgh, 1992.
2. Becher, P.F.; Wei, G.C. J. Am. Ceram. Soc., 1984, 67(12), C267.
3. Samanta, S.C.; Musikant, S. Ceram. Eng. and Sci. Proc., 1985, 6(7-8), 663.
4. Panda, P.C.; Seydel, E.R. Am. Ceram. Soc. Bull, 1985, 65(2), 338.
5. Gadkaree, K.P.; Chyung, K. Am. Ceram. Soc. Bull, 1986, 65(2), 370.
6. Bellina, 1]., Jr. ; Zeller, M.V. in Mat. Res. Soc. Symp. Proc. 97, Materials Research
Society, Pittsburgh, 1987, 265. F"
7. Hasegawa, S.; Nakamura , S.; Kawamoto, N.; Kishibe, H.; Mizokawa, Y. Surf
Sci, 1988, 206, L851.
8. Yancy, D.L.; Joshi, A. J Mater. Res, 1990, 5(10), 2197.
9. Misra, A.K. J. Mater. Res, 1990, 5(7), 1561.
10. Draper, S.L.; Gaydosh, D.J.; Nathal, M.V.; Misra, AK. J. Mater. Res, 1990,
5(9), 1976.
11. Foister, S.A.M.; Johnston, M.W.; Little, IA. .1. Mat. Sci. Lett, 1993, 12, 209. I
12. Shinoda, T.; Liu, H.; Mishima, Y.; Suzuki, T. Mat. Sci. Eng, 1991, A146, 91.
13. Kaplan, R.; Klein, P.H.; Addamiano, A. J. App]. Phys, 1985, 58(1), 321.
14. Mizokawa, Y.; Nakanishi, S.; Miyase, S. Jpn. J. App]. Phys, 1989, 28(12), 2570.
15. Mizokawa, Y. ; Nakanishi, S.; Miyase, S. Jpn. .1. Appl. Phys, 1989, 28(12), 2576.
16. Addhoum, H.; Broussaud, D. Mat. Sci. Eng, 1989, A109, 379.
17. Chouand, T.C.; Nieh, T.G. J. Mater. Res, 1990, 5(9), 1985.
18. Eustathopoulos, N.; Chatain, D.; Coudurier, L.; Mat. Sci. Eng, 1991, A135, 83.
19. Nikolopoulos, P.; Agathopoulos, S. ; Angelopoulos, G.N.; Naoumidis, A.;
Grubmeirer, H. J. Mat. Sci.,1992, 27, 139.
20. Yang, J.M.; Kao, W.H.; Liu, C.T. Metal]. Trans. A, 1989, 20A, 2459.
21. Larkin, D.J.; Interrante, L.V.; Bose, A. J. Mater. Res, 1990, 5(11), 2706.
22. Misra, A.K. Metal]. Trans A, 1990, 21A, 441. 1990
23. Bilba, K.; Manaud, J.P.; Le Petitcorps, Y.; Quenisset, J.M. Mat. Sci. Eng, 1991,
A135, 141.
24. Kieschke, R.R.; Clyne, T.W. Mat. Sci. Eng, 1991, A135, 145.
25. Misra, A.K. J. Mater. Res, 1990, 5(7), 1561.
@ Dislich, H. in Sol-Ge] Technology for Thin Films, Fibers, Preforms, Electronics,
' and Specialty Shapes, Klein, L.C.,Ed.; Noyes Publications, Park Ridge, NJ: 1988,
50
— 27. Avnir, D.; Levy, D.; Reisfeld, R. J. Phys. Chem, 1984, 88, 5956.
- 28. Lessard, R.B.; Wallace, M.M.; Oertling, W.A.; Chang, C.K.; Berglund, Nocera,
D.G.;Mat. Res. Soc. Symp. Proc., 1989, 155, 109.
-~ 29. Dulebohn, J.I.; Haefner, S.C.; Berglund, K.A.; Dunbar, KR. Chem. Mater., 1992,
4(3), 506.
—- 30. Hench, LL; West, J.K. Chem. Rev., 1990, 90, 33.
I31. Bradley, D.C.; Mehrotra, R.C.; Gaur, D.P. Metal Alkoxides; Academic Press,

London, 1978.

32.
33.

34.

35.

36.

37.

38.

39.
40.

41.
42.
43.

44.
45.
46.
47.
48.

49.

50.

51.

52.

53.
54.
55.

56.
57.
58.

29

Livage, J .; Henry, M.; Sanchez, C. Prog. Solid St. Chem, 1988, 18, 259.

Brinker, C.J.; Clark, BE; Ulrich, D.R., Eds; Mat. Res. Soc. Symp. Proc. 32,

Materials Research Society, Pittsburgh, 1984.

Brinker, 0].; Clark, BE; Ulrich, D.R., Eds; Mat. Res. Soc. Symp. Proc. 73,

Materials Research Society, Pittsburgh, 1986.

Brinker, C.J.; Clark, BE; Ulrich, D.R., Eds; Mat. Res. Soc. Symp. Proc 121,

Materials Research Society, Pittsburgh, 1988.

Zelinski, B.J.J.; Brinker, C.J.; Clark, BE; Ulrich, D.R.,Eds.; Mat. Res. Soc.

Symp. Proc. 180, Materials Research Society, Pittsburgh, 1990. r
Hampden-Smith, M.J.; K1emperer,W.G.; Brinker,C.J., Eds; Mat. Res. Soc. Symp.
Proc. 271, Materials Research Society, Pittsburgh, 1992.

Mehrotra, R.G.; Gaur, D.P.; Bohra, R. Meta] ,6-diketonates and Allied
Derivatives, Academic Press, London, 1978.

Mehrotra, R.C.; Bohra, R. Meta] Carboxylates, Academic Press, London, 1983.
Sanchez, C.; Livage, J .; Henry, M.; Babonneau, F. J. Non-Cryst. Solids, 1988,
100, 65. i..-
Doeuff, 8.; Henry, M.; Sanchez, C.; Livage, J. J. Non-Cryst. Solids, 1987, 89, 206.

Atik, M.; Aegerter, M.A. J. Non-Cryst. Solids, 1992, 147&148, 813.

Chaumont, D.; Craievich, A.; Zarzycki, J. J. Non-Cryst. Solids, 1992, 147&148,

127.

Debsikar, J.C. J. Non-Cryst. Solids, 1986, 87, 343.

Kundu, D.; Ganguli, D. J. Mat. Sci. Lett., 1986, 5, 293.

Doeuff, 8.; Henry, M.; Sanchez, 0; Livage, J. J. Non-Cryst. Solids, 1987, 89, 84.

Leaustic, A.; Riman, RE. J. Non-Cryst. Solids, 1991, 135, 259.

Gagliardi, C.D.; Dunuwila, D.;Berg1und, K.A. in Mat. Res. Soc. Symp. Proc. 180;

Zelinski, B.J.J.; Brinker, C.J.; Clark, BE; Ulrich, D.R.,Eds.; Materials Research

Society, Pittsburgh, 1990, 801.

Gagliardi, C.D.; Dunuwila, D.; Van Vlierberge-Torgerson, B.A.; Berglund, K.A.

in Mat. Res. Soc. Symp. Proc. 271, Hampden-Smith, M.J.; K1emperer,W.G.;

Brinker,C.J., Eds; Materials Research Society, Pittsburgh, 1992, 257.

Van Vlierberge-Torgerson, B.A.; Dulebohn, J.I.; Berglund, K.A. in Chemical

Processes of Advanced Materials, Hench, L.L.; West, J.K., Eds; John Wiley &

Sons, Inc., 1992, 415.

Severin, K.G.; Ledford, J .S.; Torgerson, B.A.; Berglund, K.A. Chem. Mater., in

press.

Brinker, C.J.; Scherer, G.W. Sol-Gel Science: The Physics and Chemistry of S0]-

Ge] Processing, Academic Press, Inc., London, 1990.

Yoldas, B.E. J. app]. Chem. Biotechnol, 1973, 23, 803.

Yoldas, B.E. Ceram. Bull, 1975, 54(3), 289.

Pierson, H.O. Handbook of Chemical Vapor Deposition: Principles, Technology

andApplications, Noyes Publications, Park Ridge, NJ, 1992.

Fix, R.M.; Gordon, R.G.; Hoffman, D.M. Chem. Mater., 1990, 2(3), 235.

Ghandi, S.K.; Bhat, I.B. MRS Bull, Nov. 1988, 37.

Besmann, T.M.; Stinton, D.P.; Lowden, R.A. MRS Bull, Nov. 1988, 45.

 

59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.

70.
71.

30

Interrante, L.V.; Sigel, G.A.; Garbauskas, M.; Hehna, C.; Slack, G.A. Inorg.
Chem, 1989, 28(2), 252.

Ryabova, L. A.; Sasvitskaya, J. Vac. Sci. T echnol., 1969, 6, 934.

Fischer, HE; Larkin, D.J.; Interrante, L.V.; II/IRS Bull, Apr. 1991, 59.

Vapor Deposition, Powell, C.F.; Oxley, J .H.; Blocher, J .M., jr., Eds; John Wiley
& Sons, New York, 1966.

Gardiner, R.; Brown, D.W.; Kirlin, P.S.; Rheingold, AL Chem. Mater., 1991, 3,
1053.

Cho, C.; Hwang, D.; No, K.; Chun, J.S.; Kim, S. .1. Mat. Sci, 1993, 28, 2915.
Chen, L.; Piazza, T.W.; Schmidt, B.E.; Kelsey, J.E.; Kaloyeros, A.E.; Hazelton,
D.W.; Walker, M.S.; Luo, L.; Dye, R.C.; Maggiore, C.J.; Wilkens, E.J.; Knorr,
D.B. J. Appl. Plrys, 1993, 73(11), 7563.

Feldman, L.C.; Mayer, J.W. Fundamentals of Surface and Thin Film Analysis,
Elsevier Science Publishing Co., Inc., New York, 1986, 225.

Scofield, J .H. J. Electron Spec. Relat. Phenomena, 1976, 8,129.

Penn, D.R. J. Electron Spec. Relat. Phenomena, 1976, 9, 29.

Practical Surface Analysis by Auger and X -ray Photoelectron Spectroscopy,
Briggs,D.; Seah, MP, Eds, John Wiley & Sons, Inc.Chichester, 1983, 186.
Hercules, D.M.; Hercules, 8H. J. Chem. Ed, 1984, 61, 483.

Flegler, S.L.; Heckman, J.W., jr.; Klomparens, KL. Scanning and Transmission
Electron Microscopy an Indtroduction, W.H. Freeman and Co., New York, 1993.

Chapter 2

Characterization of Aluminum Oxide Coatings Derived From Aluminum
Valerate Thin Films

2.1. Introduction

Many early studies of aluminum based sol-gel were performed using metal nitrates
or halides as sources for metal ions [1]. It is also well known that aluminum isopropoxide
and sec-butoxide can be easily hydrolyzed to give various aluminum oxide-hydroxide
phases, including boehmite [2-5]. Many researchers have examined the use of various
inorganic acids and bases (i.e. HNO3, HCl, H2SO4, HF, NH4OH) as catalysts and
peptizing agents [2-5]. Hydrolysis in a variety of organic solvents (i.e. EIZO, benzene,
alcohols) has also been studied [6,7]. The conditions chosen for these studies were largely
dictated by the desire to produce monolithic, powder or colloidal forms of alumina.
Significantly less attention has been given to the production of alumina thin films by the
sol-gel method. There is a large literature on the chemistry involved and the structure of
the gels and precursors for these syntheses [2,5,8-14]. However, there have been
relatively few studies of aluminum sol-gel chemistry in systems using modified alkoxides
or low hydrolysis ratios [8,25,26].

Aluminum carboxylates, or soaps, were studied during the 1950's due in large part
to their use as thickening agents for greases and paints [15-24]. A number of synthetic
routes to aluminum carboxylates were developed during this time. Aluminum halides and
sulfates were found to produce mono and di-substituted soaps when reacted with sodium
carboxylates [17,19] or acid anhydrides of organic acids [23]. The other major synthesis
involved the reaction of aluminum alkoxides with acid anhydrides [20,21] or directly with
the carboxylic acid [16,20-22] in an organic solvent. Throughout much of the early
literature on aluminum soaps there was a debate over whether or not aluminum could

form tri-soaps. Mehrotra and Pande, however,

31

32

showed that the previous failures to produce the tri-substituted soaps of aluminum could
be explained by the displacement of carboxylate by trace amounts of water [20]. Although
many of the soaps produced in these studies were observed to be gelatinous [19], the
potential for sol-gel derived thin film coatings was not investigated. Only very recently
has the use of organic acids for aluminum sol-gel chemistry been given attention in the
literature [25,26]. In their studies, Ayral et al. [25,26] investigated the chemistry of
alumina powders synthesized by the controlled hydrolysis of an acetate modified aluminum
alkoxide in diethyl ether.

One major use of sol-gel derived species is for the formation of thin films.
Alternating layers of silica and titania have found application as coatings to produce
automotive rear view mirrors since the 1950's [27]. This same technology has also been
successfully used for the coating of large plate glass windows with a solar reflecting titania
layer for use in modern buildings [27]. Similarly, alumina thin films have much
technological importance, such as diffusion barriers for composite processing and
insulating layers for electronics applications.

The present study will examine the chemistry involved in the hydrolysis and
polycondensation of a valerate modified aluminum isopropoxide, for the production of
aluminum oxide thin films. The structures involved in solution throughout the reaction
sequence will be followed with Fourier transform infrared spectroscopy (FTIR). We will
also examine the influence of dip coating and spin casting for the deposition of the films
produced. The films have been characterized using F TIR, X-ray photoelectron
spectroscopy (XPS) and scanning electron microscopy (SEM). Once characterized these
films will be calcined, and the structure of the oxide films produced will also be

determined.

33

2.2. Experimental

Materials. Aluminum isopropoxide 98+% and valeric acid 99+% were obtained
from Aldrich Chemical Co. and used without fiirther purification. Isopropyl alcohol
(<0.01% water) was obtained from Mallinckrodt Specialty Chemicals Co. Deionized,
distilled water was used for all syntheses.

Film Synthesis. The valerate films were synthesized based on a procedure
developed by Berglund et a]. [28-30] using agid : alkoxide : waIei ratios of 9:1:l.5.
Valerie acid was added directly to the solid alkoxide in a capped vial. After vortex stirring
and 10 minutes in a sonicating bath, the caps were loosened and the solutions were heated
to approximately 60°C. When most of the alkoxide had dissolved, the solution was
cooled to room temperature and 3.5 mL of isopropanol was added. The reaction mixture
was stirred then heated to approximately 60°C until the aluminum isopropoxide was
completely dissolved. The solution was cooled to room temperature and another 3.5 mL
of isopropyl alcohol was added. Finally, 46 ILL of water was added and the solution was
given a final stirring. Initially upon the addition of water the solution became hazy, but
cleared after it was stirred. Within 10 minutes the solution became a cloudy, white color
and after an hour an opaque gel was formed.

Coating Techniques. Quartz slides, which were 1-2 cm on a side and 1 mm thick
were used as the substrates for XPS analysis, while 13 x 2 mm KBr windows
(International Crystal, Inc.) were used as substrates for the FTIR analyses. Prior to
coating the quartz plates were sonicated in a methanol bath and then rinsed with methanol.
The KBr windows used were polished with aluminum oxide based abrasives and checked
in the FTIR for tranparency prior to coating.

Spin casting The solutions were spin cast onto KBr and/or quartz substrates

which were placed in the center of a tabletop centrifiige. Enough solution was used to

34

completely cover the substrate surface prior to spinning. Each sample was spun for 5
minutes.

Dip coating Substrates were dip coated by hand. Substrates were lowered into
the solution, held there for 5 seconds then withdrawn at a steady rate (~ 2 see withdrawl
time). The edge of the substrate was run along the rim of the vial to allow run off of
excess solution.

Processing. The films were dried in a firm hood for at least 12 hours before
analysis. Calcined films were prepared in a Lindberg tube furnace under a flow of dry air
(100 ec/min, medical grade air, AGA Gas Co.) passed through an in line molecular sieve
filter. After a 10 minute purge, the fumace was heated to 400°C and maintained at that
temperature for 1 hour. The films were then allowed to cool to room temperature before
the air flow was interrupted.

Fourier Transform Infrared Spectroscopy (FTIR). A Mattson Instruments
Galaxy 3020 Fourier transform infrared spectrometer equipped with a standard DTGS
detector was used to obtain infrared spectra. KBr crystals (International Crystal, Inc.)
were used as substrates. Each film spectrum is an average of 16 scans at 2 em’l
resolution. The reference spectrum for alumina was obtained from aluminum oxide
powder (anhydrous, Fisher Scientific Co.), which was suspended in a hexane solvent and
spray coated onto a KBr window.

X-Ray Photoelectron Spectroscopy (XPS). Films cast on quartz slides were
analyzed with a Perkin-Elmer surface science instrument equipped with a Mg anode
(1253.6 eV) operated at a power of 300 W (15 kV and 20 mA) and a model 10-360
hemispherical energy analyzer with an omnifocus small spot lens. The instrument typically
operates at pressures below 1 x 10'8 Torr in the analysis chamber. Spectra were collected
with a constant analyzer pass energy of 50 eV using a PC137 board interfaced to a Zeos
386SX computer. Four regions were examined for each sample: Si 2p, C ls, Al 2p, and

O ls. Binding energies were referenced to carbon (C ls = 284.6 eV). Empirical

35

sensitivity factors were used to calculate atomic ratios [31]. XPS spectra were fit with
20% Lorentzian-Gaussian mix Voigt functions using a non-linear least squares curve
fitting program [32].

Scanning Electron Microscopy (SEM). Samples were coated with an Emitech
K-450 carbon evaporator and mounted on an aluminum stub with a low vapor pressure
double sided adhesive tab (ME. Taylor Engineering, Inc.). Samples were then analyzed at
a 39 mm working distance using a JEOL JSM-35CF Scanning Electron Microscope. The
mierographs shown are the secondary electron images obtained with a beam energy of 10

kV.
2.3. Results and Discussion

FI'IR Spectra. Precursors. The FTIR spectra of aluminum isopropoxide, valeric
acid and isopropanol are presented in Figure 2.1. Some of the bands of the aluminum
isopropoxide spectrum (Figure 2.1a) have been assigned previously [33]. The band at
1041 cm'1 was attributed to the C-0 stretch of the alkoxy group. The series of bands at
703, 683, 616, 573 cm’1 correspond closely to those observed by Barraclough [33] and
Bell et a]. [34] and have been attributed to Al-O stretches.

The assignment of the features in the valeric acid spectrum (Figure 2.1b) has been
presented previously [3 5-3 7]. Methyl and methylene stretching frequencies are observed
at 2962, 2935 and 2876 cm'l, and are superimposed on a broad O-H absorbance centered
around 3000 cm'l. The band at 2671 cm'1 is a combination band of the C-O-H dimeric
carboxylic acid stretch at 1280 cm'1 and the C-O-H in plane bending mode at 1414 cm’l.
The strong absorbance seen at 1711 cm'1 is due to a dimeric carboxylic carbonyl
asymmetric stretch. The features centered around 1460 em‘l are due to methyl and
methylene vibrations. Skeletal vibrations are observed at 1380, 1109, and 750 cm’l. The

strong feature at 940 cm'1 can be assigned to an O-H out-of-plane bending mode.

36

 

(e)

absorbance

 

 

 

 

|
4000 3000 2000 1000

frequency (cm'l)

(3) Aluminum Isopropoxide (b) Valerie Acid (c) Isopropanol

Figure 2.1: FTIR Spectra for Reactants

37

The strongest band in the isopropanol spectrum (Figure 2.1c) is the broad O-H absorbance
seen centered at 3364 cm'l. Methyl and methylene stretching vibrations are observed at
2971, 2931 and 2885 cm'l. Features at 1467, 1408, 1379 and 1340 cm'1 are assigned to
the various C-H bending modes of the isopropyl group. The bands observed in the region
between 1110 and 1160 cm1 are associated with asymmetric C-O stretching modes of
secondary alcohols. The feature at 953 cm‘1 may be assigned to an O-H out-of-plane
bending motion. The sharp feature at 817 cm"1 may be attributed to C-H rocking motions
and C-C stretches of the isopropyl group. The features observed at 655 and 1311 cm'1
are out-of-plane hydroxyl bends [3 7].

Solutions. Figure 2.2 shows the evolution of the FTIR spectra at each step of the
synthesis. Figure 2.2a shows the spectrum for a solution composed of a mixture of
aluminum isopropoxide in an excess of valeric acid. A a 9:1 (acidzalkoxide) molar ratio
was used to prepare this solution, however it should be noted that the spectrum was
collected prior to the complete dissolution of the alkoxide. As expected, the spectrum of
this solution is dominated by valeric acid features, most notably those at 1711, 1469,
1414, 1381, 1278 and 1109 cm'l. Some aluminum isopropoxide features are also
observed as weak bands at 1194, 1167, 1127 and 947 cm'l. Intense new bands are
observed at 1650 and 1585 cm'l. The feature at 1650 cm'1 may be assigned to an
asymmetric COO stretching mode associated with an ionic carboxylate group [35, 37].
The absorbance at 1585 cm"1 is similar to the asymmetric COO stretching fi'equency
observed by Maksimov and Grigor'ev for the diacetate soaps of aluminum [24]. In a
recent study of aluminum basic acetate, Ayral et al. [25, 26] assign bands at 1640 and
1580 cm"1 to the asymmetric COO vibrational mode for bidentate bridging and bidentate
chelating acetates, respectively. The asymmetric COO bands observed at 1650 and 1585
cm‘1 can be assigned to bridging and chelating valerate ligands, respectively. New bands
at 817 and 655 cm'1 correspond to features observed in the isopropanol reference

spectrum. Other new features are observed at 1427, 986, 916, 735, 517, and 494 cm'l,

38

 

 

I 1 gig I
a s3 "
(d) 3
figs
$8
(I:) E

 

absorbance
1469
- 1370

 

 

 

 

 

(b) 5
§
i a
8
(a) 5 a a a
l l l J
2000 1500 1000 500
frequency (cm'l)

(a) Aluminum Isopropoxide and Valerie Acid (1:9) (b) Add Isopropanol (24: 1:9)
(c) Add Isopropanol (48: l :9) (d) Add Water (1.5:482119)

Figure 2.2: FTIR Spectra for Aluminum Valerate Sol-Gel Solutions

39

but are quite weak and exist as shoulders on other stronger bands The shoulder at 1427
cm'1 is similar to that observed for COO symmetric stretches as observed in the spectra of
aluminum diacetates [24]. Early IR investigation of aluminum soaps led to the assignment
of a band at 985 cm1 to a stretching mode for a 2(OH)1 bridging OH moiety [18,22].

The assignment of the 2(OH)l band was not readily made, because it is also similar
to a C-C backbone stretching frequency observed for aluminum acetates [24,3 8].
However, it is most likely related to the 2(OH)1 vibration. If this is a band due to olation
of the aluminum centers then its presence will depend on the extent of hydrolysis.
However, this band is present even in the solutions containing valeric acid and aluminum
isopropoxide. In this solution valerate ligands substitute for isopropoxide groups,

producing isopropanol.

O

AI(OPIi)3 + HO—C(CH2)3CH3 —+A1(0Pri)3_x(OOC(CH2)3CH3)x + PIiOH

Some hydrolysis is expected in these solutions, due to the esterification reaction between
the valeric acid and the isopropanol. The change in the intensity of this band as
isopropanol and water were added support this assignment. This band assignment is
further supported by the calcination temperature study (F igure2.5). This band disappears
from the FT IR spectrum upon heating to 200°C. However, the alkyl stretching
fi'equencies at 2800-3000 cm'l, as well as the symmetric and asymmetric C-O stretching
bands for the valerate ligands are still present.

It appears that the reaction of valeric acid with the isopropoxide results in the
formation of aluminum divalerates and isopropanol. Once iSOpropanol is released,
esterification with the valeric acid can produce water, which hydrolyzes the aluminum
divalerate species. The presence of the 2(OH)1 feature indicates that the hydrolysis results

in condensation by olation.

40

Figure 2.2b is the spectrum observed for a solution containing a 24 : 9 : 1 molar
ratio of isopropanol : valeric acid : aluminum alkoxide. Two new bands appear in the
spectrum. One is a fairly broad absorbance centered at 1305 cm‘1 and another,
overlapping band at 1268 cm'l. A large increase in intensity is observed for the 1380 cm’1
band and also for the band at 1585 cm'l, relative to the dimeric valeric acid C=O stretch at
1711 cm'l. The increased intensity of the 1380 cm'1 band is accompanied by the
formation of a shoulder around 1370 cm'l. The band at 987 cm'l, attributed to the
bridging hydroxyl moiety, has become more resolved from the band at 952 cm'l. The
intensity of the asymmetric carboxylate stretch at 1650 cm“1 is significantly lower in this
solution.

As water is produced from the esterification reaction, hydrolysis and olation
continue as indicated by the change in the 987 cm'1 band. The increase in the band at
1585 cm‘1 and the decrease in the 1650 cm'1 feature, indicates that the hydrolysis process
favors the removal of bridging hydroxyl species and the formation of valerate chelates.

Figure 2.2c shows the spectrum observed for a solution containing a 48 : 9 : 1
molar ratio of isopropanol : valeric acid : aluminum alkoxide. With the increased amount
of isopropanol, three significant changes in the FTIR spectra are observed. The intensity
of the isopropyl skeletal vibration at 1380 cm'1 increases relative to the 1469 cm'1 band,
which may be due to skeletal vibrations of the isopropanol or valeric backbones and/or
asymmetric COO stretches of an aluminum valerate. The feature attributed to the
presence of olated aluminum species, at 987 em’l, is sharper and the intensity of the 1650
cm"1 asymmetric COO stretch of the bidentate bridging valerate ligand decreases.

Figure 2.2d shows the spectrum observed for the final sol-gel solution containing
48 : 9 : 1.5 : 1 molar ratio of isopropanol : valeric acid : water : aluminum alkoxide. The
features are essentially the same as those found before the addition of water.

Valerate Films. Figure 2.3 shows the FTIR spectra measured for the final sol-gel

solution, as well as spin cast and dip coat films after drying in air. The spectra of the films

41

cast by the two coating methods are nearly identical. The only significant differences
between the two coating techniques is that the dip coated films have a much higher
absorbance and a small difference in the intensity ratio for the 1469 and 987 cm'1 bands.
The broad hydroxyl band observed at 3356 cm'1 in the precursor solution disappears upon
drying, but a new hydroxyl feature appears at 3691 cm'l. This sharp absorbance is
characteristic of non-hydrogen bonded hydroxyl groups and is close to the frequency of
the hydroxyl mode observed by Maksimov and Grigor'ev [24] for aluminum diacetate
soaps (3680 cm'l). The dimeric valeric acid carbonyl stretch at 1711 cm'1 observed in the
solution spectra is reduced to a weak shoulder on the asymmetric COO stretching band at
1589 cm'l. In the film spectra, the asymmetric COO stretch, which is similar to the 1591
ch band seen in the case of diacetate soaps, is the dominant feature. A band observed at
1469 cm'1 corresponds to a C00 symmetric stretching frequency as reported for diacetate
soaps of aluminum [24,3 8]. The next major band is observed at 987 cm'1 and may be due
to hydroxyl bridged aluminum moiety [24,38]. In the case of the spin cast films (Figure
2.3b), the symmetric COO stretch at 1469 cm’1 is more intense than the 987 cm"1 band.
However, in the dip coated films the elated aluminum moiety stretch at 987 cm’1 is more
intense than the 1469 cm'1 band. The band at approximately 759 cm“1 is similar to one
observed for the solution spectra. The broad band centered at 655 cm“1 due to
isopropanol is not observed, however two new bands at 671 and 580 cm'1 are observed.

Figure 2.4 gives an expanded view of the 1500 - 900 cm'1 region for the as-cast films. It
can be seen that the shoulder on the symmetric COO mode at 1469 cm'1 is composed of
at least two bands. The shoulder components lie at about 1436 and 1420 cm'l. The
shoulder at 1420 cm'1 can be assigned to an additional COO symmetric stretch mode
associated with the ionic valerate species [3 5]. This mode of vibration has also been
observed in the case of aluminum acetates [24, 38]. The weak feature observed at 1378
cm"1 has been assigned to a C-H deformation mode of the isopropyl group. Weak bands

are also observed at 1366, 1323, 1286, 1244, 1207, 1193, 1111 and 1101 cm'l. The band

42

 

——q
—
—-J

1469

(e)

 

 

 

L

 

 

 

 

 

 

 

 

 

8
g c s
a .. as
M II
a
f _. Es
s E a
(a) i LEV
I l l l
4000 3000 2000 1000

frequency (cm'l)
(a) Precursor solution (b) Spin cast film (c) Dip coat film

Figure 2.3: Aluminum Valerate Sol-Gel Solution and Dried Films

 

43

 

 

 

 

 

 

absorbance
1378
1323
1244
1193
1111
950

(a)

 

1101

 

 

 

frequency (cm'l)
(8) Spin cast film (b) Dip coat film
Figure 2.4: FTIR Spectra: Aluminum Valerate Derived films

44

at 1193 cm'1 is consistent with C00 asymmetric stretching for esters, and monomer
carboxylic acids [39]. Below the 990 cm'1 band a shoulder at 950 cm‘1 is observed.

Calcined F ilms. Figure 2.5 shows the FTIR spectra obtained for a spin cast film
after calcination, as well as a spectrum for alumina powder. The first significant changes
in the FTIR spectra occur at loo-200°C. In this range, the free hydroxyl stretch at 3691
cm'1 disappears while a broader hydroxyl stretching band centered near 3500 cm'1
appears. Also in this temperature range the band centered near 990 cm‘1 is lost. From
200°C to 300°C Al-O stretching frequencies which were two distinct bands at 855 and
664 cm‘l, begin to coalesce giving rise to a less resolved set of bands at approximately
808 and 680 cm‘l. Once calcined at 400°C all of the glxkyl‘styrgetchmflg frequencies are gone,
and the only bands left are the hydroxyl stretch near 3500 cm'l, broad pOOrly resolved
bands in the 1700-1400 cm'1 region and a single, broad Al-O stretching frequency
centered at 709 cm'l. The spectra of the films calcined at 400°C are similar to the alumina
powder spectrum shown in Figure 251‘. The same transitions were also observed for the
dip coated films.

XPS Spectra. Valerate Films. The 0 Is binding energy measured for both spin
cast and dip coated aluminum valerate films is 531.8 eV. Two types of carbon are evident
in the C Is spectra obtained for the as cast valerate films (Figure 2.6). The C 1s
component at 284.6 eV (binding energy reference) is attributed to aliphatic carbon. The
second C 1s peak (288.5 eV) is due to carboxylate carbon. The A1 2p photoelectrons are
observed at 74.1 eV.

The atomic ratios calculated for the as cast films are given in Table 2.1. The
carboxylate carbon to aluminum ratio of 1.75 for the as cast films indicates the presence of
mostly aluminum divalerate species in the film surface. The aliphatic carbon to aluminum

ratio of approximately 7.7, also supports this finding, since each valerate ligand has four

aliphatic carbon atoms. The oxygen to aluminum ratio for the films produced by both

45

 

 

absorbance

 

 

 

 

 

:1, A J

l l l

 

 

 

4000 3000 2000 1000
frequency (cm'l)

(8) Spin cast film (b) Calcined 100°C (c) Calcined 200°C
((1) Calcined film 300°C (e) Calcined film 400°C (f) Alumina standard

Figure 2.5: FTIR for Aluminum Valerate Film Calcination

46

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 Is C Is
I H I
As Cast
G
.3. ‘I F I
q 536 532 528 292 288 284 280
g I I I I I I I
C l o l ’
T L l 1 l 1 1
536 532 528 292 200 284 280

Binding Energy (eV)

Figure 2.6: Carbon and Oxygen XPS for Films

47

coating methods is approximately 5. This ratio and the high oxygen ls binding energy of
531.8 eV indicate the presence of hydroxyl species at the surface of the valerate films

Calcined Films. After calcination at 400°C, 8 C Is photoelectron peak is observed
at the same binding energy as the carboxylate carbon for the valerate films. The 0 Is
binding energy decreases to 531.1 eV for calcined films, which suggests the removal of
surface hydroxyl species during the calcination process. There is no change in the Al 2p
binding energy, as compared to the as cast films. No Si 2p photoelectrons were detected,
indicating that the films are continuous and not cracked or agglomerated.

The oxidized carbon to aluminum ratio of 0.1 suggests that at least 95% of the
valerate ligands have been removed from the surface of the films. The aliphatic carbon to
aluminum ratio was found to be about 0.6 for both the spin cast and the dip coated

samples. The oxygen to aluminum ratios for the calcined films were found to be 2.1.

Table 2.1 : Atomic Ratios for Aluminum Valerate Derived Films

 

 

C (carboxyl):Al C(aliphatic):Al O : Al
Spin Cast / As Cast 1.75i0. 15 76:03 5
Dip Coat / As Cast 1.75:1:025 7.7:t1.l 5.3 :1: 1.3
Spin Cast / Calcined 0.1 0.6:i:0.1 2.1i0.2
Dip Coat / Calcined 30.12 05:02 2.2 i 0.2

SEM Results. Valerate Films. Figure 2.7 shows an SEM image for an as cast,
dried film. The as cast dry films, both dip coated and spin cast, showed significant

cracking and peeling when examined under low magnification. The surface does not look

48

 

Figure 2.7: SEM of as Cast Films

lil

49

like a uniform film, but rather cracked film chips cover the surface.

Calcined Films. Figure 2.8 shows micrographs of calcined films. The calcined
films look much different fi'om the as cast films under low magnification (20-200X). The
calcined films appear continuous and all of the fi'agmented chips seen in the as cast films
are gone. Indeed, at high magnification (2,000 -10,000X) the calcined samples appeared
smooth, with small specks of contaminants s 5 um (Figure 2.8). Although a few small
cracks were observed at 10,000X magnification, most of the film appears to be continuous
for both spin cast and dip coated samples (Figure 2.8b). The dip coated samples appear to
have more variability in their morphology than the spun fihns. An area of patterned

cracking is observed even at low magnification over half of the dip coated film sample

(Figure 2.9).

2.4. Conclusions

FTIR analyses of the aluminum isopropoxide, valeric acid and isopropanol sol-gel
solutions indicates that both bidentate bridging and bidentate chelating ligands are present.
During hydrolysis, the bridging valerate ligands are lost preferentially to the chelating
valerate species.

FTIR and XPS results for films deposited by both dip coating and spin casting are
consistent with the presence of divalerate species observed for the solutions. The films
obtained by dip coating are thicker and show a higher degree of olation than those which
were spin cast. The higher degree of olation may be attributed to the longer drying time
of a thicker film, which allows more time for the solution reactions to occur. These dried
valerate films appear to be layered structures, based on the cracking and peeling observed
in the SEM.

When the valerate films are calcined at 400°C, approximately 95% of the surface

valerate ligands are removed, producing an aluminum oxide film. The film produced

 

50

1 II III U

 

1BHU Hlaflflg

Figure 2.8: SEM of as Calcined Films

51

 

Figure 2.9: SEM of Calcined Dip Coated Film

52

appears to completely cover the surface. Although a few 1 pm cracks were observed in

the SEM images, no photoemission fi'om the substrate was detected.

 

53

References

1.

”>199???"

.‘0

11.

12.

13.
14.
15.
16.
17.
18.
19.

20.
21.
22.
23.
24.
25.

26.
27.

28.

Gitzen, W.H. Alumina as a Ceramic Material, The American Ceramic Society,
Columbus, OH, 1970.

Yoldas, B.E. J. app]. Chem. Biotechnol, 1973, 23, 803.

Yoldas, B.E. Ceram. Bull, 1975, 54(3), 289.

Yoldas, HE J. Mat. Sci, 1975, 10, 1856.

Pierre, A.C.; Uhlmann, D.R. J. Am. Ceram. Soc., 1987, 70(1), 28.

Harris, M.R.; Sing, K.S.W. J. app]. Chem, 1957, 7, 397.

Bye, G.C.; Robinson, J.G. Kolloid Z., 1964, 198,53.

Brinker, C.J.; Scherer, G.W. Sol-Gel Science: The Physics and Chemistry of So]-
Ge] Processing, Academic Press, Inc., London. 1990.

Bye, G.C., Robinson, J.G.; Sing, K.S.W. J. app]. Chem, 1967, 17, 138.
Aldcroft, D.; Bye, G.C.; Robinson, J.G.; Sing, K.S.W. J. app]. Chem, 1968, 18,
301.

Komarneni, S.; Roy, R.; Fyfe, C.A.; Kennedy, G]. J. Am. Ceram. Soc., 1985,
68(9), C243.

Olson, W.L., Bauer, L]. in Mat. Res. Soc. Symp. Proc. 73, Brinker, C.J.; Clark,
BE; Ulrich, D.R., Eds; Materials Research Society, Pittsburgh, 1986, 187.
Nazar, L.F., Klein, L.C. J. Am. Ceram. Soc., 1988, 71(2), C85.

Assih, T.; Ayral, A.; Abenoza, M.;Phalippou, J. .1. Mat. Sci, 1988, 23(9), 3326.
Gray, V.R.; Alexander, A.E. J. Phys. Colloid Chem, 1949, 53(1), 9.

Gray, V.R.; Alexander, A.E. J. Phys. Colloid Chem, 1949, 53(1), 23.

Hood, G.C.; Ihde, A]. J. Am. Chem. Soc., 1950, 72, 2094.

Harple, W.W.; Wiberley, S.E.; Bauer, W.H. Anal. Chem, 1952, 24(4), 635.
Scott, F.A.; Goldenson, J.; Wiberley, S.E.; Bauer, W.H. J. Phys. Chem, 1954,
58, 61.

Mehrotra, R.C.; Pande, K.C. J. Inorg. Nucl. Chem, 1956, 2, 60.

Pande, K.C.; Mehrotra, R.C. Z. anorg. allg. Chemie., 1956, 286, 291.

Leger, A.E.; Haines, R.L.; Hubley, C.E.; Hyde, J.C., Sheffer, H. Can. J. Chem,
1957, 35, 799.

Maksimov, V.N.; Semenenko, K.N.; Naumova, T.N.; Novosleova, A.V. Russ. J.
Inorg. Chem, 1960, 5(3), 267.

Maksimov, V.N.; Grigor'ev, A.I. Russ. J. Inorg. Chem; 1964, 9(4), 559.

Ayral, A.; Phalippou, J.; Droguet, J.C.; in Mat. Res. Soc. Symp. Proc. 121,
Brinker, C.J.; Clark, D.E.; Ulrich, D.R., Eds; Materials Research Society,
Pittsburgh,1988, 239.

Ayral, A.; Droguet, J .C. J. Mater. Res, 1989, 4(4), 967.

Dislich, H. in Sol-Gel Technology for Thin Films, F ibers, Preforms, Electronics,
and Specialty Shapes, Klein, L.C.,Ed., Noyes Publications, Park Ridge, NJ 1988,
50.

Gagliardi, C.D.; Dunuwila, D.;Berg1und, K.A. in Mat. Res. Soc. Symp. Proc. 180,
Zelinski, B.J.J.; Brinker, C.J.; Clark, D.E.; Ulrich, D.R.,Eds.; Materials Research
Society, Pittsburgh, 1990, 801

 

29.

30.

31.

32.
33.

34.

35.

36.

37.

38.
39.

54

Gagliardi, C.D.; Dunuwila, D.; Van Vlierberge-Torgerson, B.A.; Berglund, K.A.
in Mat. Res. Soc. Symp. Proc. 271, Hampden-Smith, M.J.; K1emperer,W.G.;
Brinker,C.J., Eds; Materials Research Society, Pittsburgh, 1992, 257.

Van Vlierberge-Torgerson, B.A.; Dulebohn, J.I.; Berglund, K.A. in Chemical
Processes of Advanced Materials, Hench, L.L.; West, J.K., Eds, John Wiley &
Sons, Inc., 1992, 415.

Briggs, D.; Seah, M.P.; Practical Surface Analysis, Volume 1: Auger and X-ray
Photoelectron Spectroscopy, 2nd Ed, John Wiley & Sons Ltd, Chichester, 1990.
Wagner, C.D.; Davis, DE; Zeller, M.V.; Taylor, 1A.; Raymond, R.M.; Gale, l.
L.H. Surf Interface Anal, 1981, 3, 211.

Googly software provided by Dr. Andrew Proctor, University of Pittsburgh.
Barraclough, C.G.; Bradley, D.C.; Lewis, J .; Thomas, I.M. J. Chem. Soc., 1961,
2601.

Bell, J.V.; Heisler, J .; Tannenbaum, H.; Goldenson, J. Anal. Chem, 1953, 25(11),
1720.

Severin, K.G.; Ledford, J .S.; Torgerson, B.A.; Berglund, K.A. Chem. Mater., in
press.

Colthup, N.B; Daly, L.H.; Wiberley, S.E. Introduction to Infiared and Raman
Spectroscopy, 3rd Ed., Academic Press: San Diego, 1990.

Silverstein, R.M.; Bassler, C.G.; Morrill, T.C. Spectrometric Identification of
Organic Compounds, 3rd Ed., John Wiley & Sons, Inc., New York, 1974.

Ross, S.D.; Inorganic Infrared and Ramon Spectra, McGraw Hill, London, 1972.
Smith, A.L.; Applied Infrared spectroscopy, Fundamentals, Techniques, and
Analytical Problem-solving, John Wiley & Sons, Inc., New York, 1979.

 

Chapter 3

The Effect of Surface Composition on the MOCVD of Alumina From
Aluminum Acetylacetonate

3.1. Introduction

The production of refractory compounds at low temperatures using chemical
vapor deposition (CVD) techniques has been the focus of much attention for the last few
decades [1-3]. Today CVD technology is used to produce a wide range of coatings.
Currently there is much interest in the deposition of oxides, nitrides, borides and carbides.
Although CVD processes generally occur at much lower temperatures than the melting
point of the product, the use of metallo-organic compounds as volatile precursors has led
to even further reductions in processing temperature for many compounds.

One important refractory compound is aluminum oxide. It is commonly used in
optical devices, composite materials and as an electrical insulator. The classic CVD
process used to obtain aluminum oxide involves the hydrolysis of aluminum halides at

1050°C [1,4].

OSWC

I
I
Organometallic compounds, however, have been shown to deposit alumina at
temperatures as low as 300-500°C [5]. The deposits generally obtained at these very low
temperatures are of a polycrystalline or amorphous nature. It has been shown that at
temperatures above 800°C these films show 7 and y' alumina phases [6].

Much of the utility of organometallic precursors derives fiom the ability to design
complexes which mimic both the stoichiometry and the structure of the desired product.

This concept of synthetic design may also be applied to the incorporation of substituents

which increase the volatility of the complex [7].

55

56

Since the discovery of high temperature 1-2-3 superconductors in 1987, the
interest in MOCVD has grown due to the prospect of producing superconducting thin
films via a low temperature route [8,9]. Much of this literature is focused on volatile
metal B-diketonates [10-12]. The B-diketonates are chelating ligands which form ring like

structures when bound to metal centers.

’1
c=o~x

C/ M
\c—c/
l.

The synthesis of metal B-diketonates often involves the displacement of alkoxy ligands
from a metal alkoxide in a B-diketone solvent [13]. The chelating effect of the ligand
makes the metal center less sensitive to hydrolysis, and thus easier to handle than an
alkoxide. Like the alkoxides, these compounds possess sufficient oxygen and high enough
vapor pressures to be suitable for low temperature deposition of oxides. During the 1960's
aluminum acetylacetonate was investigated as a CVD precursor. The literature reports
that aluminum oxide deposits were formed at temperatures of 350-500°C [14].

Much of the research in deposition processes has focused on the effect of
processing parameters, such as temperature, pressure, reactor design and carrier gas flow
rate and composition [2]. Much less attention has been given to the effect of the substrate
surface structure and composition on the deposition process.

In a recent review of CVD techniques, Besmann et a]. [15] mention that the
substrate surface has an effect on growth and nucleation of films, and may act as a catalyst

for the deposition reaction, but no examples of such effects were discussed. Bradley [16]

and Fischer et a]. [17] mention hydroxyl groups as a catalyst for the decomposition of

 

57

alkoxides. Lindstrom and Johannesson determined that the nucleation of aluminum oxide
onto cemented carbide cutting tools could be enhanced by applying a titanium carbide
layer prior to the deposition process [18]. Ryabova and Savitskaya have determined that
the deposition temperature, growth rate and activation energy for the pyrolysis increase
with the conductivity of the substrate [19]. Recently the deposition of high quality
YBazCu3O7-x thin films onto various substrates has been investigated. Commonly the
deposition of these compounds is done by using single crystal dielectric materials [20].
Chen et al., however, have studied the YBa2Cu3O7,x films grown on silver substrates,
using metal B-diketonate precursors [8].

The present study will examine the deposition of aluminum oxide, using aluminum
acetylacetonate as the precursor. The effect of the substrate composition on this process
will be examined using quartz, silicon carbide and potassium bromide as substrates. The
present study will also investigate the effect of the carrier gas composition, specifically

nitrogen and air, on the film deposition onto quartz substrates.

3.2. Experimental

Substrates. Reaction bonded SiC plates (Goodfellow, Co.) were polished with 45
um and 6 pm Metadi II diamond polish (Buehler). These polished samples were sonicated
in hexane and rinsed with methanol. Samples were then placed into a reactor which could
be evacuated to <10 Torr. After evacuation the reactor was backfilled with a 100 ml/min
hydrogen (90.5%)(AGA Gas Co., Cleveland OH) flow and the temperature was raised to
500°C for 1 hour. Samples were then placed in a 50% HF / 50% water solution for 8
hours. Samples were dried and rinsed in methanol and treated again in hydrogen. Quartz
microscope slides which were cut to 1 cm x 1cm were sonicated in methanol prior to
deposition. KBr windows (International Crystal, Inc.; 13 mm x 2 mm) were used without

any further cleaning.

58

Deposition. Figure 3.1 shows the vertical reactor used for all the depositions.
The carrier gas used was varied, and consisted of either: medical grade air, or nitrogen
(99.99 %) (AGA Gas Co., Cleveland OH). The gas flow rate was set to 50 cc/min and
regulated by mass flow controllers (Porter Instrument Co.). 0.50 g of aluminum
acetylacetonate (Aldrich Chemical Co.) was used as the precursor material for production
of aluminum oxide coatings. The depositor was evacuated to a pressure of 1 x 10'2 Torr,
then backfilled with carrier gas. After a second evacuation and backfilling, the substrate
region was heated to 400i1°C. Once the substrate reached deposition temperature, the
precursor was heated to 18811°C regulated using an Omega CN 1200 temperature
controller. Once the precursor temperature was reached, the deposition was performed
for 45 minutes, after which both heaters were shut off. During the deposition the samples
were rotated 60° every 2 minutes.

Fourier Transform Infrared Spectroscopy (FTIR). A Mattson 3020 infrared
spectrometer equipped with a standard DTGS detector was used for each measurement.
KBr crystals obtained fiom International Crystal Inc. were used as substrate materials.
Each spectrum is an average of 16 scans at 2 cm'1 resolution.

The reference spectrum for alumina was obtained from aluminum oxide powder
(anhydrous, Fisher Scientific Co.), which was suspended in a hexane solvent and spray
coated onto a KBr window.

X-ray Photoelectron Spectroscopy (XPS). Samples were analyzed with a
Perkin-Elmer surface science instrument equipped with a Mg anode (1253.6 eV) operated
at a power of 300 W (15 kV and 20 mA) and a model 10-360 hemispherical energy
analyzer with an omnifocus small spot lens. The instrument typically operates at pressures
below 1 x 10"8 Torr in the analysis chamber. Spectra were collected with a constant
analyzer pass energy of 50 eV using a PC137 board interfaced to a Zeos 386SX computer.
Three regions were examined for each sample: C Is, Al 2p, and 0 Is, in addition to K 2p

and Br 3d (for depositions onto KBr) and Si 2p (for depositions onto quartz and silicon

 

59

carbide). Binding energies were referenced to carbon (C Is = 284.6 eV). Empirical
sensitivity factors were used to calculate atomic ratios [21]. XPS spectra were fit with
20% Lorentzian-Gaussian mix Voigt functions using a non-linear least squares curve
fitting program [22].

Scanning Electron Spectroscopy (SEM). The quartz and KBr samples were
mounted on an aluminum stub with a low vapor pressure double sided adhesive tab (ME.
Taylor Engineering, Inc.) and coated with gold in an Emscope sputter coater for 3 minutes
at 20 mA. Samples were then analyzed at a 39 mm working distance using a JEOL JSM-
35CF Scanning Electron Microscope. The micrographs shown are the secondary electron

images obtained with a beam energy of 10-15 kV.

 

6O

 

 

 

Rotary Feed
Exhaust/Pump I:
Tube Furnace
Substrate
Region
Precursor 1:. Gas Inlet
Vessel
Heating Mantle

 

Figure 3.1 :MOCVD Depositor

 

I—

61

3.3. Results and Discussion

FTIR Spectra. Figure 3.2 shows the FTIR spectra for the aluminum
acetylacetonate precursor, fihn deposited at 400°C on KBr and aluminum oxide powder.
The IR band assignments for aluminum acetylacetonate were published by Nakamoto et
a]. [2’]. It should be noted that the spectrum collected is missing the C=O stretch at 1545
cm‘l, C-O stretch and C-H bend at 1466 cm'l, and the out of plane bending mode at 425
cm'1 reported by Nakamoto eta]. [23]. In addition, some features are observed which
are not present in the reference spectrum for aluminum acetylacetonate, namely those at
1446, 1363, 1018, 951, 812, 800 and 786 cm'l. All of these, except for the 800 and 786
cm'1 bands, are weak and appear as shoulders on the aluminum B-diketonate bands.

The spectrum measured for the film deposited at 400°C (Figure 32b) shows bands
characteristic for the Al-O stretches, as observed for aluminum oxide powder, and a
feature at 1384 cm'1 which can be attributed to the incomplete pyrolysis of the aluminum
acetylacetonate.

XPS Spectra. The binding energy measured 0 Is for depositions on SiC and KBr
substrates (531.3 eV) is lower than for the depositions onto quartz (5317-5323 eV).
This is due to the contribution of 0 1s photoelectrons from the quartz substrate. The C Is
binding energy, due to oxidized carbon, measured for the deposition on quartz under
nitrogen (287. 1-287 .4 eV) is lower than that measured for the other depositions (287.6-
288.0 eV). The Al 2p binding energies measured for films deposited onto quartz (74.5-
75.2 eV) are higher than those measured for the depositions onto KBr and SiC (73.8-74.2
eV).

Table 3.1 gives the XPS derived atomic ratios for the coatings deposited on
quartz, silicon carbide and potassium bromide substrates, in different gas atmospheres.
The oxygen to aluminum ratio was corrected to account for the 0 Is photoelectrons

coming from the substrate. To do this the oxygen to silicon (potassium) ratio, measured

 

62
before deposition was divided by the aluminum to silicon ratio (potassium). The surface

enrichment of oxygen may be due in part to the incomplete pyrolysis of aluminum

acetylacetonate, or the adsorption of atmospheric water or C02.

Table 3.1 : Atomic Ratios for Al(acac)3 MOCVD

 

Gas / Substrate OzAI C:A1 C(ox):A1 Al:Si

N2 /SiC 1.9-5.0 $4.9 50.6 18.0-37.2
N2 / SiOz $6.2 4.0-7.2 0.5-1.1 0.4-4.2
Air / 810; 2.4-3.5 2.6—3.8 0.3-0.9 3.2-10.6

 

C(ox):K AI:K

 

Air IKBr 1.7-2.4 0.7-1.1 0.2-0.3 5.6-6.5

 

63

 

 

 

 

 

 

 

 

 

l I l
(C)
8
S
“E
8
'8
Cb)
(a)
I l l L
4000 3000 2000 1000
frequency (cm' I)
(a) Aluminum acetylacetonate (b) 400°C deposition

(c) Aluminum oxide powder

Figure 3.2: FTIR spectra for Al(acac)3 deposition

 

 

 

The carbon to aluminum ratio for the deposition on quartz is lower when air is
used as the carrier gas, compared to when nitrogen is used. This efi‘ect of the carrier gas
has been seen before for depositions using B-diketonate precursors [17]. The lowest
carbon to metal ratio, 0.7-1.1, was observed for the film produced on KBr in an air
atmosphere.

The oxidized carbon to aluminum ratio measured for the deposition on KBr under
air is found to be lower than that of the deposition onto quartz under nitrogen. These
ratios indicate that 95% of the oxidized carbon of aluminum acetylacetonate is removed in
the deposition on KBr in air, while for quartz under nitrogen only 80-90% removed. This
finding may be due more to the carrier gas than the substrate. For the depositions on
quartz under air and SiC in nitrogen the C(ox)/A1 ratio indicates 85-95+% removal of
oxidized carbon. This carbon may have been removed in the form of acetone, as
suggested by Politycki and Hieber [24]

O
a, II
Al(()2CsH7):i W Alxoy + H3C—C-CH3 + Con

and/or due to the breaking of the carbon-oxygen bond in the precursor.

The aluminum to silicon and aluminum to potassium ratios give an indication as to
the relative amount of film growth; although it does not indicate whether the film is
continuous and thin, or thick and agglomerated. The Al/SiCK) atomic ratios for the
depositions on quartz and KBr were found to lie in the 04-106 range. The aluminum to
silicon ratio measured for the deposition on silicon carbide in nitrogen was 18-3 7, possibly
indicating a more efficient film grth process. At present the origin of this effect is not
known. It is interesting to note, however, that the silicon carbide is a fairly covalent
compound, whereas quartz and KBr are more ionic in nature. Another factor which may

play a role in this effect is the sample morphology.

 

*r

65

SEM Micrographs. The SEM micrographs of the three substrate materials used
for depositions are shown in Figures 3.3 and 3.4. The silicon carbide was found to have a
fairly rough surface, with pore sizes of 1-10 pm. When imaged at twice the magnification,
the KBr is found to have small particulates, on the order of 0.3 um, and grooves less than
0.1um wide. The quartz substrate is smooth and non-porous, with stepped terrace

features.

3.4. Conclusions

The FTIR and XPS results indicate that the films deposited from aluminum
acetylacetonate at 400°C are structurally comparable to oxygen enriched aluminum oxide.
This study has shown that the deposition of aluminum oxide from aluminum
acetylacetonate produces cleaner films when air is used as a carrier gas. This study found
that the cleanest films were obtained by deposition onto the ionic KBr crystal lattice under
an air atmosphere. The most efficient deposition, in terms of relative amount of aluminum
deposited, however, was found on the more covalent silicon carbide surface under

nitrogen.

 

66

, . ‘)\ .
stucxzaae near 10.8U c5093

;8HU H4888 8888

 

Figure 3.3: SEM Images of SiC and KBr Substrates

67

 

318888 8889

Figure 3.4: SEM Images of Quartz Substrate

References

1. Powell, C.; Oxley, J .; Blocher, J. M., Jr. Vapor Deposition, John Wiley & Sons,
New York, 1966.

2. Pierson, H.O. Handbook of Chemical Vapor Deposition: Principles, Technology
andApplications, Noyes Publications, Park Ridge, NJ, 1992.

3. Vossen, J .L.; Kern, W. Thin Film Processes, Academic Press, New York, 1978

4. Messier, D.R.; Wong, P. J. Electrochem. Soc., 1971, 118, 772.

5. Matsushita, M. ;, Yoga, Y. Electrochem. Soc. Extend Abstr. N0. 90, Spring
Meeting, 1968, 230.

6. Aboaf, J. A.; J. Electrochem. Soc., 1967, 114, 948.

7. Interrante, L.V.; Sigel, G.A.; Garbauskas, M.; Hehna, C.; Slack, G.A. Inorg.
Chem, 1989, 28(2), 252.

8. Chen, L.; Piazza, T. W.; Schmidt, BE; Kelsey, J. E.; Kaloteros, A. E.; Hazelton,
D. W., Walker, M. S.; Luo, L.; Dye, R. C.; Maggiore, C. J.; Wilkins, D. J.;
Knorr, D. B. .1. App]. Phys, 1993, 73 (11), 7563.

9. Che, C.; Hwang, D.; No, K.; Chun, J. S.; Kim, S. .1. Mat. Sci, 1993, 28, 2915.

10. Gardiner, R.; Brown, D. W.; Kirlin, P. S.; Rheingold, A. L. Chem. Mater., 1991,
3,1053.

11. Snezhko, N.; Moroz, 8.; Petchurova, N. Mat. Sci. Eng, 1993, 818, 230.

12. Buriak, J. M.; Cheatham, L. K.; Gordon, R. G.; Graham, J. J .; Barron, A. R.
Eur. J. Solid State Inorg. Chem, 1992, 29, 43.

13. Mehrotra, R. C.; Bohra, R.; Gaur, D. P. Metal ,6-Diketonates andAllied
Derivatives, Academic Press, London, 1978.

14. Politycki, A.; Hieber, K. in Science and Technology of Surface Coating: a
NA T 0 Advanced Study Institute April 1972, Ed. Chapman, B. N.; Anderson,

J .C., Academic Press, London, 1974.

15. Besmann, T.M.; Stinton, D.P.; Lowden, R.A. MRS Bull., 1988, Nov., 45

16. Bradley, D.C. Chem. Rev., 1989, 89, 1317.

17. Fischer, HE; Larkin, D.J.; Interrante, L.V. MRS Bull., Apr. 1991, 59.

18. Lindstrom, J.N.; Johannesson, R.T. in Proceedings of the Conference on
Chamical Vapor Deposition, 5th International Conference 1974, Blocher, J.M.,
jr.; Hintermann, HE; Hall, L.H., Eds; The Electrochemical Society, 1974, 453

19. Ryabova, L. A.; Sasvitskaya, J. Vac. Sci. T echno]., 6, 934, 1969

20. Chen, L; Piazza, T.W.; Schmidt, BE; Kelsey, J .E.; Kaloyeros, A.E.; Hazelton,
D.W.; Walker, M.S.; Luo, L.; Dye, R.C.; Maggiore, C.J.; Wilkins, D.J.; Knorr,
D.B. J. Appl. Phys, 1993, 73(11), 7563.

21. Briggs, D.; Seah, M.P.; Practical Surface Analysis, Volume 1: Auger and X-ray
Photoelectron Spectroscopy, 2nd Ed, John Wiley & Sons Ltd, Chichester, 1990.
Wagner, C.D.; Davis, L.E.; Zeller, M.V.; Taylor, J .A.; Raymond, R.M.; Gale, L.H.
Surf Interface Anal, 1981, 3, 211.

22. Googly software provided by Dr. Andrew Proctor, University of Pittsburgh.

23. Nakamoto, K.; McCarthy, P.J.; Ruby, A.; Martell, A.E. J. Amer. Chem. Soc.,

68

1961, 83, 1066.

69

24. Politycki, A.; Hieber, K., in Science on Technology of Surface Coatings: A
NA T 0 Advanced Stua)I Institute April 1972, Chapman, B.N.; Anderson, J.C.,
Eds; Academic Press, London & New York, 1974.

Chapter 4

Future Work

4.1. Aluminum Sol-Gel Studies

Film Synthesis. The air dried films produced in this study appeared to be layered,
since a cracked and peeling layer was observed (Figure 2.8), but the substrate was not
observed using XPS. In order to produce higher quality oxide films a more complete
investigation of the synthesis and processing parameters is necessary.

One of the most important factors is the rate of hydrolysis. The chemistry of
alkoxides is well known, especially for the common alkoxides of aluminum such as the
ethoxide, isopropoxide and sec-butoxide. In general the reactivity of the metal center in
an alkoxide is found to decrease with an increase of alkyl chain length and branching. This
is due to both the electron donating capabilities of the alkoxy group and also to steric
hindrance. This would indicate that the hydrolysis process should proceed more slowly
with the sec-butoxide. This will allow us to determine the effect of the residual alkoxide
group on the aluminum sol-gel chemistry.

Another way to alter the hydrolysis chemistry of the sol-gel process is through the
use of complexing agents. In the present work it was shown that valerate ligands were
complexed to the aluminum centers. The use of acetic acid, such as in the work of Ayral
et a]. [1,2], could give information about the effect of the carboxylate chain length on the
hydrolysis rate. Gagliardi et a]. [3,4] have observed a reduction in the hydrolysis rate with
the use of longer chain carboxylate species on titanium centers. It would also be
interesting to see how the presence of a shorter carboxylate chain would effect the carbon
content of the calcined films. B-diketonates are also used as a complexing agent in order

to slow down the hydrolysis of transition metals [5-11]. Acetylacetone

7O

 

71

could be used in place of, or in conjunction with, the carboxylic acids in order to better
control the reaction kinetics of hydrolysis and condensation.

The hydrolysis rate is expected to be directly proportional to the concentration of
water present. In the present experiment water was present as a by product of the
esterification reaction between valeric acid and iSOpropanol as well as from the addition of
water in the final step of the synthesis. The synthesis could be performed without the
addition of water, simply relying on the generation of water fi'om the esterification
reaction. This technique has recently been used by Laaziz et a]. [12] for the hydrolysis of
zirconium n-propoxide.

The present study used a 9:1 molar ratio of carboxylic acid to alkoxide; however,
Gagliardi et a]. [3] have shown that this ratio has dramatic efl‘ects on the properties of
titanium films. In future studies, a variety of acid ratios could be investigated as a means
of better controlling the film properties.

Characterization. Although the FTIR and XPS experiments of the current study
have proven to be useful in the determination of structural information, they cannot
describe the complete picture. For instance the FT IR data was not able to show the
presence of any monodentate ligands due to overlapping bands. The FTIR and XPS
techniques are also not able to determine the aluminum coordination number or the
macromolecular structure of the aluminum valerate network.

Multinuclear NMR techniques have proven to be quite useful for the investigation
of inorganic polymer gels. The formation of silica gels by the sol-gel route has been
widely studied with 29Si NMR [13-16], and the structure of these systems is well known.
27A1 NMR has been used to investigate the structure of alumina gels obtained by the
hydrolysis of alkoxides by excess water [17-20]. These studies have shown the
effectiveness of this method for probing the coordination of aluminum centers in sol-gel
derived gels. Additionally, the use of 1H and 13C NMR might be useful in determining the

more precisely the binding of the valerate ligands. Although the FTIR experiments

 

72

indicate the formation of bridging and chelating valerate ligands, no quantitative
information as to the relative abundance of each type of ligand exists. Further, the
determination of monodentate species cannot be made with the FTIR experiments due to
overlap of the their C-O stretches with other intense bands. Through NMR techniques it
may be possible to determine if this type of binding is present.

Another technique which has been used to follow the sol-gel process is Raman
spectroscopy. Gagliardi et a]. [4] have examined the kinetics of the hydrolysis and
condensation processes for titanium sol-gel systems using this technique. Raman
spectroscopy could be used to study the time evolution of the aluminum sol-gel process.

The present work has focused on gaining molecular level structural information;
however, the long range order of the polymer chains in the gels and the calcined films has
not been investigated. In the case of crystalline samples, where there is very long range
order, diffraction techniques such as X-ray diffraction (XRD) may be used. Although they
are amorphous in nature, the gels and films obtained prior to calcination have some long
range structure due to the presence metal-oxygen polymer chains. The use of small angle
scattering of X-rays (SAXS) has been shown to be quite useful for probing the structure
of gels [14-16]. Several theoretical models have been developed for the interpretation of
SAXS data for polymer like chains [21,22]. Information related to the polymer branching
and fi'actal dimension could be obtained from this study.

Effect of Substrate Modification on the Sol-Gel Deposition. One technique
used in the synthesis of heterogeneous catalysts involves the use of grafting reactions [23].
This technique typically involves the reaction of a metal alkoxide with a hydroxyl group on
a solid surface. This is similar to the alkoxide sol-gel chemistry involved in the current
study. It is expected that metal alkoxide derived sol-gel films might adhere better to a
hydroxylated glass surface than to a silicon carbide surface.

The surface chemistry of silicon carbide has been examined by a number of

workers [24-28]. These studies have found that oxidation of the silicon carbide surface

73

produces a surface layer of silica, whose thickness may be tailored by temperature and
oxygen partial pressure. Further, the presence of silanol groups on silicon carbide has
been observed with FTIR [29]. Sol-gel derived films could be deposited on clean,
oxidized and hydroxylated silicon carbide substrates and their adhesion properties
examined. SEM and transmission electron microscopy (TEM) investigations of the

interface structure could be performed.

4.2 Aluminum Oxide MOCVD Studies

Characterization of Surface Coverage. Although some basic structural
information has been obtained using XPS and FTIR, the surface coverage of the aluminum
oxide films is not known. The deposited films may be very thin, but continuous or they
may be thicker, but agglomerated. The lateral resolution of our XPS instrument ( ~3mm)
and the escape depth of the photoelectrons (~ 10-100A) limits the use of XPS in this case.
If the films are very thin, (< 100A) analysis with SEM is limited. However, the scanning
Auger microprobe (SAM) has the same surface sensitivity as XPS combined with lateral
resolution approaching 100 nm. Thus, the use of SAM could provide some insight into
the surface coverage and nucleation of the films. Another technique which could indicate
the surface coverage of the films is ion scattering spectroscopy, which has monolayer
surface sensitivity.

Evaluation of the MOCVD Process. An extension of the current study is
necessary in order to better examine the issue of substrate morphology on the deposition
process. Prior to the SEM study of the substrate structure, it was hypothesized that an
explanation for the low aluminum content observed for the silica substrates might be
explained by a porous surface, which could fill with the aluminum oxide deposit but be too
deep within the surface to be observed by the XPS. In fact it is the silicon carbide

substrate which is the most porous surface, yet has the most aluminum relative to silicon

74

after deposition. The deposition could be studied as a fimction of surface morphology
simply by polishing with 1 um and .25 um diamond polish.

The present study found differences between the films deposited on silicon carbide
and quartz. As mentioned previously, silicon carbide surfaces can be modified with layers
of silica or silanol functions. An extension of the present study would be a study on the
effect of surface Si-O and Si-OH groups on the deposition process. The ability to alter the
deposition process by simple gas phase treatment could lead to the production of better
diffusion barriers.

Presently we do not know the composition of the gas phase during the deposition
process. For instance, the aluminum acetylacetonate may be partially pyrolyzed before it
gets to the substrate surface. A carefiil study of the effects of deposition temperature and
carrier gas composition may provide some insight into this.

An interesting comparison to the present study would be to use aluminum
alkoxides as precursors. The deposition process could be studied as a function of the
alkoxide chain length. This could provide a better understanding of the pyrolysis
chemistry.

Precursor-Substrate Model Studies. Model studies of single crystal substrates
might provide some insight into the chemistry involved in the CVD process. Simple gas
phase treatments could be used in order to modify single crystal SiC surfaces. After the
evaporation of small numbers (sub-monolayer) of precursor molecules (eg. Al(acac)3),
electron energy loss spectroscopy (EELS) could be used to probe surface vibrations. This
type of information could provide details about the binding mode of precursor to the
substrate. This may also lead to further information on the role of the substrate in the
deposition process. Temperature programmed desorption (TPD) experiments, in
conjunction with EELS could then be used to examine the changes occurring as the

precursor decomposes on the surface.

75

4.3 Structure of Model Metal/SiC Interfaces

In order to evaluate the effectiveness of the oxide coating as a diffusion barrier,
model interfaces could be prepared using Ni, Al, Nb, Fe and Ti. Interfaces could be
prepared by hot pressing metal foils onto oxide coated silicon carbide substrates. After
annealing at composite processing temperatures (~1000°C) these samples could be
sectioned and examined with SEM, SAM and electron microprobe analysis (EPMA). The
SEM could be used to evaluate the extent of reaction at both the metal/oxide and
oxide/SiC interfaces. The SAM and EPMA could be used to determine structural and
compositional information about the reaction zones. The progress of the reactions at the
interfacial region could be studied with an in situ experiment. Thin (<20A) layers of metal
would be evaporated onto the coated silicon carbide samples in a UHV surface science
chamber. XPS would be used to follow the reaction chemistry occurring at different

temperatures.

76

References

1.

u.)

89‘

09°

11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.

23.
24.

Ayral, A.; Phalippou, J.; Droguet, J.C.; in Mat. Res. Soc. Symp. Proc. 121,
Brinker, C.J.; Clark, DE; Ulrich, D.R., Eds; Materials Research Society,
Pittsburgh” 1988, 239.

Ayral, A.; Droguet, J.C. J. Mater. Res, 1989, 4(4), 967 .

Gagliardi, C.D.; Dunuwila, D.; Berglund, K.A. in Mat. Res. Soc. Symp. Proc. 180;
Zelinski, B.J.J.; Brinker, C.J.; Clark, DE; Ulrich, D.R.,Eds.; Materials Research
Society, Pittsburgh, 1990, 801.

Gagliardi, C.D.; Dunuwila, D.; Van Vlierberge-Torgerson, B.A.; Berglund, K.A.
in Mat. Res. Soc. Symp. Proc. 271, Hampden-Smith, M.J.; K1emperer,W.G.;
Brinker,C.J., Eds; Materials Research Society, Pittsburgh, 1992, 257.

Mehrotra, R.C.; Gaur, D.P.; Bohra, R. Metal fi-diketonates andAllied
Derivatives, Academic Press, London, 1978.

Mehrotra, R.C.; Bohra, R. Metal Carboxylates, Academic Press, London, 1983.
Sanchez, 0; Livage, J.; Henry, M.; Babonneau, F. J. Non-Cryst. Solids, 1988,
100, 65.

Doeuff, S.; Henry, M.; Sanchez, C.; Livage, J. J. Non-Cryst. Solids, 1987, 89, 206.
Atik, M.; Aegerter, M.A. J. Non-Cryst. Solids, 1992, 147&148, 813.

Chaumont, D.; Craievich, A.; Zarzycki, J. J. Non-Cryst. Solids, 1992, 147&148,
127.

Debsikar, J.C. J. Non-Cryst. Solids, 1986, 87, 343.

Laaziz, I.; Larbot, A.; Guizard, C.; Julbe, A.; Cot, L. in Mat. Res. Soc. Symp.
Proc. 271, Hampden-Smith, M.J.; K1emperer,W.G.; Brinker,C.J., Eds; Materials
Research Society, Pittsburgh, 1992.

Brinker, 0].; Clark, DE; Ulrich, D.R., Eds; Mat. Res. Soc. Symp. Proc. 32,
Materials Research Society, Pittsburgh, 1984.

Brinker, C.J.; Clark, DE; Ulrich, D.R., Eds; Mat. Res. Soc. Symp. Proc. 73,
Materials Research Society, Pittsburgh, 1986.

Brinker, C.J.; Clark, DE; Ulrich, D.R., Eds; Mat. Res. Soc. Symp. Proc. 121,
Materials Research Society, Pittsburgh, 1988.

Zelinski, B.J.J.; Brinker, C.J.; Clark, DE; Ulrich, D.R.,Eds.; Mat. Res. Soc.
Symp. Proc. 180, Materials Research Society, Pittsburgh, 1990

Komarneni, S.; Roy, R.; Fyfe, C.A.; Kennedy, G.J. J. Am. Ceram. Soc., 1985,
68(9), C243.

Olson, W.L.; Bauer, L]. in Mat. Res. Soc. Symp. Pros 73, Brinker, C.J.; Clark,
DE; Ulrich, D.R., Eds; Materials Research Society, Pittsburgh, 1986.

Nazar, L.F.; Klein, L.C. J. Am. Ceram. Soc., 1988, 71(2), C85.

Assih, T.; Ayral, A.; Abenoza, M.; Phalippou, J. J. Mat. Sci, 1988, 23(9), 3326.
Feigin, L.A.; Svergun, D.I. Structure Analysis by Small-Angle X-ray and Neutron
Scattering, Plenum Press, New York: 1987.

Schaefer, D.W.; Keefer, K.D. Phys. Rev. Lett., 1984, 53(14), 1383.

Baiker, A.; Wokaun, A. Naturwiss, 76, 168, 1989.

Costello, J.A.; Tressler, RE J. Am. Ceram. Soc., 1986, 69(9), 674.

25.

~ 26."
/
27.

28.
29.

77

Muehlhoff, L.; Bozack, M.J.; Choyke, W.J.; Yates, J.T., jr. J. Appl. Phys, 1986,
60(7), 2558.

Bellina, J.J., jr.; Ferrante, J.; Zeller, M.V. J. Vac. Sci. Technol., 1986, A4(3),
1692.

Powers, J .M.; Somorjai, G.A. Surf Sci, 1991, 244, 39.

Ledford, J .S.; Thelen, M.S.; Waner, M.J. Unpublished results.

Ramis, G.; Quintard, P.; Cauchetier, M.; Busca, G.; Lorenzelli, V. J. Am. Ceram.
Soc., 1989, 72(9), 1692.

 

 

re. - «r...
*' '1‘st .
’“ 135777325 '
with:

 

 

1‘52“ #-
""‘r' P gent-1‘ .2‘
.. _,.; w
. . cc .. c. .3

t":.~...‘,r‘- j - J'It‘rfi ‘ »

‘35; L 4, ,Ir a. grant? .136 . I‘m,

' r’
I»
"I

 

w-
«v
. 4

.«p
75:

Jn...'"1;' 1,. .’

'47?"th

. .. V, ,1 :
‘ J'

 

-‘~'I‘y~‘l); m
fuuwv—il m
~- . a s
I
s

.I

 

rm:
UM

~7-
,x

 

?'
- as
i (sh:

I.
vb? ;

a. ,8}: ex.“
4’ 4'2:er g
“$333.”:

I" ‘ ( I
guilt
”I ”if: a
«L @3135.“
‘0

..
’84
~ we}.
_ .

alder, .7

n: n

:- «rs.

-.--i

:

cavI-TWV‘W'

H‘

 

.1
.

sauna-sec:

4

.‘m

g

‘3‘

-. «vs-«waver»