 

 

 

 

 

 

 

 

 

 
 

 
 
 

 
 

 

 

 

 
 

 

 
 
 

 

 

_ . I ‘ ‘ . A , .

 
 
 
 
 
 

THESIS

°\

‘—

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

SOL-GEL DERIVED METAL OXIDE THIN FILMS
DESIGN OF NEW MATERIALS FOR CHEMICAL SENSING

presented by

Kathryn G. Severin

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Chemistry

 

Major professor

Date J0 /30 /%

MSU is an Afﬁrmative Action/Equal Opportunity Institution 0‘ 12771

 

PLACE N RETURN BOX to remove We checkout from your record.
TO AVOID FINES return on or bdore date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU IeAn Afﬁrmative Adlai/Equal Opportunity lnetituion
Wanna-o.-

 

 

SOL-GEL DERIVED METAL OXIDE THIN FILMS
DESIGN OF NEW MATERIALS FOR CHEMICAL SENSING

By

Kathryn Graessley Severin

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

1996

CO. H;
Wm
analmc
metal 0

IDCSC SC

alkoxidl
0de6 n
mom Cl
undersr;
Pftpgm

Which

p.)

Solution

combing

ABSTRACT

SOL-GEL DERIVED METAL OXIDE THIN FILMS
DESIGN OF NEW MATERIALS FOR CHEMICAL SENSING

By

Kathryn Graessley Severin

Semiconductor gas sensors are used for the detection of reducing gases such as
CO, H2, hydrocarbons, and alcohols. When these sensors are constructed from pure
polycrystalline metal oxide semiconductors, they do not have adequate selectivity for
analytical applications in complex gas streams. This research is devoted to the design of
metal oxide thin ﬁlm materials that will improve the sensitivity, selectivity, and stability of
these sensors.

The materials chosen for this research are thin ﬁlms synthesized from metal
alkoxides using sol-gel techniques. This molecular level approach allows the creation of
oxide materials, such as amorphous and mixed metal oxides, which are not accessible by
more conventional fabrication methods. The initial phase of this research focuses on

understanding the relationship between solution chemistry and the characteristics of ﬁlms

q...

N-,—o~v- ..e~ ,‘ ‘. _
M‘ s.

prepared from these solutions. The second phase focuses on modiﬁcation of ﬁlms in ways

 

which are relevant to gas sensing.
Films are made from tin, titanium, or zirconium alkoxides and valeric acid.
Solution reactions are studied with Fourier transform infrared spectroscopy (FTIR) and a

combination of EUR and X-ray photoelectron spectroscopy (XPS) is used to elucidate

chant
Lattice
XPSa
homo g

temps

quart;
coord
arc si
at tcr
10 1h:
that

SnO

how

ﬁlm structure and composition. Oxide ﬁlms are prepared by either H202 treatment or by
calcination of as cast (or “valerate”) ﬁlms at temperatures up to 1000 °C. These are
characterized with XPS, X-ray diffraction (XRD), UV-Vis spectrocopy, and ellipsometry.
Lattice oxygen reactivity, which may be correlated to sensor selectivity, is evaluated with
XPS after in situ H2 reduction. XRD and angle resolved XPS are used to evaluate the
homogeneity of mixed TiOz-SnO; ﬁlms as a function of composition and calcination
temperature.

Valerate and oxide ﬁlms are transparent, stable, and exhibit uniform adhesion to
quartz substrates. Valerate ﬁlms are comprized of an amorphous metal-oxygen network
coordinated with valerate ligands. The compositions of tin oxide and titanium oxide ﬁlms
are similar to those of polycrystalline 81102 and T10; but are largely amorphous if calcined
at temperatures <600 °C. Lattice oxygen reactivity of tin oxide ﬁlms is inversely related
to the temperature used to oxidize the ﬁlms. Results of redox cycling experiments suggest
that ﬁlms undergo very little irreversible restructuring upon H2 reduction. As cast TiOz-
SnOz ﬁlms and those calcined at temperatures 5 400 °C are amorphous and
homogeneously mixed. Calcination at higher temperatures leads to the formation of mixed

rutile phases and to a surface segregation of titanium oxide.

To Blaine and my wonderful boys Geoff and Greg.

iv

ACKNOWLEDGMENTS

This work was made possible only through the help and support of many people. I
would like express my appreciation to Dr. Jeff Ledford for giving me the opportunity to
learn about surface science, sol-gel chemistry, and sensors; for providing the ideas which
were the foundation of this research project; and for giving the encouragement and
guidance necessary for its completion. Thanks also to Professor John McCracken for
taking on the role of advisor upon the departure of Ledford and to my committee:
Professors Berglund, Pinnavaia, and Harrison for their direction.

Additionally, I would like to thank Ron Haas for his patience in the never-ending-
battle to keep the VG surface science instrument running; Russ Geyer for building a
multitude of excellent sample holders, manipulators, bellows, etc. for the VG; and the
glass shop for providing what must have seemed like an inﬁnite supply of quartz slides.

“Successful completion of my research is owed in large part to the other Ledford
group members, especially the other “orphans”: Paul Park, Greg Noonan, and Per
Askeland. They provided not only moral support, but a sounding board for ideas and
insights for data interpretation. Additionally, Ed Townsend (and Maria and Sara), Mike
Thelen, and honorary member Dana Spence contributed greatly to the enjoyment of being

a Ledford group member.

CSCZ

suppc

Nam

Sanity was provided by family and friends. Not only were Blaine, Geoff and Greg
understanding and encouraging, but weekends at the lake with Laurie, Steve and Jason
Frey were great escapes from the world. The encouragement and support of my parents,
Helen and Bill Graessley, was invaluable. Thanks to Cindy Dauphin for helping me to
maintain a sense of humor and for providing perspective on my doctorin’ work.

Funding for this research was provided by the Center for Fundamental Materials
Research at Michigan State University and is gratefully acknowledged. Additional
support was provided by an August F. and Ernest J. Frey Reseach Award and a College of

Natural Sciences Fellowship.

LIST (
LISTC

CHAP

CHAP’I

CHAPT

CA) L») L») LA) La.)

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................... ix
LIST OF FIGURES ..................................................................................................... x
CHAPTER 1 DESIGN OF NEW MATERIALS FOR CHEMICAL SENSING ......... l
1.1 Overview .............................................................................................. 1
1.1.1 Sol-gel Chemistry of Metal Alkoxides ....................................... 2
1.1.2 Materials for Semiconductor Gas Sensors ................................. 4
1.2 Semiconductor Gas Sensors .................................................................. 9
1.3 X-ray Photoelectron Spectroscopy (XPS) ............................................. ll
. 1.3.1 Angle Resolved XPS (AR-XPS) ................................................ 15
1.4 References........... ................................................................................. 21
CHAPTER 2 CHARACTERIZATION OF TITANIUM AND ZIRCONIUM
VALERATE SOL-GEL FILMS ........................................................... 23
2.1 Abstract ................................................................................................ 23
2.2 Introduction .......................................................................................... 23
2.3 Experimental ........................................................................................ 27
2.4 Results and Discussion .......................................................................... 30
2.5 References ............................................................................................ 56
CHAPTER 3 SYNTHESIS AND CHARACTERIZATION OF TIN
VALERATE AND TIN OXIDE THIN FILMS .................................... 60
3.1 Abstract ................................................................................................ 60
3.2 Introduction .......................................................................................... 61
3.3 Experimental ........................................................................................ 63
3.4 Results and Discussion .......................................................................... 66
3.5 References ............................................................................................ 88

vii

CH.A

CHA

CH;

CHAPTER 4 SYNTHESIS AND CHARACTERIZATION OF SOL-GEL

4.1
4.2
4.3
4.4

4.5
4.6

CHAPTER 5
5.1
5.2

5.3
5.4

5.5

5.6

CHAPTER 6

6.1
6.2

DERIVED TITANIUM-TIN OXIDE FILMS ....................................... 91
Abstract ................................................................................................ 91
Introduction .......................................................................................... 92
Experimental ........................................................................................ 95
Results and Discussion .......................................................................... 100
4.4.1 Effect of Preparation on Film Homogeneity ............................... 100
4.4.2 Effect of Calcination Temperature on Titanium Valerate

Films ......................................................................................... 106
4.4.3 Effect of Calcination Temperature on Tin Valerate Films ........... 112
4.4.4 Effect of Mixing and Calcination at 400 °C on Film

Properties ................................................................................. 1 16
4.4.5 Effect of Mixing and Calcination at 600 °C to 1000 °C .............. 120
Concluding Remarks ............................................................................. 134
References ............................................................................................ 136

EFFECT OF PREPARATION TEMPERATURE ON THE

REACTIVITY OF TIN OXIDE FILMS ............................................... 139
Abstract ................................................................................................ 139
Introduction .......................................................................................... 140
Experimental ................................................................................... 143
Results and Discussion .......................................................................... 146
5.4.1 H2 Reduction of Polycrystalline Tin Oxide ................................. 146
5.4.2 H2 Reduction of Tin Oxide Films ............................................... 153
Concluding Remarks ............................................................................. 165
References ............................................................................................ 169
CONCLUSIONS .................................................................................. 171
Project Summary .................................................................................. 171
Future Work ......................................................................................... 174

viii

Table 2.1

Table 3.1

Table 4.1

Table 4.2

Table 4.3

LIST OF TABLES

Atomic Ratios of Carbon and Oxygen Species Calculated from XPS
Intensity Ratios. .................................................................................... 52

Composition of Films: Ratios‘ of Ligands and Backbone Oxygen to
Tin and Estimated Average Tin Oxidation State as Determined by
Quantitative XPS Analysis. ................................................................... 77

Thickness “(1” (nm) and refractive index “n" of ﬁlms as a function
of composition and calcination temperature. Standard deviations
are < 5% for thicknesses and < 3% for refractive indices. ...................... 110

Composition of Valerate Films: Ratios of Ligands and Backbone
Oxygen to Metal and Estimated Average Metal Oxidation State as
Determined by Quantitative XPS Analysis ............................................. 117

Band Gap Energies for ﬁlms as a Function of Calcination
Temperature and Composition. ............................................................. 119

ix

ﬁgure

LIST OF FIGURES

Figure 1.1 Reactions between reducing gases (RH) and metal oxide surfaces
which lead to a response in a semiconductor gas sensor ........................ 5

Figure 1.2 Band structure of a metal oxide material showing both bulk and
surface states. Both the non—equilibrium ﬂat band picture and
equilibrium picttue with band bending are shown. Occupied bands
and surface states are indicated with hatch marks .................................. 10

Figure 1.3 Top: XPS spectrum of a Ti/Sn oxide ﬁlm prepared from an
equimolar mixture of Sn and Ti isopropoxides. Bottom: C ls and
0 Is spectra of a tin valerate ﬁlm. ......................................................... 13

Figure 1.4 The sampling depth in XPS is 37tcosE), where 7t is the mean free
path of the photoelectrons in the material and 0 is the angle
between the surface normal and the analyzer. In AR-XPS the
sample is rotated with respect to the analyzer to obtain different
sampling depths. ................................................................................... 16

Figure 1.5 Trends in concentration measured with AR-XPS for several
possible scenarios. These scenarios give rise to enhanced measured
concentrations of one component in a binary mixture of A and B. ......... 17

Figure 2.1 FTIR spectra of (a) valeric acid; (b) Ti(OPr‘)4; (c) isopropanol; (d)
Zr(OPr“)4 in n-propanol; and (d) n-propanol. ........................................ 31

Figure 2.2 FTIR spectra of (a) Ti(OPr‘)4; (b) 1:1 mixture and (c)l:1.5 mixture
of Ti(OPr')4 and valeric acid; ((1) 1:15:15 mixture of Ti(OPr')4,
valeric acid, and water. ......................................................................... 34

Figure 2.3 F'TIR spectra of (a) Zr(OPr")4 in n-propanol; (b) 1:1 mixture and

(c) 1:2 mixture of Zr(OPr”)4 and valeric acid; ((1) 1:2:1.5 mixture of
Zr(OPr")4. valeric acid. and water. ........................................................ 38

Figure 2.4 FTIR spectra of (a) TiOD, (b) Ti24, and (c) ZrOD ﬁlms ........................... 44

Figure 2.5

Figure 2.6

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4
Figure 3.5
Figure 3.6
Figure 3.7
Figure 4.1

Figure 4.2

Figure 4.3

FTIR spectra of hydroxyl region: in situ heating of 2100 ﬁlm (a)
before heating; (b) heated to 50°C; (c) heated to 100°C; (d) cooled
to room temperature in dry nitrogen; (e) exposed to ambient air for
30 minutes. ......................................................................................

Oxygen ls photoelectron spectra measured for (a) Ti00, (b) Ti24,
and (c) Zr00 ﬁlms. ...........................................................................

FTIR spectra of (a) valeric acid and (b) Sn(OPri)4 in isopropanol
and spectra of solutions containing different concentrations of
valeric acid and water. Sn(OPr')4 : valeric acid : water ratios are:
(c)1:0.5:0,(d)1:1:0,and(e)1:1:2. Spectraarenormalized
to the isopropanol band 1769 cm". ..................................................

FTIR spectra of (a) SnV and (b) SnC films. .....................................

FTIR spectra of sequential treatments of an SnV ﬁlm: (a) initial
spectrum collected at 25°C; heated in nitrogen to: (b) 50°C, (c) 75
°C, and (d) 100°C; (e) cooled to 25°C in nitrogen, and (f) exposed
to ambient air at 25°C. .....................................................................

Valence band photoelectron spectra of SnV ﬁlm (a) initially and (b)
after 12h of X-ray exposure. ............................................................

Valence band photoelectron spectra of (a) tin(IV) oxide, (b) SnC

ﬁlm, and (c) SnP ﬁlm. ......................................................................

FTIR spectra of SnV ﬁlm on Si crystal (a) before and (b) after
hydrogen peroxide treatment (SnP), and (c) the uncoated Si crystal.

FTIR spectra of tin valerate gel (a) before and (b) after hydrogen
peroxide treatment. ..........................................................................

FTIR spectra of valerate ﬁlms: (a) SnVal, (b) STVal, (c) EMVal,

(d) TSVal, and (e) TiVal ..................................................................

AR-XPS analysis of 50%Ti 50% Sn valerate ﬁlms prepared using
Methods 1 and 2 I and using Method 3 O. The curves represent
the best ﬁt to the model. ..................................................................

XRD patterns of calcined TiVal ﬁlms: (a) TiC250, (b) TiC400, (c)
TiC600, (d) TiC800, and (e) TiC1000 ..............................................

xi

..... 48

..... 51

..... 67

..... 70

..... 74

..... 80

..... 83

..... 84

..... 86

..... 101

..... 104

..... 107

Figur

Figur

Flgur

Flgur

Figur

Figur

Figure 4.4

Figure 4.5

Figure 4.6
Figure 4.7
Figure 4.8

Figure 4.9

Figure 4.10
Figure 4.11

Figure 4.12

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

XRD patterns of ﬁlms calcined at 1000 °C: (a) SnC1000, (b)
STC1000, (c) EMC1000, (d) TSC1000, and (e) TiC1000. .................... 109

XRD patterns of calcined SnVal ﬁlms: (a) SnC250, (b) SnC400, (c)
SnC600, (d) SnC800, and (e) SnC1000. ............................................... 114

UV-Vis spectra of films calcined at 400 °C: (a) SnC400, (b)
STC400, (c) EMC400, (d) TSC400, and (e) TiC400..............._. ............. 119

XRD patterns of calcined STVal ﬁlms: (a) STC400, (b) STC600,
(c) STC800, and (d) STC1000 .............................................................. 126

AR-XPS analysis of ISTVal, OSTC400, ‘STC600, OSTCSOO,
and OSTC1000 ﬁlms ............................................................................ 127

XRD patterns of calcined EMVal ﬁlms: (a) EMC400, (b) EMC600,
(c) EMC800, and (d) EMC1000. .......................................................... 129

AR-XPS analysis of IEMVal, OEMC400, 4EMC600,
OEMCSOO, and OEMCIOOO ﬁlms. ...................................................... 130

XRD patterns of calcined TSVal ﬁlms: (a) TSC400, (b) TSC600,
(c) TSC800, and (d) TSC1000. ............................................................. 132

AR-XPS analysis of ITSVal, OTSC400, ATSC600, OTSC800.
and OTSC1000 ﬁlms....'...........~. ............................................................ 133

XPS valence band spectra of polycrystalline tin oxide pellet (a)
initially; after sequential H2 reductions for 1 h. each at (b) 200 °C,
(c) 250 “°C,. and (d) 300 °C: and (c) after reoxidation in 20% 02
(He) for 1 h. at 200 °C .......................................................................... 147

XRD pattern of polycrystalline SnOz (a) initially, with most intense
cassiterite peaks labelled, and (b) after H; reduction at 350 °C for
0.5 h. Peaks due to B-tin are designated with a dot (0) and major
B-tin peaks are identiﬁed ....................................................................... 148

XPS valence band spectra of (a) SnP, (b) SnC200, (c) SnC250, (d)
SnC400, and (e) SnC600 ﬁlms. ............................................................. 155

Changes in oxide concentration with reduction temperature as
determined with XPS for ISnP, OSnC200, OSnC400, and
OSnC600 ﬁlms. ................................................................................... 156

xii

Figur

E1 gur

Figure 5.5 XPS valence band spectra of tin oxide materials after H2 reduction
at 300 °C: (a) SnP, (b) SnC200, (c) SnC250, (d) SnC400, and (e)
SnC600 ﬁlms. ....................................................................................... 157

Figure 5.6 Oxidation - reduction cycling of SnP ﬁlm: Clinitial ﬁlm; I200 °C,
0250 °C, and 0300 °C H2 reduction; and 0200 °C 20% O2(He)
oxidation. Oxide/tin atomic ratio is determined with XPS analysis
after each sequential in situ treatment. .................................................. 159

Figure 5.7 Oxidation - reduction cycling of SnC200 ﬁlm: Dinitial ﬁlm; I200
°C, 0250 °C, and 0300 °C H2 reduction; and 0200 °C 20%
O2(He) oxidation. (a) Oxide concentration and (b) tin oxidation
state as calculated using charge balance from XPS determined
oxide and hydroxyl concentrations after sequential treatments. .............. 161

Figure 5.8 Oxidation - reduction cycling of SnC400 ﬁlm. Elinitial ﬁlm; I200
°C, 0250 °C, and 0300 °C H2 reduction; and 0200 °C 20%
O2(He) oxidation. Oxide/tin atomic ratio is determined with XPS
analysis after each sequential in situ treatment. Scale is adjusted to
that of Figure 5.7a to avoid an artiﬁcal over emphasis of
differences. ........................................................................................... 164

Figure 5.9 AR-XPS analysis of the formation of metal in an SnC400 ﬁlm after
H2 reduction for 4h at 350 °C. .............................................................. 167

Figure 5.10 Chemical shift between Sn 3d5,2 metal and oxide peaks and the

oxide/tin oxide atomic ratio as a function of sampling depth for
SnC400 ﬁlm after H2 reduction for 4h at 350 °C. .................................. 168

Figure 6.1 Oxide concentration as a function of reduction temperature for

OTiC400. DSnC400, OTSC400. IEMC400, and OSTC400 ﬁlms.
The standard deviation in the reported atomic ratios is i005. ............... 176

xiii

1.1

580510:
alkoxi
prcpa
impor
allow
rcc0g
Suital
S€nsi

Sensi

Sen:

Che

Chapter 1

Design of New Materials for Chemical Sensing

1.1 Overview

This research is part of a larger study to develop new materials for chemical
sensing. The materials that are central to this research effort are made from metal
alkoxides using sol-gel techniques. Sol-gel chemistry is very versatile and materials can be
prepared which are stable, transparent, porous and conductive, properties which are
important for chemical sensing. The synthesis can be carried out at room temperature
allowing the incorporation of sophisticated organic and organometallic molecular
recognition sites. Additionally, the rheological properties of the precursor solutions are
suitable for the formation of thin ﬁlms. Films can be coated onto optical ﬁbers for optical
sensing, surface acoustic wave devices for mass sensing, and arrays for conductimetric
sensing.

This study is separated into two parts. First, the basic chemistry involved in
formation of these ﬁlms is explored and its relationship to the composition and structure of
the resultant ﬁlms is established. Then, the study focuses on modiﬁcation of the ﬁlms in
ways which may improve their sensitivity, selectivity and stability when used as
semiconductor gas sensor substrates. While the desire to make new materials for chemical
sensing is the motivation behind this study, these ﬁlms have a wide range of physical and

chemical properties which are potentially USCfIJI for a variety of applications. Among

cnher

cmah

invoh
conde
fbnne
nunen

reacux

matena

The ﬁr:

others, these ﬁlms can be used as membranes, optical and protective coatings, and
catalysts.

1.1.1 Sol-gel Chemistry of Metal Alkoxides. Traditionally, sol-gel chemistry
involves the reaction of silicon alkoxides with water [I ]. The hydrolysis and
condensation reactions which result are slow, and amorphous silicon oxide glasses are
formed. Extension of this chemistry to metal alkoxides opens up the possibility of creating
materials with a wide range of properties. However, transition metal alkoxides are more

reactive than silicon alkoxides and, unless some control is exerted, polycrystalline oxide

 

and oxohydroxide precipitates are formed when they are exposed to moisture [2 ]. Their
hydrolysis rates can be slowed down by replacing reactive alkoxide groups with strong
ligands such as .carboxylates or B-diketonates. Gels and ﬁlms can then be prepared from
these modiﬁed metal alkoxides [3 - 5 ].

It is generally acknowledged that there are three steps in the formation of sol-gel
materials-from metal alkoxides (M(OR)4) and carboxylic acids (R'COOH) [1-3, 5, 6].
The ﬁrst is nucleophilic substitution of alkoxide ligands by carboxylate ligands.

M(OR)4 + x R'COOH —) M(OOCR'),(OR)4_X + x ROH 1.1
Next, the addition of water causes hydrolysis of the carboxylated metal alkoxides and
alkoxide ligands are replaced with hydroxyl groups. 1
aM-OR + H-OH —> sM-OH + ROH 1.2
The parent alcohol is a byproduct of both of these reactions. Hydrolysis initiates the third
step: condensation.

sM-OH + HO-ME —> EM-O-ME + H20 1.3a

Hydrox
product
solution
mixture
hydroxi
substrat

1
free ﬁlm
hydroly;
(Tit OPr

TiiOPr‘]

knowiec
{mined 1
by anal}-
imated

mecham

EM-OR + H0-M—z--—) sM-O-Ma + ROH 1.3b
Hydroxyl groups react with either hydroxyl or alkoxy groups on other metal centers
producing M-O-M bonds and water or the parent alcohol, respectively. The resultant
solution consists of a metal oxide network dissolved in an alcohol - carboxylic acid
mixture. The metal centers may be coordinated with carboxylate, alkoxide, and/or
hydroxyl ligands. Films can be formed by spin casting, dip coating, or spray coating
substrates with these solutions.

Berglund et a1. [7 , 8 ] found that high quality thin ﬁlms (stable, transparent, crack-
ﬁ'ec ﬁlms which exhibit uniform adhesion to glass substrates) could be spin cast from
hydrolyzed. mixtures of valeric acid (C4H9COOH) and titanium(IV) isopropoxide
(T i(OPri)4). Optimum reactant molar ratios were found to be 9 valeric acid : 1.5 water : 1
mom...

The research presented in Chapter 2 is undertaken to achieve a more detailed
knowledge of the chemistry involved in the production of these ﬁlms, as well as ﬁlms
formed from zironium(IV) n-propoxide (Zr(OPr“)4) and valeric acid. This is accomplished
by analyzing the solutions at various stages in the synthetic process with Fourier transform
infrared spectroscopy (F TIR). Information obtained is related to the rate, extent, and
mechanism of the reactions in these systems.

Also established in Chapter 2 are relationships between the solution chemistry and
the structure and composition of the ﬁlms produced. This knowledge should lead to
better control of the sol-gel process and ideally, to the ability to produce ﬁlms which are

tailor-made for particular applications. The composition, including the nature and

C hapt
relex'a
functrt

functic

concentration of ligands, and the structure of titanium valerate and zirconium valerate
ﬁlms are elucidated using FTIR and X-ray photoelectron spectroscopy (XPS).

1.1.2 Materials for Semiconductor Gas Sensors. The research presented in
Chapters 3, 4, and 5 is guided by the desire to design materials which have properties
relevant to semiconductor gas sensing. In general gas sensors must perform two
functions: a receptor function allowing recognition of a particular gas and a transduction
ﬁrnction in which the gas concentration is converted to signal output. In the case of
semiconductor gas sensors, reactions on the surface of a metal oxide semiconductor lead
to changes in surface conductivity.

Currently, there are three types of semiconductor gas sensors in use commercially.
These include sensors for the detection of reducing gases (i.e. H2, CO, hydrocarbons,
ethanol) which are based upon SnO2, In2O3, or F e203; H2S sensors based upon SnO2 and
W03; and humidity sensors. Additionally, though air/ﬁre] ratio (A/F) is automobile
exhaust is now monitored with ZrO2zY203 solid electrolyte sensors, TiO2, CeO2, and
szOs semiconductor gas sensors are currently under development for this application
[9 ].

For the detection of reducing gases an n-type semiconductor is used. Reactions
between reducing gases and metal oxide surfaces which lead to sensor response are
illustrated in Figure 1.2 [9]. Reducing gases (RH) can dissociatively chenrisorb on metal
oxide (M-O) surfaces and act as electron donors. Alternatively, they can react with either

ionosorbed oxygen or lattice oxygen and when the products desorb, inject electrons into

Mechc

Rec

rig

Mechanism

- RHO
‘I’ /'

 

 

-M-O-M-O-M-O— -M-O—M—O—M—O—
I44 -1\|/I-O-I\I/I- RH ’_6_l_l_l l_l
if I I I I I I [I i) ’14—? i4—
-M-O—M—O—M-O- —M—O—M—O-M-O-
RH RH
Tit—ﬂ“? Ti? T?’
T’I‘TTi‘N‘ ‘i’N‘VN‘T‘Y"
—M—o-M—0-M—o— —M-0-M—0—M-o—
Reactions

RH(g) ‘—> R+ (ads) + H+ (ads) + 2 e'
RH(g) + 0’ (ads) —> ROH (g) + e'

RH(g) + 02' (lat) —> ROH (g) + 2 c"

Figure 1.1 Reactions between reducing gases (RH) and metal oxide surfaces which
lead to a response in a semiconductor gas sensor.

the surfac
Settion l

T
oudeIS
high ter
polycrys
are cor.
because

than thl

prepar
‘v'alcraI
ﬁlms t
in air

COIla:

the surface. The effect of these reactions on the surface conductivity is discussed in
Section 1.2.
The semiconductor used most commonly for the detection of reducing gases is tin

oxide (SnO2). Sn02 is chemically inert and therefore stable when sensors are operated at

high temperatures in air. Sn02 sensing elements are in the form of compressed
polycrystalline powders or thin ﬁlms [10 ]. The use of ﬁlms is advantageous because they
are compatible with the fabrication of microsensors, have rapid response rates, and,
because these sensors function at elevated temperatures, they have lower operating costs
than their pressed powder counterparts [11 , 12 ].

Chapter 3 focuses on the preparation of tin oxide ﬁlms. First, tin valerate ﬁlms are
prepared and characterized using the same methods described for titanium and zirconium
valerate ﬁlms. Then, two methods are used to remove the organic components of the
ﬁlms to produce the tin oxide ﬁlms needed for gas sensing. The ﬁrst method is calcination
in air at 400 °C. ' This is a conventional method but calcination commonly leads to the
collapse of sol-gel materials which lowers their surface area [1]. This is an advantage for
most coating applications but a disadvantage for both sensor and heterogeneous catalyst
applications; A high surface area maximizes interaction between the substrate and the gas
and means improved response for sensors (both in time and senSitivity) [l3 ] and a more
active catalyst. In an attempt to remove the organic components while limiting the extent
of collapse, a room temperature chemical oxidation with hydrogen peroxide is

investigated. -

llusi
forse
thanh
the C(
anhu

prodt

the nu
heate.
prepa
lidié
E\ em
idenu
can
abﬂnt
SeleCtI

Chapter 4 explores the preparation of mixed metal oxides using sol-gel techniques.
This is a promising approach for tailoring the physical and chemical prOperties of materials
for semiconductor gas sensing. The system chosen for this study is tin oxide mixed with
titanium oxide. These oxides were chosen because Takahashi and Wada [14 ] found that
the conductivity of TiO2-SnO2 ﬁlms was strongly dependent upon the components of the
ambient gas, making them ideal candidates for gas sensing materials. Indeed, gas sensors
produced from mixed TiO2-SnO2 materials have been shown to have better sensitivity
[15 ,16 ] and response times [16, 17 ] than the single metal oxide materials. Second, both
oxides have rutile structures and may crystallize in mixed phases rather than crystallize
separately; This improves the chances of maintaining a homogeneous mixture of the
oxides even after calcination at elevated temmratures.

Important questions addressed in Chapter 4 are: how homogeneously mixed are
the metal oxides and how, does the homogeneity and crystallinity change as the ﬁlms are
heated in air? Films with Ti:Sn atomic ratios of 100:0, 75:25, 50:50, 25:75, and 0:100 are
prepared and their homogeneity is evaluated with XPS, F TIR, and angle resolved XPS
(AR-XPS). AR-XPS is used to detect the surface segregation of one of the oxides, an
event which commonly occurs in heterogeneous mixed materials [18 ]. The presence,
identity, and particle sizes of crystalline phases are determined with X-ray diffraction
(XRD).

One of the most important features of any chemical sensor is its selectivity or its
ability to detect a particular molecule in the presence of others [19 ]. Commonly the
selectivity of semiconductor gas sensors is improved by the addition of noble metal

catalysts and metal oxide promoters [20]. Other potentially useﬁrl strategies for

jmprml
semicor
occurnt
the mos
an impc

lattice c

coordrn
POW?
are obt
reactivi
200 CC
used {c
[Educm
Oxide 5
are IWI
Signiﬁc;
Select“,
redUCltc

Stable sr

improving sensor selectivity can be found by analogy to selective oxidation catalysis. Like
semiconductor gas sensors, catalyst performance depends upon the selectivity of reactions
occurring on metal oxide surfaces [9]. It is known that lattice oxygen reactivity is one of
the most important factors in the selectivity of catalysts [21 - 23 ]. It may therefore play
an important role in the selectivity of semiconductor gas sensors and ﬁlms with a range of
lattice oxygen reactivities may be selective for different gases.

Since the reactivity of oxygen in defect sites is greater than that of fully
coordinated bulk sites [21], amorphous materials may be more reactive than
polycrystalline materials. In Chapter 5 tin oxide ﬁlms with varying degrees of crystallinity
are obtained through oxidation of amorphous ﬁlms at temperatures up to 600 °C. The
reactivities of these materials are evaluated using H2 reduction at temperatures between
200 °C and 400 °C in a reactor attached to the surface science instntment. XPS analysis is
used to determine the stoichiometry and the oxidation state of tin before and after H2
reduction. The effect of calcination temperature on the structure and reactivity of tin
oxide ﬁlms is evaluated and compared to that of a polycrystalline tin oxide powder. There
are two important issues. . First, do ﬁlms prepared at different temperatures have
signiﬁcantly different oxide reactivities? If not, this is not a promising approach to making
selective sensor materials. Second, do ﬁlms undergo irreversible restructuring upon
reduction? Film stability is evaluated through oxidation-reduction cycling. Potentially

stable substrates are those whose reactivity is unchanged by repeated treatments.

1.2 Sen
Sen
reattions c
surfaces at
reactionsc
oxide sent
Ti
oxide sen
bonding I
EDEIgy g;
With ban
ﬁner
“Till em
IRRICC c
IOOm It
middle

band

can be
{mm

SUI-fa

1.2 Semiconductor Gas Sensors

Semiconductor gas sensors respond to changes in conductivity which result from
reactions on their surface [9, 24 ]. Reactions between reducing gases and metal oxide
surfaces are shown in Figure 1.1. The increase in conductivity which results from these
reactions can be understood by examining the band structure and surface states of a metal
oxide semiconductor (Figure 1.2).

The right hand side of the top ﬁgure illustrates the bulk band structure of a metal
oxide semiconductor. For tin oxide, the top of the valence band is comprised of non-
bonding 0 2p orbitals and the bottom of the conduction band of Sn Ss orbitals. The
energy gap between these two bands (band gap) is approximately 3.6 eV. While materials
with band gaps of this magnitude are generally classiﬁed as insulators, tin oxide behaves
as an 'n-type semiconductor because it naturally has a high concentration of defect states
with energies <0.1 eV below the conduction band [25 ]. These defects are primarily
lattice oxygen vacancies. Electrons from these states can populate the conduction band at
room temperature. In the absence of defects, the Fermi level would be located in the
middle of the band gap. Their presence moves the Fermi energy closer to the conduction
band.

The left side of the top ﬁgure shows the energies of the surface states. The surface
can be thought of as a defect in the crystal in which the top layer of atoms has been
removed. Even though restructuring occurs to lower the surface energy, energy levels in
these coordinatively unsaturated surface sites will be different than those in the bulk.

Surface metal atoms can be viewed as electron donor sites, lattice oxygen as acceptor

FlatI

Bane

FlSure ]

Flat Bands

10

ConducﬁonBand
99999999
++++++++

 

 

 

 

 

 

 

 

 

 

 

< Er
& .
\ \ \ \ \
Val Band
\\\\ \enee \ k\\\
Surface Bulk
Band Bending
II E
+ Conduction Band
I: + 9 9 9 9 9 9 9 9
w +, + «r- + + + + + + «r-
E. /” E.
. i‘ \ \
§ Veer Bed&\\\
Surface Bulk
Figure 1.2 Band structure of a metal oxide material showing both bulk and surface

states. Both the non-equilibrium ﬂat band picture and equilibrium picture
with band bending are shown. Occupied bands and surface states are

indicated with

hatch marks.

sites Su:

chemisorbI

gases act a

and the lov

since the F t

levels. elect
causes band
at the surfac
semiconduc
the adsorbec
Operating tei
Whe:

causes the in.
at the suffac
adsmbed 0X
PTOdUCts deg

lower the p01

13 x“,

In th;

phototflemo,

 

PhotoejectrOn

have a mean I;

 

11

sites. Surface states are also created by the presence of adsorbed species. Oxygen
chenrisorbed ﬁ'om ambient air acts as an acceptor at the surface. Cherrrisorbed reducing
gases act as donors. At the surface, the Fermi level is between the highest occupied state
and the lowest unoccupied state. As drawn, this is obviously not an equilibrium situation
since the Fermi level at the surface is below that of the bulk. In order to equalize the two
levels, electrons ﬂow from the near surface region to acceptor states on the surface. This
causes band bending, as shown in the bottom illustration, and a potential barrier is created
at the surface which decreases the surface conductivity. This is the situation for an n-type
semiconductor gas sensor when ambient oxygen is adsorbed on its surface. The nature of
the adsorbed oxygen species is temperature dependent but is predominantly O' at normal
operating temperatures (~ 300 °C) [26 ].

When a reducing gas chemisorbs on the surface, it acts an electon donor. This
._ causes the injection of electrons into the near surface region, lowering the potential barrier
at the surface. This in turn, increases the surface conductivity. Both lattice oxygen and
adsorbed oxygen are acceptor sites. When they react with a reducing gas and the
products desorb, electrons from those sites remain. They enter the near surface region,

lower the potential barrier, and increase the surface conductivity.

1.3 X-ray Photoelectron Spectroscopy (XPS).

In this technique, a sample is irradiated with X-rays causing the emission of
photoelectrons. The kinetic energy of these electrons is measured by the analyzer.
Photoelectrons are exponentially attenuated by the material from which they emerge and

have a mean free path (A) which is strongly dependent upon their kinetic energy. Ninety

ﬁve perce
of the sur

100 A

IKE). the

The bindi
originated
relaxation
unique set
can be dg
phmoemjs
phOIOEICCI

Sig

(III) in the

12

ﬁve percent of the photoelectrons which reach the detector have emerged ﬁ'om within 31
of the surface. Typically, this limits the sampling depth of XPS to a maximum of 50 -
100 A.
Binding energies (BE) can be determined from photoelectron kinetic energies
(KB), the energy of the X-rays (hv), and the spectrometer work function (Imp):
BE=hV'¢Ip'KE 1.4
The binding energy is characteristic of the atomic orbital from which the electron
originated. Binding energies differ from ionization energies by ﬁnal state effects because
relaxation processes occur simultaneously with photoemission. Since each element has a
unique set of binding energies associated with it, the elemental composition of a surface
can be determined from its XPS spectrum. Shown at the top of Figure 1.3 is a
photoemission spectrum of a mixed TiO2-Sn02 ﬁlm. Peaks shown are due to
photoelectrons from O] 3, Sn 3d, and Ti 2p orbitals.
Signal intensities (1;) in XPS spectra can be related to the number density of atoms
(m) in the surface of a sample by the following relationship [27 ]:
1. = 10n.7\2(8.)D(€.) 1.5
where 10 is the X-ray ﬂux, 0,- is the photoionization cross section (probability that a
emission will occur from this orbital; this is dependent upon the excitation energy), 117(5))
the mean free path of the photoelectron (this is a function of its kinetic energy (8.)), and
D(£i) the transmission function of the instrument.
XPS is seldom used to measure absolute concentrations of elements because the 10

and D691) are diﬂicult to determine. Of these only the kinetic energy dependence of D(£,-)

\A:.l.:U~—_—

A-Z-U~F-

Bi;

F’SUre L3

l3

XPS Spectrum of Ti/Sn Oxide Film

 

 

 

 

 

 

 

 

 

 
  

 

 
   

T l
Sn 3d
_ 0 Is _
l
.15?
3:3 _ _
.5 Ti 2p
Carbon ls Oxygen 13
I I I I I I I I
SnV Alkyl SnV
Backbone
a?!
E. .
_ Carboxyl

    

 

 

 

 

288

284

Binding Energy (eV)

536 534 532 530 528 526

 

 

 

 

 

Binding Energy (eV)

Figure 1.3 Top: XPS spectrum of a Ti/Sn oxide ﬁlm prepared from an equimolar
mixture of Sn and Ti isopropoxides. Bottom: C 1s and 0 Is spectra

of a tin valerate ﬁlm.

remains It

The inter
the bracl
theoretic
universal
though 1‘.
with him
factors

COmpos:

empm'cz

en“for
Spectra
Carbon
SUch a
“Suit,

ngEr

14

remains if the relative concentrations (C) of two elements A and B are calculated:

9a -14 ___°8A'B£_Bo's
CB 13 OAK/4820.5 1.6

The intensity of a peak is taken to be its area after background subtraction. The term in
the bracket on the right is the sensitivity factor. Sensitivity factors can be determined
theoretically using cross sections determined by Scoﬁeld [28 ] and mean free paths from a
universal curve [29 ]. Unfortunately, mean ﬁee paths can be material dependent and
though it is commonly assumed that D(s;) is proportional to at“, its relationship changes
with kinetic energy and varies diﬁ‘erently for different instruments [30 ]. Better sensitivity
factors can be determined empirically by the analysis of standard materials of known
composition. For example, from the spectrum in Figure 1.3 it was determined, using
empirical sensitivity factors based on the analysis of polycrystalline SnO2 and TiO2, that
the relative concentrations of the elements in the surface of this ﬁlm are 1 Ti : 1 Sn : 4 0.
These were the relative concentrations expected since this oxide ﬁlm was made from an
equimolar mixture of Ti(OPri)4 and Sn(OPri)4.

Even more speciﬁc chemical information can be obtained because chemical
environment inﬂuences binding energies. Also shown in Figure 1.3 are the C Is and 0 1s
spectra of a tin valerate ﬁlm. In the C Is spectrum two peaks are present, one due to alkyl
carbon and the other to carboxyl carbon. Carbon bonded to an electronegative element
such as oxygen has less electron density around it than carbon bonded to C or H. As a
result, the electrons are more tightly bound. For this reason, the carboxyl C 1s peak has a

higher binding energy than the alkyl C Is peak. Similarly, when oxygen is bonded to Sn

 

(backbone

charge and

 

relative co
their relatiI
1.3
varied by I
way depth
Cc
Splitter a\
by XPS -
difﬁcult
because
over SDU

man'th'rrur

SUrfaCe 1
(compo:
AB rat;
the acn

(F4

“jrh n0

RaIIOS

15

(backbone oxygen) as opposed to C or H (ligand oxygen), it has a higher net negative
charge and therefore, the 0 Is peak for backbone oxygen has a lower binding energy. The
relative concentrations of the different types of carbon or oxygen can be determined from
their relative peak areas.

1.3.1 Angle Resolved XPS (AR-XPS). The sampling depth of XPS can be
varied by rotating the sample with respect to the analyzer as shown in Figure 1.4. In this
way depth proﬁling can be done in the near surface region of the sample [31 ].

Commonly depth proﬁling is done by bombarding a surface with ions which
sputter away the sample. Relative concentrations of sample components are determined
by XPS analysis as the layers are removed. Unfortunately, with sputter proﬁling it is
diﬁcult to distinguish between ion induced damage and actual sample composition
because lighter atoms are preferentially sputtered. AR-XPS analysis has an advantage
over sputter depth proﬁling because it is non-destructive. It is limited, however, to the
maximum sampling depth of XPS, which is typically 50 - 100 A.

AR-XPS can be used to qualitatively assess the distribution of species in the near
surface region of the sample. A few of the many possible scenarios for a binary mixture
(components A and B) are illustrated in Figure 1.5 along with the trends expected in the
NB ratios measured with sampling depth. (Trends are are shown as linear functions but
the actual dependence is more complex (vide inﬁa)). The ratios are normalized by
dividing by the relative concentrations in the bulk mixture. In a homogeneous mixture
with no surface segregation, the normalized ratio should be 1 for all sampling depths.

Ratios other than I can be the result of one component migrating to the surface

 

ﬁgure 14

 

 

hv

Figure 1.4

16

 

 

 

 

 

 

 

hv

 

The sampling depth in XPS is 3Acos6, where A is the mean free path of the
photoelectrons in the material and 0 is the angle between the surface
normal and the analyzer. In AR-XPS the sample is rotated with respect to
the analyzer to obtain different sampling depths.

17

Some Possible Scenarios

 

 

 

 

 

 

 

(D
e: 1
E
.2
r E 4
<
S "
“D
it.
E"
3
2 2
0 Sampling Depth (A) 50

Figure 1.5 Trends in concentration measured with AR-XPS for several possible
scenarios. These scenarios give rise to enhanced measured concentrations
of one component in a binary mixture of A and B.

Icon‘

fonn

enhai
conc:
depth
asth.
sufat

suiac

Scena
rano

suiac

18

(conversely, the other component migrating away from the surface) or due to the
formation of crystallites which have a different A/B ratio than in the original bulk mixture.

In Scenario 1, component A has segregated at the surface forming a surface layer
enhanced in A. The layer may have a uniform concentration or there may be a
concentration gradient. As long as the layer thickness is less than the maximum sampling
depth (3 A), the measured A/B ratio would be greater than 1 at the surface and approach 1
as the sampling depth increases. Scenario 2 is a similar situation, except that B is the
surface enriched species. In this case, the normalized A/B ratio is less than 1 at the
surface and increases with sampling depth.

Crystallites which are enriched in component B have formed at the surface in
Scenario 3. As a result of the crystallization, the bulk of the sample is enriched in A. The
ratio measured for shallow angles will be less than I because the crystallites are at the
surface. It is possible that as the sampling depth increases the ratio becomes greater than
1. This inversion of the measured ratio would happen if there are few crystallites and they
have diameters much greater than 31. In this case for the greater sampling depths, only
the surface of the crystallites is sampled and the majority of the signal is from the B-
deﬁcient amorphous bulk of the sample.

In Scenario 4, crystallites enriched in B have formed again resulting in a bulk
which is enriched in A. In this case, the crystallites have formed below the maximum XPS
sampling depth. In this case, the measured ratio is independent of the sampling depth but

is greater than 1.

surface
concent
differen
An equ;

from 1hr

a materi

Where I
it; is th.
eIipressir
ratio (R;

Peaks L15

rePlaced

15 e

OfRsand

19

Quantitative information can be obtained if a simple model is used to describe the
surface segregation of one component of a binary mixture (Scenarios 1 and 2 with no
concentration gradient). In this model, a single surface layer of thickness “d” has a
different relative concentration of the two components (R.) than found in the bulk (Rb).
An equation was derived, as follows, to get an estimate of the thickness of a surface layer
from the relative concentrations measured as a function of analysis angle with AR-XPS.

It is commonly acknowledged that photoelectrons are exponentially attenuated by

a material [31]:

I." —
1,. =f£2§(z)el‘dz 1.7

i

where I, is the peak intensity of element i, If is its intensity in an elemental bulk standard,
A, is the effective escape depth, and )C (2) is the mole fraction of i at depth 2. This
expression can be written in terms of ratios of intensities (R, and Rf ) and mole fraction
ratio (Rx(z)) for a binary mixture when A, is the same for both species. This is valid if the
peaks used for quantitation have similar binding (kinetic) energies. Additionally, A, can be
replaced with 7." cost) to indicate its angular dependence:

_ R7 ~ 4.2:
RI—mj; Rx(z)e" “dz 1.8

R? is essentially a sensitivity factor so R,/Rj’ can be replaced with R, the relative

concentrations. Assuming the surface layer has a thickness d and a relative concentration

of R, and the bulk has a relative concentration Rb, then this expression becomes:

 

.. 1 “Fire rail.—
RTA°cosG[R‘-I°e dz+Rbfe 0dz] 1.9

Sohing th

Estimates
between IT.

Wt
particular .

compositio

 

 

20

Solving the integrals results in the following equation:
-_d__ ._E_
R=R‘[1-e 2L"°°‘°)+Rb(e mm] 1.10

Estimates of R, , d, and Rb can be obtained by minimizing the sum of the differences

between measured relative concentrations and those generated using this equation.
Whether it is valid to use this equation to estimate a surface layer thickness for a

particular sample should be judged based upon the trends in the AR-XPS data and the

composition and size of any crystallites present in the sample.

1.4

. J

Lo)

LII

‘J

10

1.4

10.

11.

12.

l3.

14.

15.

16.

21

References
Brinker, C.J.; Scherrer, G. Sol-Gel Science, the Physics and Chemistry of Sol-Gel
Processing, Academic Press: San Diego, 1990.

Bradley, D.C., Mehrotra, R.C.; Gaur, D.P. Metal Alkoxides; Academic Press:
London, 1978.

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

Mehrotra, R.C.; Gaur, D.P.; Bohra, R. Metal ,B-diketonates and Allied
Derivatives, Academic Press, London, 1978.

Mehrotra, R.C.; Bohra, R. Metal Carboxylates, Academic Press, London, 1983.
Sanchez, C.; Livage, J. New]. Chem, 1990, 14, 513.

Gagliardi, C.D.; Berglund, K.A. in Processing Science of Advanced Ceramics,
Aksay, I.A.; McVay, G.L.; Ulrich, DR, Eds; Mater. Res. Soc. Proc. 155:
Pittsburgh, PA, 1986; 127.

Gagliardi, C .D.; Dunuwila, D.; Berglund, K.A. in Better Ceramics Through
ChemistryIV, Zelinsky, B.J.J.; Brinker, C.J.; Clark, DE; Ulrich, D.R., Eds;
Mater. Res. Soc. Proc. 180: Pittsburgh, PA, 1990; 801.

Madou, M.J.; Morrison, S.R. Chemical Sensing with Solid State Devices;
Academic Press, Inc.: San Diego, CA, 1989.

McAleer, J. F.; Moseley, P, T.; Norris, J. O. W.; Williams, D. E. J. Chem. Soc,
Faraday Trans. 1 1987, 83, 1323.

Flietner, B.; Eisele, 1. Thin Solid Films 1994, 250, 258.

Stoev, 1.; Kohl, D. Sens. Actuators B 1990, 2, 233.

Pink, H.', Treitinger, L. Vite, L. Japan. J. Appl. Phys. 1980, 19(3), 513.
Takahashi, Y.; Wada, Y. Yogyo-Kyokai-Shi 1987, 95(9), 864.

Yoshimura, N.; Sato, S.; Itoi, M.; Taguchi, H. Sozai Busseigalcu Zasshi 1990, 3,
47.

Chung, W.-Y.; Lee, D.-D.; Sohn, B.-K. Ihin Solid Films 1992, 221, 304. Chung,
W.-Y.; Lee, D.-D.; Choi, D.-H. Sens. ActuatorsB 1993, 13-14, 517.

31,

 

.\'a

Sea

Jar.
Ya:

Ma
198
5822'
.\l .

Kur

 

 

 

Aso
Fan
“in
Mai
C he
Per
Scc
See
See

Jo'r

loh

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

-- 27.

28.

29.

30.

31.

22

Nakajima, A. J. Mater. Sci. Lett. 1993, 12, 1778.

Seah, M. P. in Practical Surface Analysis, Vol. 1, Briggs, D.; Seah, MP, Eds;
John Wiley & Sons, Inc.: New York, 1990, p.311.

Janata, J. Principles of Chemical Sensors; Plenum Press: New York, 1989.
Yamazoe, N.; Kurokawa, Y.; Seiyama, T. Sens. Actuators 1983, 4, 283. Huck,
R.; Bottger, U.; Kohl, D.; Heiland, G. Sens. Actuators 1989, 17, 355.
Matsushima, S.; Maekawa, T.; Tamaki, J .; Miura, N.; Yamazoe, N. Chem. Lett.
1989, 845. Maekawa, T.; Tamaki, 1.; Miura, N.; Yamazoe, N.; Matsushima, S.
Sens. ActuatorsB 1992, 9, 63. Cheong, H.-W.; Choi, J .-J .; Kim, H. P.; Kim, J.-
M.; Churn, G.-S. Sens. ActuatorsB 1992, 9, 227.

Kung, H.H. Transition Metal Oxides: Surface Chemistry and Catalysis; Elsevier:
Amsterdam, 1989.

Aso, 1.; Nakao, M.; Yamazoe, N.; Seiyama, T. J. Catal. 1979, 57, 287.
Fattore, V.; Fuhrrnan, Z.A.; Manara, G.; Notari, B. J. Catal. 1975, 37, 215.
Windischmann, H.; Mark, P. .1. Electrochem. Soc. 1979, 126(4), 627.
Maier, J.; Gopel, W. J. Solid State Chem. 1988, 72, 293.

Chang, S, C. J. Vac. Sci. Technol. 1980, 17(1), 366.

Penn, DR. .1. Electron Spectrosc. Relat. Phenom. 1976, 9, 29.

Scoﬁeld, J. H. J. Electron Spectrosc. Relat. Phenom. 1979, 8, 129.

Seah, M. P.; Dench, W. A. Surf Interface Anal. 1979, 1., 2.

Seah, M. P. in Practical Surface Analysis, Vol. 1, Briggs, D.; Seah, M.P., Eds;
John Wiley & Sons, Inc: New York, 1990, p.231.

Hofmann, S. in Practical Surface Analysis, Vol. 1, Briggs, D.; Seah, M.P., Eds;
John Wiley & Sons, Inc: New York, 1990, p. 151.

Chara

2.1 Ab

FT}
films prep
titanium(l\

temperaturI

 

 

 

 

 

FTIR. The
5010110115 u
metaloxyg
for the PFC!
in all cast
SIrUCIure d
are higher
SolquOnS c

ire more h

2.2 Int

The
and 0Tgani

aIlse {mm

Chapter 2

Characterization of Titanium and Zirconium Valerate Sol-Gel Films

2.1 Abstract

FTIR and XPS have been used to characterize titanium and zirconium valerate thin
ﬁlms prepared using sol-gel techniques. Films were prepared by hydrolysis of
titanium(IV) isopropoxide or zirconium(IV) n-propoxide in excess valeric acid at room
temperature. Film solution chemistry, from precursors to cast ﬁlms, was followed with
F TIR. The structure and chemical composition of ﬁlms spin cast from fresh and day old
solutions were determined. Results of these studies suggest that all ﬁlms consist of a
metal-oxygen polymer backbone coordinated with bidentate valerate ligands. No evidence
for the presence of alkoxide ligands has been found. A small amount of water is adsorbed
in all cast ﬁlms. While solution aging experiments indicate that the zirconium ﬁlm
structure does not change with solution reaction time, carboxylate ligand concentrations
are higher in titanium ﬁlms made ﬁ'om aged solutions. Titanium ﬁlms made ﬁ'om aged
solutions contain slightly less than 1.5 valerate ligands per titanium atom. Zirconium ﬁlms

are more highly carboxylated with almost two valerate groups per metal center.

2.2 Introduction
The sol-gel process is an attractive approach for the synthesis of glasses, ceramics,
and organic-inorganic composite materials [1 - 4]. The advantages of sol-gel techniques

arise ﬁom the highpurity of the metal alkoxide precursors, the molecular homogeneity of

23

the intenn
ln additior
preparatiot
effort has
techniques
control of

tailor-made

 

 

Hyc
studied ant
strucrure is
transition n
chemistry
Coordinatic
ability to it
result in at
greater the
employed,
Oftlansitic

Ch
reactions I
dikergn es
Speties are

deCOUph'nQ

24

the intermediate sols, and the low processing temperatures necessary to prepare materials.
In addition, the rheological properties of many sols and gels make them suitable for ﬁlm
preparation using methods such as spin casting and dip coating [4-7 ]. Thus considerable
effort has been devoted to the development of new thin ﬁlm materials using sol-gel
techniques. Understanding the chemistry involved in ﬁlm synthesis will lead to better
control of the sol-gel process and, ultimately, to the ability to produce ﬁlms which are
tailor-made for particular applications.

Hydrolysis and condensation reactions of silicon alkoxides have been extensively
studied and, consequently, the relationship between sol-gel chemistry and polysilicate
structure is reasonably well understood [8 ~12 ]. In comparison, the sol-gel chemistry of
transition metal alkoxides has been less completely characterized. Transition metal sol-gel
chemistry differs ﬁ'om that of silicon because transition metals have a variety of
coordination states available and lower electronegativities than silicon [13 -16 ]. Both the
ability to increase coordination number and the electropositive nature of transition metals
result in alkoxide hydrolysis and condensation rates which are usually orders of magnitude
greater than that of silicon alkoxides. Therefore, unless some form of chemical control is
employed, hydroxide and/ or oxohydroxide species precipitate immediately upon hydrolysis
of transition metal alkoxides.

Chemical control of transition metal alkoxide hydrolysis and condensation
reactions is often achieved by adding complexing reagents such as organic acids or B-
diketones [17 -33 ]. When these ligands react with metal alkoxides, new molecular
species are produced with altered stnrctures and reactivities. Complexation promotes

decoupling between hydrolysis and condensation [34] and facilitates the preparation of

. A

umtomt I
hydrolysis
Thus. thel.
backbone L

ThI

and zircon

reported It

 

 

 

 

 

 

from mixtI.
However. I
ﬁlm ptOpe
Films prod
also hydro
those pteI
carbOtylic
acid and s
A
lm’olx'ed
Zirconjurr
there are
metal m,

Subsutum

25

uniform, transparent coatings [35 ]. Strong complexing ligands are more stable toward
hydrolysis than alkoxide ligands due to chelate, electronic, and steric hindrance effects.
Thus, the organic ligands become anchored to the transition metal oxo-polymeric
backbone and form organic-inorganic networks.

The role of acetic acid as a complexing agent in the sol-gel chemistry of titanium
and zirconium alkoxides has been examined in some detail [22-32]. We have previously
reported that optically transparent, porous ﬁlms can be produced at ambient temperatures
ﬁ'om mixtures of acetic acid, water, and titanium, zirconium, or hafnium alkoxides [3 2].
However, these ﬁlms tend to be of poor quality and are soluble in water. We found that
ﬁlm properties could be improved by complexing with longer chain carboxylic acids.
Films produced using valeric acid (n-C4H9COOH) are not only transparent and porous but
also hydrophobic and show greater integrity and better adhesion to glass substrates than
those prepared with acetic, propanoic, or butyric acid [33]. Our previous work with
carboxylic acid/transition metal alkoxide sol-gel chemistry also focused on the effect of
acid and water concentration on the structure of solutions, monoliths, and ﬁlms [3 2,33].

A major goal of this work is to gain a more detailed understanding of the chemistry
involved in the production of ﬁlms ﬁom titanium(IV) isopropoxide (Ti(OPr‘)4) or
zirconium(IV) n-propoxide (Zr(0Pr“)4) and valeric acid. It is generally acknowledged that
there are three principal reaction steps in the synthesis of sol-gel materials from transition
metal alkoxides and carboxylic acids [5,13,15,18,34,36 ]. The ﬁrst is a nucleophilic
substitution of an alkoxide ligand by a carboxylate ligand:

M(OR)4 + x R'COOI-I —9 won)” (000K), + x ROH (2.1)

 

 

Canh
imnmesci
gmmsr

respectn

aceﬁc
carbo
53¢:

Cher

hm
as
U:

ti

26

Alcohol is a by-product of this reaction. The second is the hydrolysis of one or more of
the remaining alkoxide groups:

aM-OR + H-OH -) aM-OH + ROH (2.2)
initiates condensation. In condensation, hydroxyl groups react with hydroxyl or alkoxide
groups on other metal centers producing M-O-M bonds and water or alcohol,
respectively:

EM-CLH + EM-OH —-) EM-O-ME + H20 (2.3a)

EM-OR + EM-OH —> EM-O-Ma + ROH (2.3b)

Esteriﬁcation reactions also play a ﬁmdamental role in the solution chemistry of
acetic acid modiﬁed alkoxides [27-22,36]. Esters can be formed by the reaction of a
' carboxylic acid with either an alcohol molecule or an alkoxide ligand producing water or a
hydroxyl ligand, respectively [29]. ‘ The importance of esteriﬁcation reactions in the
chemistry of valeric acid modiﬁed alkoxides is not known.

Carboxylate ligands can adopt a variety of coordination modes and as bidentate
bridging ligands, can act as network formers. The separation (Av) between the the
asymmetric (v..(COO)) and symmetric (v.(COO)) carboxyl stretch frequencies can be
used as an indicator of carboxylate coordination mode [37 ]. Separations much greater
than that of the ionic form of the ligand tend to indicate monodentate coordination,
otherwise bidentate coordination is implied. It has also been suggested that bidentate
bridging ligands typically exhibit Av values similar to ionic species, while bidentate

chelating ligands show smaller splittings [3 7]. This reasoning has been applied to

determine

 

coordinan
hand in th
bond angi’
nith coor
hands of c.
suggest rh
Tl
(Fink) s
and Zr((
mechani,
muﬂyes
mama

elucida

2.3

plOp;
Mdn
Chen
Vale

COr

27

determine coordination mode of acetate ligands in sols and gels [22,28,32]. Carboxylate
coordination mode can also be inferred from the frequency of the v..(COO) absorbance
band in the infrared spectrum of a material. Gﬁgor’ev found that the greater the 0C0
bond angle, the higher the v.(COO) frequency [38 ]. Since bond angle tends to decrease
with coordination mode: monodentate>bidentate bridging> bidentate chelating, v..(COO)
bands of considerably higher ﬁequency than in the spectrum of the ionic form of the ligand
suggest the presence of monodentate ligands.

The ﬁrst half of this paper is devoted to a Fourier transform infrared spectroscopy
(FTIR) study of the sol-gel solution chemistry speciﬁc to valeric acid modiﬁed Ti(OPri)4
and Zr(0Pr")4. Information is obtained which can be related to the rate, extent, and
mechanism of the sol-gel reactions in these systems. The second part of the paper
involves the characterization of titanium and zirconium valerate ﬁlms. The structure and
chemical composition of these ﬁlms, including the nature and concentration of ligands, are

elucidated using a combination of FTIR and X-ray photoelectron spectroscopy (XPS).

2.3 Experimental

Materials. Titanium(IV) isopropoxide, zirconium(IV) n-propoxide 70% in n-
propanol, isopropanol (99.5%, HPLC grade), and valeric acid (99+%) were obtained ﬁ'om
Aldrich Chemical Company and used without ﬁrrther puriﬁcation. All syntheses used
chemicals obtained from freshly opened bottles and distilled, deionized water. Sodium
valerate was prepared by mixing valeric acid with sodium carbonate (Aldrich Chemical

Company).

So
techniques
carboxylat
tials Moi
the ﬁlms
vortex mix
The reactn
titanium st
yellow

Tu
to titaniun‘
was the 5a:
oftitaniun:
0f both 55
ldentical “
SOlutiong a

Filr
or KB” 3
dessicanOr

complete a;

in iﬁphCate

 

 

28

Solution Preparation. The methods used to synthesize ﬁlms were based on
techniques developed in these laboratories for the preparation of transition metal
carboxylate thin ﬁlms [33]. Reactions were carried out at room temperature in capped
vials. Molar ratios of metal alkoxide : valeric acid : water of 1 :9: 1.5 were used to prepare
the ﬁlms. Valerie acid was added to the alkoxide, immediately followed by water. A
vortex mixer was used to vigorously stir solutions following the addition of each reactant.
The reaction between valeric acid and both metal alkoxides was exothermic. The resulting
titanium solutions were clear and colorless; zirconium solutions were clear and slightly
yellow.

Two sets of titanium solutions were prepared. In the ﬁrst, isopropanol was added
to titanium isopropoxide before the addition of valeric acid. The alcohol concentration
was the same as in the 70% Zr(OPr")4 solution (2.3 alcohol : l alkoxide). The second set
of titanium solutions was prepared without the addition of alcohol. Results of the analysis
of both sets of titanium solutions and the ﬁlms prepared from these solutions were
identical within experimental error. Results presented in this paper are for titanium
solutions and ﬁlms prepared without added isopropanol.

Film Preparation. Solutions were deposited onto the desired substrate (quartz
or KBr) and spun for ﬁve minutes. Films were airodried overnight and stored in a
dessicator. All ﬁlms prepared in this study were transparent, colorless, and exhibited
complete and uniform coverage of quartz substrates. Each solution and ﬁlm was prepared

in triplicate. In order to simplify discussion, titanium and zirconium ﬁlms cast immediately

 

after solut
solutions a
Fo
obtained
spectromet
software p
2cm" Fit
solution pl
reactions '
features u
Performer
holder cc
Omega C
)
analyzed
Preeisim
Using a :
kV’ and
each hit
energies
“lih a I

quan'llltg

29

after solution preparation are denoted TiOO and Zr00, respectively. Films prepared ﬁom
solutions aged for 24 h before casting are denoted Ti24 and Zr24.

Fourier Transform Infrared Spectroscopy (FTIR). F TIR spectra were
obtained using a Mattson Instruments Galaxy 3020 Fourier transform infrared
spectrometer. Data acquisition and processing were performed using an Enhanced First
software package. The mid-IR region (4000-400 cm'l) was examined with a resolution of
2 cm". Films were cast on KBr windows; solution spectra were obtained from a drop of
solution placed between two KBr windows. To identify artifacts which could be due to
reactions with KBr, several ﬁlms were cast on Ge and analyzed with FTIR. Spectral
features were the same for ﬁlms cast on both substrates. In situ heating experiments were
performed under nitrogen purge (99.5%, AGA Gas Co.) using a stainless-steel sample
holder equipped with cartridge heaters (Omega). The temperature was ﬁxed using an
Omega CN 1200 controller.

X-ray Fhotoelectron Spectroscopy (XPS). Films cast on quartz slides were
analyzed with a Perkin-Elmer surface science instrument equipped with a model 10-3 60
precision energy analyzer and an omnifocus small spot lens. All spectra were collected
using a non-monochromatized Mg anode (1253.6 eV) operated at a power of 300 W (15
kV and 20 mA) with an analyzer pass energy of 50 eV. Three- regions were scanned in
each ﬁlm: C Is, 0 Is and the most intense metal core band (Ti 2p or Zr 3d). Binding
energies were referenced to adventitious carbon (C 1s = 284.6 eV) and were measured
with a precision of i 0.1 eV. Empirically derived sensitivity factors [39 ] were used for

quantitative XPS calculations involving the Cls and 015. For the Ti 2p and Zr 3d,

Si

f0

bu
f0;

as;

30
sensitivity factors were derived ﬁ'om the analysis of TiOz and Zr02 (Aldrich Chemical
Company, 99.99+%). XPS peaks were ﬁtted with 20% Lorentzian-Gaussian mix Voigt
functions using a non-linear least squares curve ﬁtting program [40]. Each reported
result is based on the analysis of at least three ﬁlms. It must be noted that the use of XPS

to evaluate ﬁlm composition assumes that the ﬁlms are homogeneous.

2.4 Results and Discussion

FI'IR Spectra of Reactants. The FTIR spectra of reactants, as well as those of
isopropanol and n-propanol, are shown in Figure 2.1. The assignment of FTIR absorption
bands for valeric acid (Figure 2.1a) is straightforward [41 ,42 ]. Three methyl and
methylene C-H stretching absorbances are observed at 2962, 2935, and 2876 cm'l
superimposed on an O-H absorbance centered near 3000 cm". The strong absorbance
observed at 1711 cm'1 is due to a dimeric carboxylic COO asymmetric stretch. Bending
and stretching vibrations of C-O-H groups are observed at 1414, 1280, and 940 cm".
Skeletal vibrations are evident at 1382, 1109, and 752 cm".

In the FTIR spectrum of Ti(OPri)4 (Figure 2.1b) absorbance bands are assigned as
follows: the band at 1126 cm’1 is due to a combination of skeletal and C-0 stretches
[43 ]that observed at 1002 cm'1 is a v(C-O)Ti vibration [44], a skeletal vibration band is at
852 cm'1 [44 ], and a v(Ti-O) band appears at 621 cm'1 [43]. Due to steric hindrance by
bulky isopropoxide groups, Ti(OPrJ)4 is monomeric with a coordination number of
four,[30,45 ]. Isopropanol (Figure 2.1c) can be most easily distinguished from the

alkoxide by the sharp bands at 953 and 818 cm".

e: .525: <
9

Figure 2

31

 

 

 

 

 

 

 

 

 

l l l l 1
$
05
(e) -
:S'é
O m
"7 see
5‘“ V
3%
(d)
8 ON
G 00
.8 a
‘52 (C)
.D L
< § g
g s
(b)
E ‘1' c
E 8 o
" a
(a)
l J l L l
3600 2500 . 1600 900 400

Frequency (cm")

Figure 2.1 FTIR spectra of (a) valeric acid; (b) Ti(OPri)4; (c) isopropanol; (d)
Zr(OPr“)4 in n-propanol; and (d) n-propanol.

At

 

I
assigned ll

cm" is att
1083 and
tibrations.
Alcohol at

based upoi

 

consistent‘
indicate th:
The zircon
Propanol [
solvents. 1

Sol
alliOxide-a
5P€Ctra of
reat‘tiOn n
The ConCe
readily C01
concentrat
Solutjons a
that Coflde

32

Absorption bands in the spectrum of Zr(OPr“)4 in n-propanol (Figure 2.1d) are
assigned based upon the analysis of similar alkoxides [13,43]. The strong band at 1135
cm“1 is attributed to a combination of v(C-O)Zr and skeletal stretches, bands observed at
1082 and 1007 cm'1 are also associated with the alkoxide and may be due to v(C-O)Zr
vibrations, and those observed at 597, 549, 505, and 463 cm“ are attributed to vat-0).
Alcohol absorbances between 1100 and 900 cm'1 make it difﬁcult to determine structure
based upon v(C-O)Zr bands, however, multiple bands in the low frequency region are
consistent with an oligomeric structure. Ebulliometric studies by Bradley and Carter [46 ]
indicate that Zr(OPr")4 is oligomeric in n-propanol with a molecular complexity of 2.44.
The zirconium coordination sphere is also thought to be increased by solvation in n-
propanol [47]. This limits the extent of oligomerization compared to that in non-polar
solvents. N-propanol (Figure 2.1e) can be identiﬁed by a sharp band at 969 cm".

Solution Reactions. Solution reactions were investigated by preparing a series of
alkoxide-acid mixtures with increasing valeric acid concentrations. In addition to the
spectra of 1:1 (alkoxide : acid) mixtures of each alkoxide and valeric acid, spectra of
reaction mixtures with the lowest detectable level of unreacted valeric acid are shown.
The concentration of these solutions indicates the number of valerate ligands which can be
readily coordinated to each alkoxide. To the latter solution, water was added at a molar
concentration of 1.5 water per metal alkoxide. This is the concentration of water in ﬁlm
solutions and is sufﬁcient water to completely hydrolyze carboxylated alkoxides provided
that condensation also occurs. Analysis of these solutions avoids interference due to

excess valeric acid present in the 1:9 ﬁlm mixtures. It is recognized that the presence of

excess
about

acid cc

ﬁlms. 5
\‘alerat
Absorl

respeC‘

1.] ar
reSpec:

miX'tur

A

”—4
1.10
a4
Ur

formeC
bands .
‘0 bid.
VJ. C 0

lower 1

Cm") r
Inter-es
Split in

there -

33

excess valeric acid may alter the solution chemistry, however, ﬁtndamental information
about the interaction of solution species can be obtained by analyzing mixtures with lower
acid concentration.

In order to evaluate the coordination mode of valerate ligands in solutions and
ﬁlms, sodium valerate was prepared and analyzed with FTIR. The absorbances for sodium
valerate can be assigned by analogy to those‘of sodium acetate reported by Spinner [48 ].
Absorbances at 1562 and 1421 cm" (Av = 141 cm") are v..(COO) and v.(c00) bands,
respectively, of the ionic valerate groups.

Titanium Solutions. Figure 2.2 shows the FTIR spectra of Ti(0Pri)4 (Figure 2.2a),
1:1 and 1:1.5 mixtures of this alkoxide and valeric acid (Figures 2.2b and 2.2c,
respectively), and the latter mixture after the addition of water (Figure 2.2d). For a 1:1
mixture (Figure 2.2b), no free valeric acid is observed (1711 cm") and v.(c00) bands
(1575 and 1529 cm!) are present. This means that all of the valeric acid has reacted and
formed valerate ligands which are bonded to titanium. The frequency of the v..(COO)
bands are close to that of ionic valerate (1562 cm") and, therefore, are assumed to be due
to bidentate valerate ligands. Based upon its frequency, the stronger, higher energy
v..(COO) band (1575 cm") is probably due to bridging valerate ligands. Whether the
lower energy band is due to bridging or chelating ligands cannot be determined.

The appearance of absorption bands characteristic of isopropanol (953 and 818
cm") indicates that valerate ligands have displaced alkoxide ligands producing alcohol.
Interestingly, the alkoxide absorbance at 1002 cm'1 in titanium(IV) isopropoxide is now
split into two distinct bands at 1014 and 993 cm'l. This suggests that in the 1:1 solution,

there are two types of alkoxide ligands present, probably terminal and bridging alkoxides,

 

002:2..Cwﬁ <

34

 

 

 

 

 

1 1 1 1
0')
<1- 3
— e
E
((0
§
5
8 (C) "‘
5. .,,
.3 l\
o .‘C‘. E".
m cm
a a "‘8‘
(b)
g N
: §
91, a
0° C
(a)
1 1 i 1
1600 1200 800 400

Frequency (cm‘)

Figure 2.2 FTIR spectra of (a) Ti(OPri)4; (b) 1:1 mixture and (c)1;1.5 mixture of
Ti(OPr‘)4 and valeric acid; (d) 1:1.5:1.5 mixture of Ti(OPr’)4, valeric acid,
and water.

respe‘c
have 1
bully
but the
to six
monon

acid mt

valeric
readily
per titai
\leOC
band at
iS consi
Coordin;
obsen-e<
band at
this Pape
F
There is
hand at g

read”)? ht

35

respectively. The assignment is based upon other alkoxides: typically terminal alkoxides
have higher vibrational frequencies than bridging alkoxides [13]. The replacement of a
bulky isopropyl group with a valerate ligand not only increases the coordination to ﬁve,
but the decrease in steric hindrance also allows a further increase in coordination number
to six with the formation of alkoxide bridges. The transformation of Ti(OPri)4 from a
monomeric to an oligomeric structure was also reported by Sanchez er al. [49 ] for acetic
acid modiﬁcation of this alkoxide.

The spectrum of the 121.5 solution contains a band at 1711 cm'1 due to dimeric
valeric acid. Apparently, titanium centers do not accommodate a second valerate ligand as
readily as they accept the ﬁrst. On average, there are between 1 and 1.5 valerate ligands
per titanium atom in this solution. There are two new bands (1636 and 1596 cm") due to
v..(COO) absorbances. While the latter may be ascribed to bidentate bridging ligands, the
band at 1636 cm'1 probably indicates the formation of monodentate valerate ligands. This
is consistent with steric hindrance at the titanium center which would make bidentate
coordination more difficult. Alcohol and alkoxide bands are very similar to those
observed in the 1:1 solution spectmm. It should be noted that there is a small absorbance
band at 1734 cm". : The assignment of this band will be discussed in some detail later in
this paper.

Figure 2.2d shows the spectrum of the 1:1.5 mixture after the addition of water.
There is still a substantial terminal alkoxide band at 1014 cm'1 but the bridging alkoxide
band at 993 cm'1 is no longer evident. Apparently, bridging alkoxide groups are more

readily hydrolyzed than terminal alkoxides. Condensation is occurring as evidenced by the

36

presence of a rising background in the low energy portion of the spectrum that is
commonly attributed to phonon bands in a titanium oxygen network [50 ]. The increase in
valeric acid concentration indicates that valerate ligands are also displaced by the addition
of water. Sanchez et a]. [30] also observed that ﬁ'ee acid was formed when greater than 1
mole of water per metal alkoxide was added to acetate modiﬁed titanium alkoxides. The
high frequency v.(COO) bands present in the solution spectra before the addition of water
(1636 and 1596 cm'l) are no longer present. Either these carboxylates are preferentially
hydrolyzed or the decrease in steric hindrance which occurs when isopropyl groups are
hydrolyzed allows conversion from monodentate to bidentate coordination. Two
v.(COO) bands of about equal intensity are present and of the same frequency as that in
the 1:1 solution spectrum (Figure 2.2b). As stated previously, the higher energy band is
probably due to bidentate bridging ligands and the lower may be due to either bridging or
chelating ligands. ‘

It should be noted that small absorbance bands due to alkoxide ligands are present
in the freshly prepared titanium ﬁlm solution (1 Ti(OPri)4 : 9 valeric acid : 1.5 water)
spectra (not shown). The FTIR spectrum of a solution which has been aged for 24 h does
not contain bands characteristic of alkoxide ligands.

Zirconium Solutions. Figure 2.3a shows the FTIR spectrum of Zr(OPr")4 in n-
propanol. When a stoichiometric amount of valeric acid is added to the alkoxide (Figure
2.3b), it reacts completely (no band at 1711 cm'l) and forms valerate ligands (v.(COO)
bands at 1562 and 1535 cm’1 with a 1600 cm'1 shoulder). As in the titanium solutions, the

higher ﬁ'equency bands are probably due to bidentate bridging ligands and the lower may

37

be due to either bridging or chelating valerates. The relative intensity of alcohol bands
between 1200 and 800 cm'1 also increases with respect to the alkoxide bands in this region
of the spectrum. This is consistent with the displacement of alkoxide ligands and
formation of alcohol. Unlike titanium, none of the alkoxide (C-O)Zr features split and
their relative intensities do not appear to change with the addition of valerate ligands. This
implies that alkoxide ligand coordination and oligomerization is unchanged by
carboxylation. However, absorbance bands due to n-propanol also occur in this region of
the spectrum making it difﬁcult to say this unequivocally. The maximum intensity of the
bands between 700 and 500 cm'1 is shifted to a slightly higher frequency than in the
alkoxide spectrum. This is may be due to coordination with valerate ligands [13].

The spectnun of the 1:2 mixture is shown in Figure 2.3c. Small bands at 1711
cm'1 (unreacted valeric acid) and 1734 cm'1 (vide infra) are present. These bands are not
present in the spectrum of a 121.5 mixture (not shown). Nearly two valerate ligands are
readily coordinated to zirconium atoms in this solution. A broad band due to v..(COO)
occurs between 1589 and 1535 cm'1 indicating valerate ligands with bidentate
coordination only. In comparison to the 1:1 solution spectrum, the relative intensities of
the alkoxide bands are decreased and the alcohol bands increased. This is most clearly
seen by comparing the intensity of the alkoxide band at 1135 cm'1 with the alcohol bands
in this region.

Figure 2.3d shows the spectrum of this solution after the addition of water. Unlike
the titanium solutions, addition of water completely removes all alkoxide ligands from
zirconium. Alkoxide bands at 1139, 1082, and 1007 cm'1 are not evident. Indeed, this

portion of the spectrum resembles that of n—propanol (Figure 2.1e). The alkoxide bands

38

 

v
In
to 2
ln
I!)
v—
M
3 i:
(d) "
am
323."
~~
'11—!
V—
0 (cm:
g "
.D N
a ea
.9 .‘2
<

(b)

1135

(a)

 

 

W

l l l l

1600 1200 800 400
Frequency (cm'l)

 

Figure 2.3 FTIR spectra of (a) Zr(OPr")4 in n-propanol; (b) 1:1 mixture and (c) 1:2
mixture of Zr(OPr")4 and valeric acid; (ti) 12:15 mixture of Zr(OPr")4,
valeric acid, and water.

39

between 600 and 450 cm'1 are also absent. Absorption bands at 635 and 471 cm’1 are
probably due v(Zr—O) vibrations of oxy bridges formed upon condensation. As in titanium
solutions, some of the valerate groups are hydrolyzed as well, as seen by the increase in
the intensity of the band at 1711 cm". v..(COO) bands are present in the same range as
before the addition of water, indicating only bidentate coordination.

Comparison of Titanium and Zirconium Solution Reactions. The results of
the FTIR analysis of Ti(OPri)4-valeric acid and Zr(OPr")4-valeric acid solution spectra are
consistent with the three step reaction scheme presented in the introduction. However,
our study has revealed that there are differences in the reactivity of the two alkoxides. In
particular, Zr(OPr")4 readily accepts a higher level of carboxylate ligands than Ti(OPri)4
and Zr(OPr")4 alkoxide ligands are more rapidly hydrolyzed than those on Ti(OPri)4.
These differences can be related to both the chemical nature of the metal atoms and to
steric factors associated with their ligands.

As stated in the introduction, transition metal alkoxides are more reactive toward
nucleophilic substitutidn reactions than silicon alkoxides due to the more electropositive
nature of transition metals and the ability to expand their coordination sphere. For
analogous reasons, in general, zirconium alkoxides would be expected to be more reactive
than titanium alkoxides. Zirconium is less electronegative than titanium, and zirconium is
able to accommodate a larger coordination sphere than titanium. Coordination numbers
of up to 8 are possible for zirconium, whereas the optimum coordination number for

titanium is 6 [51 ].

40

The nature of the ligands on the metal center also aﬂ‘ects the reactivity of a
particular alkoxide. The reactivity of alkoxy ligands with respect to nucleophilic
substitution decreases in the order: tertiary>secondary>primary [15]. This is related to the
ability of the alkoxide to form a good leaving group. Therefore, isopropyl ligands would
be expected to be more reactive than n-propyl ligands in nucleophilic substitution
reactions. The reactivity of alkoxy ligands is also inﬂuenced by the stmcture of the
alkoxide compound. Monomeric alkoxides are more reactive than oligomeric alkoxides.
For these reasons, Ti(OPri)4 is expected to be very reactive in nucleophilic substitution
reactions. Indeed, 1:1 mixture of this alkoxide with valeric acid undergoes a strongly
exothermic reaction producing a dimeric complex in which titanium has a coordination
number of six.

Addition of a second valerate ligand apparently occurs more readily for the
zirconium valerate alkoxide complex than for its titanium counterpart. Not only is
zirconium larger than titanium, but it is coordinated with less bulky primary alkoxide
ligands. It is also possible that the oligomeric nature of Zr(OPr")a is unchanged by
reaction with valeric acid. Relatively labile solvate molecules, as well as alkoxide ligands,
could be displaced in the substitution reaction to accommodate bidentate valerate ligands
without a change in the oligomeric nature or coordination number. However, stable
dimeric titanium complexes would need to be broken in order to accommodate the
addition of a second bidentate valerate ligand. Evidence of the inability of titanium to

accommodate further carboxylate ligands is the presence of monodentate coordination in

41
the 111.5 Ti(OPrj)4 valeric acid solution. No monodentate valerate ligands are apparent in
the zirconium solution spectra.

Alkoxide ligands are evident in the titanium solution spectra after the addition of
water. Aging of the ﬁlm solutions for 24 h leads to the removal of the alkoxide ligands.
In contrast, addition of water to the zirconium solutions results in the complete and
immediate removal of all alkoxide ligands. This diﬁ'erence may be related the hydrolysis
and/or condensation rate of the materials. Since carboxylates are more electronegative
ligands than alkoxides there will be a higher partial charge on the highly carboxylated
zirconium center than on titanium. Consequently, it would be expected to be more
reactive in nucleophilic substitution reactions such as hydrolysis. The hydrolysis rate also
depends upon the concentration of water. Sufﬁcient water for complete hydrolysis is
present only if condensation occurs as well. Therefore, if condensation is slow, it will in
turn limit the hydrolysis rate.

The spectra of solutions containing unreacted valeric acid (Figures 2.2c, 2.2d,
2.3c, and 2.3d) show an absorbance band at 1734 cm". Bands of this frequency are
commonly attributed to the presence of esters. Esteriﬁcation is important in acetic acid
transition metal alkoxide sol-gel chemistry. However, higher order carboxylic acids (i.e.:
valeric acid) are known to form esters less readily than acetic acid [52]. Esteriﬁcation is
typically a slow reaction at room temperature and in the absence of an acid catalyst [52 ].
Though our solutions are warmed brieﬂy by the exothermic reaction of the alkoxides with
valeric acid, our solutions are not externally heated or allowed to stand for more than 24 h

before casting.

42

To see how readily esters could be prepared and to identify absorbance due to
esters, mixtures of valeric acid and the alcohols were prepared with concentrations similar
to that in the solutions used to cast ﬁlms (for purposes of this calculation only, it was
assumed that all alkoxide groups react to form alcohol in solution). The alcohol - valeric
acid mixtures were heated at 40°C for 1/2 h with and without concentrated sulfuric acid
catalyst. No changes were observed in the spectra of uncatalyzed mixtures. Spectra of
the catalyzed mixtures contained new bands which could be attributed to the presence of
esters. The most prominant new bands in the spectrum of the n-propanol mixture are at
1739 and 1180 cm'.1 and are typical of v(C=O) and v(C-O) vibrations, respectively, in
ester molecules [41]. Similarly, bands are observed at 1734 and 1183 cm'1 in the spectmm
of the acid catalyzed isopropanol-valeric acid mixture.

The presence of an absorption band at 1734 cm'1 does not necessarily indicate the
presence of esters in a solution. Very dilute solutions of valeric acid in both isopropanol
and n-propanol were prepared and analyzed with FTIR. In the spectra of these solutions,
a shoulder at 1734 cm'1 is evident on the C-0 stretch band of dimeric valeric acid (171]
cm"). The shoulder is not present in the spectrum of a dilute solution of valeric acid in
CCLr. Heating and aging the valeric acid/alcohol mixtures did not change the relative
intensity of the two bands. Based upon these results, the band at 1734 cm'1 in these dilute
solutions is not attributed to esters but to valeric acid which is involved in carbonyl-
alcohol H-bonds [4 1 ].

We believe that the bands at 1734' cm'1 in the solution spectra (Figures 2.2 and 2.3)

are also due to the presence H-bonded valeric acid-alcohol species. First, in these spectra

43

and in all ﬁlm solution spectra there was no evidence of corresponding v(C-O) bands.
Second, esters proved to be difﬁcult to produce from valeric acid and these alcohols in the
absence of an acid catalyst. Therefore, we do not believe that esteriﬁcation plays a
signiﬁcant role in the chemistry of valeric acid-alkoxide solutions used to prepare the
ﬁlms.

FTIR Characterization of Valerate Films. The inﬁ'ared spectra of titanium
(Ti00 and Ti24) and zirconium (Zr00) valerate ﬁlms are presented in Figure 2.4. All FTIR
absorbance bands in ﬁlm spectra can be assigned to valerate ligands, hydroxyl groups
(either hydroxyl ligands or water), or to the metal-oxygen backbone. No ﬁlm spectrum
contains bands that can be attributed to alkoxide groups, valeric acid, or alcohols.

Valerate Ligands. The three bands observed at 2959, 2933, and 2873 cm" are
identical in all titanium and zirconium ﬁlms. The relative intensity and position of these
peaks are similar to those observed for valeric acid (Figure 2.1a) and, therefore, are
attributed to valerate ligands. Less intense bands observed in all ﬁlm spectra at 1319,
1302, 1109, and 752 cm'I are also due to valerate ligands.

The most intense absorbances in the ﬁlm FTIR spectra are v(COO) bands
associated with valerate ligands (1650 - 1200 cm"). In the spectrum obtained for the T100
ﬁlm (Figure 4a), v..(COO) and v,(COO) frequencies of comparable intensity are observed
at 1530 cm'1 and 1448 cm", respectively (Av = 82 cm"). For the Ti24 ﬁlm (Figure 2.4b),
the v.(COO) band is less intense than the v.(COO) band. In addition, the v..(COO) band
in the spectrum of the T124 ﬁlm consists of two distinct bands at 1559 and 1528 cm'1 with
a shoulder at 1577 cm'1 (Av = 111, 80, and 129 cm", respectively) [53 ]. The frequency

range spanned by the v..(COO) bands and v.(COO) bands are unchanged by solution

 

 

 

 

 

 

 

 

l l l l l
(c)
a
a
e
O
a (b)
< .
- (a) l
l l l l l
3600 2500 1600 900 400

Frequency (cm")

Figure 2.4 FTIR spectra of (a) T100, (b) T124, and (c) Zr00 ﬁlms.

45

aging. Thus we believe that solution aging does not necessarily lead to new valerate
coordination conﬁgurations but rather to a redistribution of ligands amongst
conﬁgurations. Comparing these Av values with that in sodium valerate (141 cm")
suggests that titanium ﬁlms contain only bidentate ligands. While the presence of
chelating ligands cannot be ruled out, we believe it is possible that our ﬁlms contain only
bidentate bridging ligands which vary slightly in bond angle giving rise to several distinct
maxima. Doeuﬁ‘ et a]. [3 6] reported three distinct v..(COO) bands present in a crystalline
acetate alkoxy titanium complex known to contain only bridging acetates of varying bond
angle.

In the spectrum obtained for the Zr00 ﬁlm (Figure 2.4c), the v..(COO) band at
1554 cm'1 and v.(COO) band at 1447 cm'1 are of similar intensity. Solution aging has little
effect on the carboxylate stretch region of the FTIR spectra of the zirconium ﬁlms. The
spectrum obtained for the Zr24 ﬁlm shows v..(COO) and v,(COO) bands at 1559 and
1445 cm", respectively. Av values are 107 and 114 cm'1 and indicate that the zirconium
ﬁlms contain only bidentate valerate ligands. Paul et a1. [54] report values of 1540 and
1420 cm'1 for the asymmetric and symmetric COO stretches of zirconium(IV) propionate
(Av = 120 cm") and ascribed these bands to the presence of bidentate ligands.

Spectra of solutions used to prepare these ﬁlms (not shown) were also collected
but, due to the presence of excess valeric acid and alcohol, they are of limited value in
providing information about the polymer structure in solution. However, asymmetric

carboxylate stretches can be readily seen in the solution spectra. Comparing this region of

46

the solution spectra with that in the ﬁlm spectra, suggests that carboxylate coordination
modes are similar before and after spin casting.

Hydroxyl Groups. Broad bands due to H-bonded hydroxyl groups are observed
near 3400 cm'1 in the spectra of all ﬁlms. The intensity of the hydroxyl stretch observed in
the zirconium ﬁlm spectra is essentially unchanged by solution aging and greater than that
observed in the titanium ﬁlm spectra. In zirconium ﬁlms, the sharp band observed at 3639
cm'1 is typical of "free" or non-H-bonded hydroxyl groups [41]. Water has a characteristic
absorbance near 1600 cm'1 [41]. Unfortunately, the strong carboxylate absorbances in this
region make it impossible to use this band to distinguish between hydroxyl ligands and
water.

In order to determine whether these absorbances are associated with water or
hydroxyl ligands, changes in the hydroxyl stretch absorbances were observed while heating
titanium and zirconium ﬁlms in dry nitrogen. Heating titanium ﬁlms at 50°C essentially
removed the hydroxyl stretch band centered at 3400 cm". This band is partially restored
by exposure to ambient air after cooling. Hydroxyl groups should not be removed at such
low temperatures and, therefore, we ascribe the hydroxyl bands observed in the FTIR
speCtra of titanium ﬁlms to sorbed water.

Figure 2.5 shows the effect of heating on the hydroxyl region of the Zr00 ﬁlm
spectrum. The intensity of the hydroxyl absorbances decreased upon heating slowly to 50
°C (Figure 2.5b); however, the hydroxyl bands are evident unless the ﬁlm is heated to 100
°C (Figure 2.5c). Additionally, when the ﬁlm is heated rapidly to 75°C, the broad

absorbance disappears before the loss of the sharp band (not shown). A small band at

47

3639 cm'l is restored upon cooling (Figure 2.5d) and exposure to ambient air for 30
minutes results in the partial restoration of the broad hydroxyl absorbance (Figure 2. Se).
Continued exposure to ambient air did not cause any additional increase in the intensity of
this band. Again, the low temperature required for removal of the hydroxyl absorption
bands indicates that water, not hydroxyl ligands are the primary source of the bands.

We tentatively ascribe the non-H-bonded hydroxyl absorbance to water
coordinated to Zr centers. We attribute H-bonded hydroxyl absorbances to sorbed water.
During slow heating an equilibrium is maintained between the two types of water. Rapid
heating results indicate that non-H-bonded water is more difﬁcult to remove, perhaps due
to a combination of steric hindrance and coordination with Zr. It also may be energetically
more stable since the sharp absorbance band returns ﬁrst upon cooling. Partial restoration
of the broad absorbance upon exposure to ambient air is further evidence that this band is
due to sorbed water.

Water may be present due to condensation reactions occurring during spin casting
or due to trapping of water present in precursor solutions. Water is probably also present
due to exposure to ambient moisture during spincasting and drying. Zirconium can
accommodate a larger coordination sphere than titanium. As a consequence, hydration of
zirconium may occur leading to ﬁlms with a higher water content than the titanium ﬁlms.
Factors such as porosity may also contribute to the extent of water sorption in the ﬁlms.

Backbone. The low frequency region of the T100 and Ti24 spectra (Figures 2.4a
and 2.4b, respectively) contains a broad absorbance between 980 and 530 cm'1 that is
commonly attributed to an envelope of phonon bands in the titanium oxide network [50].

Similar features have been reported by Doeuﬁ‘ et al. [22] in the spectra of titanium acetate

48

 

 

(d)

 

\e
l

Absorbance
@

(b)

 

 

 

 

(a)

 

 

 

1 1 1 1
4000 3500 3000 2500
Frequency (cm'l)

Figure 2.5 FTIR spectra of hydroxyl region: in situ heating of Zr00 ﬁlm (a) before
heating; (b) heated to 50°C; (c) heated to 100°C; (d) cooled to room
temperature in dry nitrogen; (e) exposed to ambient air for 30 minutes.

49

gels and were also present in our solution spectra (Figure 2.2d). Though the shape of this
broad absorbance is essentially unchanged with increased precursor solution age, its
relative intensity is decreased with respect to carboxylate absorbances. The low frequency
region of the FTIR spectra of the zirconium ﬁlms (Figure 2.4c) show two major bands at
654 and 462 cm'1 which we attribute to backbone v(Zr-O). These bands are similar to
those in the solution spectra after the addition of water (Figure 2.3d). Atik and Aegerter
[23] ascribed bands at 666 and 363 cm" to Zr-O-Zr stretches and Sanchez and In [55]
attributed a broad band between 600 and 300 cm'1 to these vibrations in a zirconium oxide
based network.

XPS Characterization of Valerate Films. The XPS metal core peak binding
energies measured for titanium (Ti 2pm = 458.7 eV) and zirconium (Zr 3d5rz = 182.3 eV)
are independent of solution aging and typical of values measured for tetravalent forms of
the elements [56 ]. The XPS C ls spectra measured for the ﬁlms consist of peaks at 284.6
eV (reference peak) and 288.5 eV due to alkyl and carboxylate carbon, respectively. No
features due to alkoxide carbon (~ 286 eV) are observed in the spectra which is consistent
with the FTIR analysis of the ﬁlms. The XPS O ls spectra measured for the ﬁlms consist
of peaks at 530.1 eV and 531.8 eV that are attributed to backbone and ligand
(carboxylate, hydroxyl, or water) oxygen, respectively. Figure 2.6 shows the XPS O ls
spectra of T100, T124, and Zr00 ﬁlms. The relative intensity of the ligand oxygen peak
increases with respect to that of the backbone oxygen peak with titanium solution aging.
For the zirconium ﬁlms, the relative intensities of the backbone and ligand oxygen peaks

are independent of the age of solution used to cast the ﬁlm.

50

Atomic ratios derived from XPS intensity ratios are presented in Table 2.1. When
combined with interpretation based upon the FTIR analysis of the ﬁlms, these values
provide a quantitative assessment of the chemical species present in the ﬁlms. FTIR
analysis indicates that the ﬁlms consist primarily of a metal oxygen network with valerate
ligands coordinated to metal centers. Water is also present in the ﬁlms. The
concentration of valerate ligands is indicated by carboxyl carbon/metal atomic ratios
determined by the XPS analysis. Since each carboxylate carbon is associated with two
ligand oxygens, a ﬁlm containing only carboxylate ligands is expected to have a ligand
oxygen/carboxyl carbon ratio close to two. Thus, ligand oxygen/carboxylate carbon
values greater than two indicate the presence of water or hydroxyl groups in the ﬁlms.

It is important to note that if water is present in valeric acid when it is mixed with
the alkoxide, levels of carboxylation are lower. This is in agreement with the results of
Doeuﬁ‘ et al. [27] who found that adding titanium alkoxide to a mixture of water and
acetic acid led to the formation of a gel with a low level of carboxylation. In order to
achieve the level of carboxylation indicated by our results, the use of dry reactants is
imperative.

Titanium Films. The concentration of valerate ligands in the ﬁlms [57] increases
from the T100 ﬁlm (1.16 i 0.10) to the T124 ﬁlm (1.42 i 0.05) indicating that these
ligands add to titanium during solution aging. Accompanying the increase in valerate
ligand content is a corresponding decrease in the backbone oxygen concentration with
solution aging. Backbone oxygen to titanium atomic ratios decrease from 1.40 i 0.01 for

T100 ﬁlms to 1.12 i 0.03 for Ti24 ﬁlms. The change in the relative concentrations of

 

 

 

 

.3:
5 (C)
E
“C
d)
.21
E .
E (b)
Z
(a)
l l J
536 532 528
Binding Energy (eV)

Figure 2.6 Oxygen ls photoelectron spectra measured for (a) T100, (b) T124, and (c)
Zr00 ﬁlms.

52

Table 2.1 Atomic Ratios of Carbon and Oxygen Species Calculated from XPS

Intensity Ratios.
Atomic Ratios“
kb 11 11 Wm LigantLQngen Liam
F_i_ Metal Metal Metal Carboxyl Carbon
T100 1.4 1.2 2.2 1.9
T124 1.1 1.4 2.6 1.8
ZrOO 1.0 1.8 2.5 2.1
Zr24 1.0 1.9 2.5 2.0

* Relative standard deviations for atomic ratios are 5 1- 10%.

53

these two species is also apparent by comparing the FTIR spectra of the T100 and T124
ﬁlms (Figure 2.4).

These changes can be understood based upon solution compositions. Recall that
the FTIR spectra of freshly prepared titanium solutions contain absorbances due to
alkoxide ligands. This indicates that hydrolysis and condensation reactions have not gone
to completion in these solutions. We believe that the removal of the solvent that occurs
during spincasting and drying facilitates completion ‘of these processes and the formation
of backbone oxygen linkages. Because valerates are strong ligands, their presence limits
the extent of condensation. In contrast, alkoxide groups are absent in the aged ﬁlm
solutions. Film solutions contain a large excess of valeric acid and during solution aging,
valerate ligands may displace remaining alkoxide and/or hydroxyl ligands. The extent of
carboxylation is increased while the extent of condensation has been accordingly
decreased.

A slow rate of carboxylation for titanium valerate alkoxide complexes is consistent
with their slow rate of hydrolysis since both are nucleophilic substitution reactions.
Though isopropoxide ligands form good leaving groups, their bulkiness may hinder the
approach of .nucleophiles to titanium. In addition, solution studies indicated that a second
valerate ligand was not readily added to titanium since it required the breaking of relatively
stable dimeric units. Bridging alkoxide ligands were preferentially hydrolyzed and
subsequent condensation reactions lead to completely new oligomeric structures. These

species may be more easily carboxylated than the titanium valerate alkoxide dimers.

54

The ligand oxygen/carboxyl carbon ratio for titanium ﬁlms is slightly less than two
indicating that little water is present in these ﬁlms. Sorbed water evident in the FTIR
spectra is pumped away in the vacuum required for XPS analysis. In situ F TIR heating
experiments indicated that water is not strongly bound to the titanium ﬁlms.

Zirconium Films. The compositions of zirconium ﬁlms prepared from ﬂesh and
day old solutions are nearly identical. This is consistent with FTIR results which indicate
that aging has little effect on the structure of the ﬁlms. The number of valerate ligands per
zirconium is the same within experimental error in zirconium ﬁlms prepared ﬁ'om fresh
(1.80 i 0.12) and aged (1.90 i 0.15) solutions and all zirconium ﬁlms have a backbone
oxygen/metal atomic ratio of approximately one.

The ligand oxygen/carboxyl carbon ratio for zirconium ﬁlms is slightly greater than
that for titanium ﬁlms. However, its value also indicates that little water is present in these
ﬁlms. Again, sorbed water observed with FT IR is removed by the vacuum prior to XPS
analysis. A small amount may remain in the zirconium ﬁlms. This is consistent with the
higher concentration of water in the zirconium ﬁlms as indicated by the FTIR analysis and
the higher temperature required for its removal. The presence of a small concentration of
hydroxyl ligands can not be ruled out.

Additional Comments. It should be noted that ﬁlm compositions calculated using
XPS intensity ratios have the appropriate number of ligands per metal center. Titanium
and zirconium have +4 oxidation states, backbone oxygen has a -2 oxidation state, and a
carboxylate group can be assigned a charge of -1. Using this accounting system, the anion

charge calculated for the T100 ﬁlm (1.4 backbone oxygens and 1.2 carboxylate species per

55

metal center) is -4.0. Similar calculations for the T124, Zr00, and Zr24 ﬁlms lead to
charges of -3.7, -3.9, and -3.9, respectively. Within experimental error, these results
indicate the formation of neutral molecules and ﬁrrther demonstrate the use of XPS to
determine ﬁlm composition.

The levels of carboxylation determined for the ﬁlms are consistent with that
observed in the solution studies. According to the solution studies, between 1 and 1.5
valerate ligands displace alkoxide ligands on T1(OPrj)a. The T100 and T124 ﬁlms contain
1.2 and 1.4 valerate ligands per metal center. Zr(OPr")4 were shown to react readily with
just under two valeric acid molecules which is the level of carboxylation measured in the

zirconium ﬁlms.

2.5

10.

ll.

12.

13.

14.
15.

16,

56

References

Ulrich, D.R. J. Non-Cryst. Solids 1988, 100, 174.

Better Ceramics lhrough Chemistry V, Hampden-Smith, M.J.; Klemperer, W.G.;
Brinker, C.J.; Eds; Mater. Res. Soc. Symp. Proc. 271: Pittsburgh, PA, 1992.

Ultrastructure Processing of Advanced Ceramics, Mackenzie, J .D.; Ulrich, D.R.,
Eds; North Holland: New York, 1988.

Sol-Gel Technology for Thin F ilms, Fibers, Preforms, Electronics, and Specialty
Shapes, Klein, L.C., Ed.; Noyes Publications: Park Ridge, New Jersey, 1988.

Brinker, C.I.; Scherrer, G. Sol-Gel Science, the Physics and Chemistry of Sol-Gel
Processing, Academic Press: San Diego, 1990.

Scriven, L.E. in Better Ceramics Through Chemistry 111, Brinker, C.J.; Clark,
D.E.; Ulrich, D.R., Eds, Mat. Res. Soc., Pittsburgh, PA, 1988, 717.

iBrinker, 0.1.; Hurd, A.J.; Schunk, 13.11.; Frye, G.C.; Ashley, cs. J.Non-

Cryst.Solids 1992, 147&148, 424.

Hench, L.L.; West, J.K. Chem. Rev. 1990, 90, 33.

Schaeﬁ‘er, D.W. Science 1989, 243, 1023.

Klemperer, W.G.; Ramamurthi, SD. in Better Ceramics 7hrough Chemistry 111;
Brinker, 0].; Clark, D.E.; Ulrich, D.R.; Eds; Materials Research Society,
Pittsburgh, 1988, l.

Klemperer, W.G.; Mainz, V.V.; Millar, D.M. in Better Ceramics Through
Chemistry II; Brinker, C.J.; Clark, D.E.; Ulrich, D.R.; Eds; Materials Research
Society, Pittsburgh, 1986, 3.

Klemperer, W.G.; Mainz, V.V.; Millar, D.M, ibid, 15.

Bradley, D.C., Mehrotra, R.C.; Gaur, D.P. Metal Alkoxides; Academic Press:
London, 1978.

Mehrotra, R.C. J. Non-Cryst. Solids 1988, 100, 1.
Livage, 1.; Henry, M.; Sanchez, C. Prog. Solid St. Chem. 1988, 18, 259.

Hubert-Pfalzgraf, L.G. NewJ. Chem. 1987, 11, 663.

l7.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30 .‘

31.

32.

57

Mehrotra, R.C.; Gaur, D.P.; Bohra, R. Metal ,B-diketonates and Allied
Derivatives, Academic Press, London, 1978.

Mehrotra, R.C.; Bohra, R. Metal Carboxylates, Academic Press, London, 1983.

Sanchez, C.; Livage, J.; Henry, M.; Babonneau, F. J. Non-Cryst. Solids 1988, 100,
65.

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

Doeuﬁ‘, S.; Henry, M.; Sanchez, C.; Livage, J. J. Non-Cryst. Solids. 1987, 89, 84.
Doeuﬁ‘, S.; Henry, M.; Sanchez, C.;L1vage, J. J. Non-Cryst. Solids 1987, 89, 206.
Atik, M.; Aegerter, MA. J. Non-Cryst. Solids. 1992, 147&148, 813.

Chaumont, D.; Craievich, A.; Zarzycki J. Non-Cryst. Solids 1992, 147&148, 127.
Yoldas, BE. J. Mater. Sci. 1986, 21, 1080.

Leaustic, A.; Riman, R.E. J. Non-Cryst. Solids. 1991, I35, 259.

Assink, R.A.; Schwartz, R.W. Chem. Mater. 1993, 5, 511.

Laaziz, I.; Larbot, A.; Julbe, A.; Guizard, C.; Cot, L. J. Solid St. Chem. 1992, 98,
393.

Doeuﬁ‘, 8; Henry, M.; Sanchez, C. Mat. Res. Bull. 1990, 25, 1519.

Sanchez, C.; Babonneau, F.; Doeuﬁ‘, S.; Leaustic, A. Ultrastructure Processing of
Advanced Ceramics, Mackenzie, J .D.; Ulrich, D.R., Eds; North Holland: New
York, 1988; 77.

Larbot, A.; Alary, J.A.; Guizard, C .; Cot, L.; Gillot J. Non-Cryst. Solids 1988,
104, 161.

Gagliardi, C.D.; Berglund, K.A. in Processing Science of Advanced Ceramics,
Aksay, I.A.; McVay, G.L.; Ulrich, D.R., Eds; Mater. Res. Soc. Proc. 155:
Pittsburgh, PA, 1986; 127.

33.
34.
35.

36.
37.
38.

39.
40.
41.
42.

43.

45.

46.
47.

48.

58

Gagliardi, C.D.; Dunuwila, D.; Berglund, K.A. in Better Ceramics Through
Chemistry IV, Zelinsky, B.J.J.; Brinker, C.J.; Clark, BE; Ulrich, D.R., Eds;
Mater. Res. Soc. Proc. 180: Pittsburgh, PA, 1990; 801.

Sanchez, C.; Livage, J. NewJ. Chem. 1990, I4, 513.

Schroeder, H. in Physics of Ihin Films, Hass, G., Ed, Vol. 5, Academic Press,
New York, 1969, 87.

Doeuﬁ‘, S.; Dromzee, Y.; Taulelle, J.; Sanchez, C. Inorg. Chem. 1989, 28, 4439.
Deacon, G.B.; Phillips, R.J. Coord Chem. Rev. 1980, 33, 227.
Grigor’ev, A.I. Russ. J. Inorg. Chem. 1963, 8(4), 409.

Wagner, C.D.; Davis, L.E.; Zeller, M.V.; Taylor, J.A.; Gale, L.H. Surf Interface

Anal. 1981, 3, 211.

Software provided by Dr. Andrew Proctor, University of Pittsburgh, Pittsburgh,
PA.

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 8: Sons, Inc.: New York, 1974.

Barraclough, C.G.; Bradley, D.C.; Lewis, J.; Thomas, I.M. J. Chem. Soc. 1961,
2601.

Zeitler, V.A.; Brown, CA. J. Phys. Chem. 1957, 61, 1174.

Babonneau, F.; Doeuﬁ‘, S.; Leaustic, A.; Sanchez, C.; Cartier, C.; Verdaguer, M.
Inorg. Chem. 1988, 27, 3166.

Bradley, D.C.; Carter, D.G. Can. J. Chem. 1961, 39, 1434.
Livage, J .; Sanchez, C. J. Non-Cryst. Solids 1992, I45, 11.

Spinner, E. J. Chem. Soc. 1964, 4217.

49.

50.

51.

52.

53.

54.
55'.
56.

57.

59

Sanchez, C.; Toledano, P.; Ribot, F. in Better Ceramics Through Chemistry IV,
Zelinsky, B.J.J.; Brinker, C.J.; Clark, D.E.; Ulrich, D.R., Eds; Mater. Res. Soc.
Proc. 180: Pittsburgh, PA, 1990; 47.

McDevitt, N.T.; Baun, W.L. Spectrochimica Acta 1964, 20, 799.

Greenwood, N.N.; Earnshaw, A. Chemistry of the Elements; Pergamon Press:
New York, 1984.

Morrison, R.T.; Boyd, R.N. Organic Chemistry, 3rd Ed; Allyn and Bacon, Inc:
Boston, 1974.

In all ﬁlm spectra, the v.(COO) band is as broad as the v..(COO) band, however,
only one well-resolved maximum is observed. All Av values are reported with
respect to this maximum which is on the high energy side of the peak. Therefore,
separations of the outer peaks may actually be greater that the values reported.

Paul, R.C.; Baidya, O.B.; Kumar, RC; Kapoor, R Aust. J Chem. 1976, 29,
1605.

“ Sanchez, c.; In, M. J Non-Cryst. Solids 1992, 147&14s, 1.

Handbook of X-ray Photoelectron Spectroscopy; Wagner, C.D.; Riggs, W.M.;
Davis, L.E.; Moulder, J .F .; Muilenburg, G.E., Eds; Perkin Elmer, 1979.

In this discussion, values are reported to two decimal places and the standard
deviation of the values measured for three different ﬁlms is given to indicate the
actual precision of the. measurements.

Chapter 3

Synthesis and Characterization of Tin Valerate and Tin Oxide Thin
Films

3.1 Abstract

Tin valerate ﬁlms have been prepared by spincasting solutions produced from the
hydrolysis of tin(II) methoxide or tin(IV) isopropoxide in alcohol and valeric acid. Tin
oxide ﬁlms were made from tin valerate ﬁlms using two methods: calcination at 400°C in
air and by room temperature hydrogen peroxide treatment. All tin valerate and tin oxide
ﬁlms are transparent and exhibit uniform adhesion to quartz substrates. FTIR and XPS
analyses indicate that the structure and composition of valerate ﬁlms prepared from either
alkoxide are similar despite differences in precursor oxidation state and concentrations of
acid and alcohol used for the syntheses. Tinivalerate ﬁlms are comprised of an oxygen-tin
polymer backbone coordinated with valerate ligands. Both monodentate and bidentate
carboxylates are present, the former apparently stabilized by hydrogen bonding with
sorbed water. Quantitative XPS analysis indicates that valerate/tin and hydroxyl/tin ratios
are 0.26 and 0.55, respectively. This level of ﬁlm carboxylation is consistent with the
extent of carboxylate formation in tin(IV) isopropoxide/valeric acid solutions. Hydrogen
peroxide treatment of tin valerate ﬁlms removes organic ligands and increases the level of
backbone oxygen. Hydrogen peroxide treatment or calcination results in the formation of

amorphous tin (IV) oxide ﬁlms.

60

61

3.2 Introduction

Tin oxide is a commonly used material in conductimetric sensor [1 ] and
heterogeneous catalyst applications [2 ]. Properties such as transparency and
semiconductivity also make tin oxide a ﬁmdarnental component of solar cells [3 ],
transparent electrodes [4 ], thin ﬁlm resistors [5 ], and far IR detectors [6 ]. Thin ﬁlms of
tin oxide are conventionally prepared by reactive sputtering [7], spray pyrolysis [8 ], or
CVD [9 ]. While these methods produce high quality oxide ﬁlms, they do have limitations
that can be addressed by the development of sol-gel techniques. First, sol-gel derived
materials can be prepared with high surface areas and controlled porosity, properties
which are important in catalyst and sensor applications [10 ,11 ]. Second, electronic and
optical properties can be tailored by mixing tin oxide with other metal oxides or by doping
with noble metals. Sol-gel synthesis of these materials would be advantageous since
molecularly mixed oxides can be produced from metal alkoxide precursors [12 ].

Sol-gel synthesis of materials from silicon and transition metal alkoxides has
been extensively investigated [10,13 ]. Under the proper conditions, addition of water to
alkoxides causes hydrolysis and condensation reactions. Transition metal alkoxides are
highly reactive and tend to form oxyhydroxide precipitates when exposed to moisture.
Limiting alkoxide reactivity inhibits precipitate formation and allows the preparation of
polymeric gels and thin ﬁlms [13]. Control is achieved through complexation of the metal
alkoxides with strong ligands such as carboxylates and B-diketonates. Valerie acid

(pentanoic acid) modiﬁed transition metal alkoxides produce especially stable, uniform,

62

and transparent ﬁlms which exhibit good adhesion to glass and quartz substrates [14 , 15 ].
Previously we have shown that titanium and zirconium valerate ﬁlms are comprised of a
metal-oxygen network coordinated with valerate ligands [15].

Despite the importance of tin oxide thin ﬁlms in a number of technologies, the
study of tin alkoxide sol-gel routes’to these materials has been relatively limited [16 ,17 ].
Like transition metal alkoxides, tin alkoxides are known to react rapidly with water [16].
It is reasonable to suppose that methods demonstrated to produce high quality ﬁlms from
transition metal alkoxides could be successfully applied to the synthesis of ﬁlms from tin
alkoxides. In this work uniform, transparent tin valerate ﬁlms have been prepared ﬁ'om
tin(IV) isopropoxide (Sn(OPri)4) or tin(II) methoxide (Sn(OMe)2) and valeric acid using
methods found to produce high quality ﬁlms ﬁ'om titanium and zirconium alkoxides
[14,15]. Two diﬁ‘erent methods: room temperature hydrogen peroxide treatment and
calcination in air at 400°C were used to produce oxide ﬁlms from tin valerate ﬁlms.

All ﬁlms in this study were characterized using a combination of bulk and surface
spectroscopies in a manner similar to that used to study titanium and zirconium valerate
ﬁlms [15]." Structural information was obtained by Fourier transform infrared
spectroscopy (F TIR).- . X-ray photoelectron spectroscopy (XPS) was used to provide
information about the chemical species present in ﬁlm surfaces and their relative
concentrations. The oxidation state of tin in the ﬁlms was examined using valence band

XPS.

63

3.3 Experimental

Materials. Tin(II) methoxide (Johnson Matthey) and tin(IV) isopropoxide 10%
w/v in isopropanol (Chemat Technologies), absolute methanol (Photrex reagent, J .T.
Baker), hydrogen peroxide, 30% (Baker analyzed reagent, J.T. Baker), and valeric acid
(99+%, Aldrich Chemical Company) were used without further puriﬁcation. Distilled,
deionized water was used for all syntheses. Tin(IV) oxide (99.99S+%, Aldrich Chemical
Company) was pressed into pellets and calcined in dry air (so cm3/min, medical grade,
AGA Gas Co.) at 425°C for 4 h before XPS analysis. i

Film Preparation. The methods used to synthesize ﬁlms were based on
techniques developed for the preparation of transition metal carboxylate thin ﬁlms [14].
Modiﬁcations were necessary to obtain clear, particle free solutions and transparent,
uniform ﬁlms using the tin alkoxide precursors. Molar ratios of 15 methanol : 9 valeric
acid : 1.5 water : 1 metal alkoxide were used to prepare ﬁlms from Sn(OMe)2. Films
synthesized from Sn(OPri)4 were prepared using mixtures of 40 isopropanol : 20 valeric
acid : 1.5 water : 1 metal alkoxide. Reactions were carried out at room temperature in
capped vials. Sn(OMe)j was dispersed in methanol using sonication. Valerie acid was
added to the alkoxide-alcohol mixtures followed by water. A vortex mixer was used to
vigorously stir solutions following the addition of each reactant.

Addition of valeric acid to Sn(OMe)2-methanol suspensions caused brief but
vigorous foaming and produced a faintly yellow, opalescent, particle free solution. No
further changes were observed upon addition of water to these mixtures. No reaction was

visually apparent when mixing Sn(OPr‘)a and isopropanol with valeric acid; however,

54

addition of water resulted in the formation of a white precipitate which dissolved
immediately upon mixing. Though initially these solutions had an opalescent appearance
(perhaps due to unreacted water dispersed in the solutions), within ﬁve minutes the
solutions were transparent and colorless.

Freshly prepared solutions were dispersed on the desired substrate (quartz or KBr
crystal) and spun for ﬁve minutes All ﬁlms prepared in this study were transparent,
colorless, and exhibited complete and uniform adhesion to quartz substrates. F ilrns were
air-dried overnight and stored in a desiccator. Films studied without further treatment are
referred to as tin valerate ﬁlms (SnV). Calcined ﬁlms (SnC) were prepared by heating
SnV ﬁlms in a tube furnace in dry air (so cm3/min, medical grade, AGA Gas Co.) at
400°C for 4 h. Peroxide treated ﬁlms (SnP) were obtained by immersing SnV ﬁlms coated
on quartz or Si crystals in a 30% hydrogen peroxide solution for 4h followed by drying at
110°C in air for 30 min. The FTIR spectrum of Si crystals was found to be unchanged by
hydrogen peroxide treatment.

To clarify the analysis of the hydrogen peroxide treated ﬁlms, tin valerate gels and
peroxide treated gels were prepared. Tin valerate gels were synthesized from a 1.1:] .5
mixture of Sn(OPri)4 (as received in isopropanol), valeric acid, and water. Gels were
allowed to dry at room temperature and were spread on KBr crystals for analysis.
Peroxide treated gels were obtained using the same procedures described for the
preparation of SnP ﬁlms.

Fourier Transform Infrared Spectroscopy (FTIR). FTIR spectra were

obtained using a Mattson Instruments Galaxy 3020 Fourier transform infrared

65

spectrometer. Data acquisition and processing were performed using an Enhanced First
software package. The mid-IR region (4000-400 cm") was examined with a resolution of
2 cm". Spectral features were the same for ﬁlms cast on both KBr and 81. For titration
experiments, solutions were prepared in capped vials and stirred with a vortex mixer.
FTIR spectra of all liquids were obtained from samples prepared by placing a drop
between KBr windows. In situ heating experiments were performed under nitrogen purge
(99.95%, AGA Gas Co.) using a stainless-steel sample holder equipped with cartridge
heaters (Omega). The temperature was ﬁxed using an Omega CN 1200 controller.

X-ray Photoelectron Spectroscopy (XPS). Films cast on quartz slides were
analyzed with a VG Microtech spectrometer using a Clam2 hemispherical analyzer. All
XPS spectra were collected using a Mg anode (1253.6 eV) operated at a power of 300 W
(15 kV and 20 mA emission current) with an analyzer pass energy of 50 eV. Four regions
were scanned for each ﬁlm: C Is, 0 Is, Sn 3d, and the valence band. Binding energies
were referenced to adventitious carbon (C 1s = 284.6 eV) and were measured with a
precision of i 0.1 eV. Quantitative XPS calculations were performed using sensitivity
factors determined by the analysis of titanium valerate ﬁlms [15] and tin(IV) oxide. XPS
peaks were ﬁtied with 20% Lorentzian-Gaussian mix Voigt functions using a non-linear
least squares curve ﬁtting program [18 ]. This program was also used to remove X-ray
satellites ﬁom valence band spectra. It must be noted that the use of XPS to evaluate ﬁlm
composition assumes that the ﬁlms are homogeneous.

X-ray Diffraction (XRD). XRD diffraction patterns of tin oxide ﬁlms on quartz

slides were obtained with a Rigaku XRD diffractometeri employing Cu Ker radiation (A?

66

1.541838 A). The X-ray was operated at 45 kV and 100 mA. Diffraction patterns were

collected with a scan rate of 0.20 degree (29)/min and DD and DS slit widths of 1°.

3.4 Results and Discussion

Solution Chemistry. We have previously reported an FTIR investigation of
solution chemistry leading to the production of titanium and zirconiumvalerate polymers
[15]. In addition to examining ligand formation and hydrolysis reactions, we found that
FTIR analysis of solutions containing diﬁ‘erent valeric acid : metal alkoxide ratios provided
an estimate of the extent of carboxylation of the metal valerate polymers. A similar study
of tin ﬁlm solutions is presented here.

° FTIR spectra of solutions that contain varying concentrations of Sn(OPri)4, valeric
acid, and water are shown in Figure 3.1. For valeric acid (Figure 3.1a), the strong
absorbance at 1711 cm“1 is attributed to a dimeric asymmetric carbonyl stretch [19 ,20 ].
Sn(OPr‘)a is dissolved in a large excess of isopropanol (~1 :40) and, for this reason, its
FTIR spectrum (Figure 3. lb) largely resembles that of the alcohol solvent. Valerate
ligands formed when valeric acid and Sn(OPri)a are mixed (Figures 3.1c - 3. le) have
asymmetric carboxyl stretch (v.(COO)) vibrations which occur between about 1700 and
1500 cm". Corresponding symmetric stretch (v.(COO)) bands (1500 to 1250 cm") are
not shown because they are obscured by strong isopropanol absorbances [21 ].
Carboxylate coordination mode can be inferred from the frequency of the v..(COO)
absorbance band. Grigor’ev reported that the v..(COO) ﬁ'equency decreased with
decreasing OCO bend angle [22 ]. Since bond angle tends to decrease with coordination

mode as follows: monodentate > bidentate bridging > bidentate chelating, v..(COO) bands

67

 

 

Absorbance
A
O
V

(b)

(a)

J

1 1 1 1
1800 1700 1600 1500
Frequency (cm'l)

 

 

 

Figure 3.1 FTIR spectra of (a) valeric acid and (b) Sn(OPri)4 in isopropanol and
spectra of solutions containing different concentrations of valeric acid and
water. Sn(OPr‘)a : valeric acid : water ratios are: (c) 1 : 0.5 : 0, (d) 1 : 1 :
0, and (e) 1 : 1 : 2. Spectra are normalized to the isopropanol band 1769
cm .

of consider
presence 0|
band is lit
In 1
ratio (Figu
unreacted
bidentate v
and 1600 c
FTIR speCt
3.1d)conta
cm") Valera
Apparently
Additionallj
doubling ti
Concentratn
increased,
accommoda
isOpropélnol
metal Ceme
Snloprilt r.

each dim er
I

 

68

of considerably higher frequency than observed for the ionic form of the ligand suggest the
presence of monodentate ligands. For sodium valerate, the frequency of the v..(COO)
band is 1562 cm'1 [15].

In the spectrum of a solution prepared with a 0.5 valeric acid : Sn(OPri)4 mole
ratio (Figure 3.1c), the absence of a band at 1711 cm'1 indicates that there is little
unreacted valeric acid in the solution. The absorbance at 1541 cm'1 suggests that
bidentate valerate ligands have been formed. The broad, low intensity band between 1700
and 1600 cm'1 may be due to a small concentration of monodentate valerate ligands. The
F TIR spectrum observed for an equimolar mixture of Sn(OPri)a and valeric acid (Figure
3.1d) contains bands characteristic of both bidentate (1541 cm") and monodentate (1659
cm") valerate ligands as well as a small band due to unreacted valeric acid (1711 cm").
Apparently less than one valerate ligand per tin atom can be formed in this solution.
Additionally, a comparison of valerate band intensities in Figures 3.1c and 1d shows that
doubling the initial valeric acid concentration has little effect on the bidentate ligand
concentration; however, the concentration of monodentate ligands has been slightly
increased. This suggests that there is a limit to the number of bidentate ligands that can be
accommodated by the metal centers. Hampden-Smith et a1. [23] reported that in
isopropanol, Sn(OPri)4 is present as solvated dimers of the form [Sn(OPri)4 (HOPri)]2 with
metal centers bridged by two isopropoxide ligands. Considering the dimeric nature of
Sn(OPri)4, our results suggest that one bidentate valerate ligand readily coordinates with

each dimer. Since both tin atoms are expected to be equivalent, we believe that the

Sol 0
Snt'O
hydrc
was 2
with

suggt
maxir
Th5:
Ugant
soluti
by a
lSOpr
Valet;

COncr
from
is ShC

in ﬁlr

Unlike

69

bidentate valerate ligand may bridge the two metal centers. Further addition of valerate
ligands favors monodentate coordination.

Figure 3.1e shows the FTIR spectrum obtained for an equimolar mixture of
Sn(OPri)4 and valeric acid after addition of sufﬁcient water to completely hydrolyze
Sn(OPri)4. The increase in relative intensity of the band at 1711 cm'1 indicates that water
hydrolyzes some of the valerate ligands. Hydrolysis of a fraction of the valerate ligands
was also observed in similarly prepared titanium and zirconium solutions [15]. Reaction
with water leads to a lowering of the intensity of the bidentate v..(COO) band. This
suggests that bidentate valerate ligands are preferentially hydrolyzed. In addition, the
maximum intensity of the monodentate v.(COO) band has shifted to a lower frequency.
This may indicate that water is H-bonded to the dangling oxygen of monodentate valerate
ligands [21] (vide infra). Addition of water to the 0.5 valeric acid : Sn(OPri)a mole ratio
solution also results in the hydrolysis of valerate ligands (spectrum not shown) as indicated
by a small band at 1711 cm". The intensities of the v.(COO) bands relative to
isopropanol absorbances are the same as seen in Figure 3.1e. Apparently less than 0.5
valerate ligands remain bound to tin aﬁer the addition of water. This suggests that the
concentration of valerate ligands in the ﬁlms will be less than 0.5 per tin atom.

Structure of Tin Valerate Films. The FTIR spectra of tin valerate ﬁlms prepared
from Sn(OMe)2 or Sn(OPri)4 are essentially identical. A representative SnV ﬁlm spectrum
is shown in Figure 3.2a. The relative intensities of some of the absorbance bands can vary
in ﬁlms prepared at different times (vide inﬁa). Since the FT IR spectra of ﬁlms made
from a methoxide in methanol and an isopropoxide in isopropanol are identical, it is

unlikely that alkoxides, esters or alcohols are more than trace components of the ﬁlms.

A:

stillll
. i‘lll 1111'

I. 1.111 l
illlill
.llrll.l|ll.lt.lll

OUE:£.—Cm£ <

70

 

 

 

 

Frequency (cm'l)

Figure 3.2 FTIR spectra of (a) SnV and (b) SnC ﬁlms.

1 1 1 1 b 1
O
.3
(b)
8
E 8 s
8 N ('12 o 3.3 m
g _ 1 § '9 52 R
(a)
1 i i l l l L
3600 2500 1600 900 400

71

Indeed, there are no absorbances in any ﬁlm specn'um that can be attributed to the
presence of these species. All bands in the SnV spectrum can be attributed to vibrational
modes of either valerate ligands, sorbed water and/or hydroxyl ligands, or tin-oxygen
bonds.

Valerate Ligands. As in titanium and zirconium valerate ﬁlm spectra [15], three
bands at 2959, 2933, and 2873 cm'1 are present in the SnV spectrum which are similar in
frequency, shape, and relative intensity to the OH stretching absorbances observed for
valeric acid. Alkyl group vibrations at 1109 and 752 cm'1 are also present in all valerate
ﬁlm spectra and in the valeric acid spectrum. The absence of the characteristic valeric acid
absorbance at 1711 cm'1 indicates that these bands are [due to valerate ligands in the ﬁlms,
not to valeric acid. Additionally, bands which can be attributed to coordinated valerate
ligands can be readily identified in SnV film spectra. Bands at 1623 and 1542 cm'1 are due
to v..(COO) vibrations and those at 1450 and 1358 cm'1 to v.(COO) vibrations of the
valerate ligands [21, 24]. The frequencies of the asymmetric stretch vibrations are very
similar to those obserVed for these polymers in solution (Figure 3.1e).

The coordination mode of carboxylate ligands to a metal center is commonly
inferred by comparing the difference between the asymmetric and symmetric COO stretch
vibrations (Av) with that of the ionic form of the ligand [21,24]. A separation greater than
that in the spectrum of the ionic species indicates monodentate ligands, whereas smaller
separations are indicative of bidentate coordination [24]. The value of Av for sodium
valerate is 142 cm'1 [15]. The bands at 1623 cm'1 and 1358 cm'1 are proposed to be a pair

of asymmetric and symmetric COO stretches, as are those at 1542 cm'1 and 1450 cm".

The SCP
inner pe

these an

carboxy
modes j
Alcock
includin
1575, 1
approac
1635, l

ETOUps.

SUetchi
(Figure
absorb;
distin g1
119]. 1
the 8pc

SchICS

[hEm f

72

The separation of the outermost peaks is 265 crn'l indicating monodentate ligands. The
inner peaks are 92 cm'1 apart, a separation characteristic of bidentate ligands. Whether
these are bridging or chelatin g bidentate ligands cannot be determined from Av.

The presence of both monodentate and bidentate carboxylate ligands in tin
carboxylate compounds is not unusual. Garner et al. [25] reported both coordination
modes present in Sn(OzCEt)a with absorbances at 1690, 1565. 1445, and 1301 cm“.
Alcock et al. [26] studied the structure and infrared spectra of metal carboxylates
including two tin compounds: Sn(02CMe)a which has COO stretching vibrations at 1635,
1575, 1513, and 1400 cm'1 and is bidentate chelating with one asymmetric ligand
approaching a monodentate structure, and (MeaN)Sn(02CMe)5 with absorption bands at
1635, 1570, 1430, and 1365 cm'1 due to two bidentate and three monodentate acetate
groups.

Sorbed Water and/or Hydroxyl Ligands. A broad absorbance due to O-H
stretching vibrations is observed near 3000 cm'l in the spectra of all tin valerate ﬁlms
(Figure 3.2a). Since there is little, if any, valeric acid or alcohol present in these ﬁlms, this
absorbance is attributed to hydroxyl ligands and/or sorbed water. In general, water can be
distinguished from hydroxyl ligands by the 5(H-O-H) band at approximately 1620 cm'1
[19]. Unfortunately, strong absorbances due to valerate ligands dominate this region of
the spectrum making it impossible to use this feature to differentiate between the two
species in these ﬁlms.

Water and hydroxyl ligands can be distinguished by the energy required to remove

them from the ﬁlm [15]. Most sorbed water will be removed by heating at 100°C,

wherea.
a tin va
and the
relative
v.,(CO(
band as
ﬁlm is l
the fillr
carboxy

minutes

subsmn
H:b0nc
band i
dcstabi
OXYgCr
monod
Water }
fealUrc
Simull;

Cay-hex

73

whereas, hydroxyl ligands will not. Figure 3.3 shows the changes in the FTIR spectrum of
a tin valerate ﬁlm heated from 25°C (Figure 3.3a) to 100°C (Figure 3.3d) in dry ninogen
and then cooled to room temperature (Figures 3.3e and 3.31). Heating decreases the
relative intensities of both the broad hydroxyl absorbance centered at 3000 cm'1 and the
v..(COO) band at 1623 cm". An absorbance between 1400 and 1300 cm:1 (the v.(COO)
band associated with the v..(COO) band at 1623 cm“) also decreases in intensity as the
ﬁlm is heated. These absorption bands are largely removed by heating at 100°C. Cooling
the ﬁlm to room temperature in dry nitrogen results in a small increase in the hydroxyl and
carboxylate features (Figure 3.3e). Subsequent exposure of the ﬁlm to ambient air for 15
minutes restores these bands (Figure 3.30.

The hydroxyl absorbance band can be largely attributed to sorbed water since it is
substantially decreased by heating the ﬁlms to 100°C. Its frequency is typical of strongly
H-bonded O-H groups [20]. The simultaneous decrease in hydroxyl band and v..(COO)
band intensities suggests that the monodentate form of the carboxylate ligand is
destabilized by the removal of water. Sorbed water may be H-bonded to the dangling
oxygen of the monodentate carboxylic acid. As water is removed from the ﬁlm, the
monodentate form is destabilized and a bidentate coordination is assumed. (Although
water has a vibrational mode at about 1620 cm'1 and part of the loss of the carboxylate
feature at 1623 cm'1 may be due to loss of water, this would not account for the
simultaneous loss of the band between 1400 and 1300 cm". Complete removal of

carboxylate groups could also account for these spectral changes. This is unlikely,

Figure

74

 

 

 

 

 

 

 

W
(e) A M
g (d) -
E
8
.0
<1:
(C) . _— M
(b)
(a)
3600 2500 1600 900

Frequency (cm-1)

Figure 3.3 FTIR spectra of sequential treatments of an SnV ﬁlm: (a) initial spectrum
collected at 25°C; heated in nitrogen to: (b) 50°C, (c) 75°C, and ((1) 100°
C; (e) cooled to 25°C in nitrogen, and (f) exposed to ambient air at 25°C.

howel

cxposr

obsen
prepar
relativl
results

functit

[he tir.
3.221).
Ore] e
\‘(Sn4
simjla
low er
of vib

aS\we

COnsi
m0dil
COOTC

51ml];

75

however, since monodentate carboxylate features are restored after the heated ﬁlm is
exposed to ambient air.)

Variation in the relative intensity of the carboxylate stretching ﬁequencies was
observed for ﬁlms prepared at different times. The SnV ﬁlm shown in Figure 3.3a was
prepared under higher humidity conditions than the one shown in Figure 3.2a and it has a
relatively higher monodentate carboxylate concentration. This is consistent with the
results of the thermal study. The concenu'ation of water in these ﬁlms may, in part, be a
function of ambient conditions when the solution is prepared and the ﬁlm is cast.

Metal-Oxygen Bonds. The low energy region of the FTIR spectrum observed for
the tin valerate ﬁlm shows a broad absorbance band between 800 and 400 cm'1 (Figure
3.2a). A broad band in this frequency range was also reported for tin gels prepared by
Orel et al. [27] and by Giuntini et al. [28]. Orel et al. [27] ascribed bands 551 cm'1 to
v(Sn-O) modes and those at 667 cm'1 to v(Sn-O-Sn) vibrations. Giuntini et al. [28] made
similar assignments to bands at 540 cm“1 and 680 cm", respectively. The maximum of the
low energy absorbance band in the SnV spectrum occurs at 573 cm". The broad envelope
of vibrational bands includes Sn-O vibrations associated with hydroxyl and valerate ligands
as well as Sn-O-Sn vibrations associated with a metal oxygen backbone.

In summary, the structure of the SnV ﬁlms as determined by FTIR analyses is
consistent with that expected from the hydrolysis and condensation of valeric acid
modiﬁed tin alkoxides. SnV ﬁlms are comprised of a metal - oxygen backbone
coordinated with both monodentate and bidentate valerate ligands. This structure is

similar to that of titanium and zirconium valerate ﬁlms [15], however, no evidence of

76

monodentate valerate ligands was found in the titanium or zirconium ﬁlms. Sorbed water
is present in the SnV ﬁlms and some tin centers may be coordinated with hydroxyl ligands.

XPS Analysis. XPS was primarily used to identify and quantitate the species in the
surface of the ﬁlms. In the C Is spectra of SnV ﬁlms, peaks observed at 284.6 eV
(reference peak) and 288.5 eV are attributed to alkyl/adventitious carbon and carboxyl
carbon (valerate group), respectively. Oxygen in ﬁlms can be categorized into two types:
backbone oxygen (530.6 eV) and ligand oxygen (531.9 eV) [15]. FTIR analysis of SnV
ﬁlms suggests that ligand oxygen in these ﬁlms includes both valerate and hydroxyl
oxygen. Since XPS measurements were carried out under UHV conditions, most sorbed
water was removed from the ﬁlms prior to the analysis and, as a consequence, does not
contribute signiﬁcantly to the O 1s spectra. The XPS Sn 3d5n binding energy measured
for these ﬁlms (486.7 eV) is 0.3 - 0.4 eV higher than typically reported for tin oxide
materials [29]. The separation of 43.9 eV between the backbone oxygen Is and Sn 3d5n
peaks is characteristic of tin oxides [30].

Table 3.1 shows concentrations of valerate ligands, hydroxyl ligands, and
backbone oxygen relative to the tin concentration in the SnV ﬁlms as calculated ﬁ'om XPS
intensity ratios. The valerate concentration was determined directly from the carboxyl C
Is / Sn 3dm intensity ratio. The difference between the total ligand oxygen and valerate
oxygen (twolper valerate group) established the concentration of hydroxyl ligands in the
ﬁlms. As with the FTIR analysis, XPS analysis of SnV ﬁlms made from either alkoxide
precursor gave identical results, within experimental error.

The level of carboxylation in the SnV ﬁlms (0.26 valerate ligand per tin) is quite

low compared to values reported for titanium (1.2) and zirconium (1.8) valerate ﬁlms

Table

(AK/3:11
.21.]

l

 

(A
23

l

' Rela
10%.

77

Table 3.1 Composition of Films: Ratios' of Ligands and Backbone Oxygen to Tin
and Estimated Average Tin Oxidation State as Determined by Quantitative

 

 

 

 

 

XPS Analysis
Valerate 11111292911 £3521;me Tin Oxidation
ﬁlm Sn Sn Sn State
SnV 0.26 0.55 1.6 4.0
SnP - 0.52 1.8 4.1
SnC - 0.33 1.85 4.0

 

 

 

 

 

' Relative standard deviations for valerate, hydroxyl, and backbone oxygen ratios are S :1:
10%, S :1: 10%, and S. i 5%, respectively.

 

[19. H1
vdueext
concenn
reaCtion
mmcm
thezuc
Snﬂlh’
concer

acklis

ﬁhns
each
neun~

cmcu

ident
ﬁn h
PTEC
obse
YEdu
01nd

Oxidp

78

[15]. However, the relative extent of carboxylation of SnV ﬁlms is consistent with the
value expected from the FTIR study of the valeric acid-Sn(OPri)a solutions. Low valerate
concentrations in tin valerate ﬁlms may be related to high levels of alcohol in tin alkoxide
reaction mixtures. It is well-known that the maximum replacement of alkoxide ligands
with carboxylates can only be achieved if alcohol is removed during synthesis [21]. While
the alcohol concentration of the Sn(OPr‘)4 solutions is nearly three times that of the
Sn(OMe)2 solutions, in both Sn(OMe)2 and Sn(OPr‘)a precursor solutions, the initial
concentration of alcohol is approximately twice that of valeric acid. In contrast, valeric
acid is the major component of zirconium and titanium ﬁlm solutions [15].

Quantitative XPS results can also be used to infer the oxidation state of metals in
ﬁlms [15, 31]. Assuming a charge of -l for carboxylate and hydroxyl ligands and -2 for
each backbone oxygen, a net negative charge in each film can be determined. Electrical
neutrality requires that metal atoms have a charge equal and opposite to this. These
calculations indicate that tin has an average oxidation state of approximately +4 in all ﬁlms
in this study (Table 3.1). As stated previously, SnV ﬁlms made from either precursor are
identical according to FTIR and XPS analyses. This suggests that the oxidation state of
tin is the same in: all valerate ﬁlms even though the oxidation states are different in the
precursor alkoxides. Oxidation of Sn(OMe)2 is consistent with the vigorous bubbling
observed when it is mixed with valeric acid: protons oxidize Sn2+ to Sn“ and are in turn
reduced to form hydrogen gas. It is also known that most Sn2+ solutions are readily
oxidized to Sn“ and must be protected from exposure to air [32]. Therefore, a tin

oxidation state of +4 in a ﬁlm derived from Sn(OMe)2 is not surprising.

to use
basis .
oxidal

The 11

the 0)
used '
Sn“ 1
band

ﬁlms.

Oxygt
oxide
minu
With

3&811)
leads
Slate

phOtC
This .

featill

79

Since Sn2+ and Sn“ oxides have similar XPS Sn 3d5n binding energies, it is difﬁcult
to use Sn core level binding energies to conﬁrm the oxidation state of tin calculated on the
basis of ligand and metal charge balancing. It has been established, however, that the
oxidation state can be determined from the valence band spectra of tin oxides [30, 33-35].
The lowest binding energy feature in the valence band spectrum of SnOz is a prominent 0
2p - derived peak with a binding energy of approximately 5 eV. In SnO, there is a strong
Sn SS - derived feature at about 2 eV. Valence band spectra have been used to determine
the oxidation state of tin in different forms of tin oxide. For example, Themlin et al. [34]
used the areas of these two features to determine the relative concentrations of Sn2+ and
Sn4+ in single crystal and polycrystalline tin oxides. Sanjinés et al. [35] used the valence
band spectrum to determine the oxidation state of tin in amorphous sputter deposited
ﬁlms.

While SnV ﬁlms contain organic ligands, they are primarily comprised of metal
oxygen polymers with limited valerate ligand coordination. They are, therefore, largely tin
oxide - like materials. The valence band spectrum of a SnV ﬁlm collected in the ﬁrst 15
minutes of X-ray exposure time (Figure 3.43) is similar to that of SnOz. This is consistent
with our charge balance calculation (Table 3.1). We have found that the SnV ﬁlms are
easily photoreduced during XPS analysis. Extended exposure of SnV ﬁlms to X-rays
leads to the formation of a peak indicative of Sn2+ (2 eV) and a decrease in the oxidation
state calculated from core level intensities. An extreme example of the effects of X-ray
photoreduction on the valence band spectrum of an SnV ﬁlm is shown in Figure 3.4b.
This valence band spectrum was collected after 12 hours of X-ray exposure and contains

features similar to those reported for Sn“ oxides [30,34]. The calculated oxidation state

Figur

 

rll l lllllll .l1 . .
495:3:—

80

 

 

 

 

>n
3:
CD
5
a (b)
1—4
(a)
10.0 ' 5.0 0.0
Binding Energy (eV)

Figure 3.4 . Valence band photoelectron spectra of SnV ﬁlm (a) initially and (b) after

12h of X-ray exposure.

ofdni

analys

ﬁhnsl
speccr
34001
1620
hganc
spect
and z

OCCUl

band

Char

Cons

0ft,

lhg

81

of tin in this photoreduced ﬁlm is approximately +2.8. For this reason, all core band XPS
analyses of SnV ﬁlms were conducted within the ﬁrst 5 minutes of X-ray exposure.

Structure of Tin Oxide Films. Calcined Tin Films. XRD analysis of the SnC
ﬁlm showed no peaks other than a broad feature due to the quartz substrate. The FTIR
spectrum of the SnC ﬁlm is shown in Figure 3.2b. A broad O-H stretch absorbance near
3400 cm”1 is typical of hydroxyl ligands and/or sorbed water in tin oxide [36]. The band at
1620 cm", characteristic of sorbed water, is relatively small indicating that hydroxyl
ligands predominate. No bands characteristic of organic groups are observed in the SnC
spectrum. The spectrum of a calcined tin ﬁlm shows a strong, broad band between 800
and 400 cm'1 due to v(Sn-O-Sn) and v(Sn-O) vibrations. The maximum absorbance
occurs at 587 cm", a slightly higher frequency than seen for the SnV ﬁlms.

Giuntini et al. [28] noted an increase in the relative intensity of the v(Sn-O-Sn)
band at 680 cm'1 compared to the v(Sn-O) band at 540 cm'1 upon heating gels to 350°C.
Changes were attributed to the formation of O-Sn-O bridges from structural hydroxyl
groups. Orel et al. [27] noted similar changes for their gels and correlated it with
crystallization. We believe that the similarity of the low frequency bands observed in the
FTIR Spectra of 811C and SnV ﬁlms suggests that the SnC ﬁlms are amorphous. This is
consistent with XRD results obtained for the calcined ﬁlms.

The XPS C Is spectrum obtained for SnC ﬁlms contains peaks that can be ascribed
to adventitious carbon (284.6 eV) and carbonate species (288.8 eV). These features are
of very low intensity compared to the carbon peaks observed in the SnV ﬁlms. Based on

the trace levels of carbonate species detected in SnC ﬁlms (<0.04/Sn), SnC ls oxygen

peaks
The t
decre.
cakﬁr
uladt
the h
hs v2

SnO;

thel
oftl
cast
pert
Ire:
(:0:

Spa

in

t‘r

82

peaks are ascribed to backbone oxygen (530.3 eV) and hydroxyl groups (531.9 eV) only.
The binding energy of the Sn 3d5,2 peak is 486.4 eV. typical of tin oxides [29]. The
decrease in hydroxyl concentration and increase in backbone oxygen observed following
calcination (Table 3.1) suggests that removal of the carboxylate and hydroxyl ligands leads
to additional metal - oxygen backbone formation. Charge balance calculations based upon
the XPS results leads to a +4 oxidation state for tin in SnC ﬁlms. This is consistent with
its valence band spectrum (Figure 3.5b) which is very similar to that of polycrystalline
SnO; (Figure 3.5a).

Hydrogen Peroxide Treated Films. The effect of hydrogen peroxide treatment on
the FTIR spectrum observed for a tin valerate ﬁlm is shown in Figure 3.6. The spectrum
of the untreated SnV ﬁlm on Si (Figure 3.6a) is similar to the spectrum obtained for a ﬁlm
cast on KBr (Figure 3.221). Figure 3.6b is the spectrum of the same ﬁlm after hydrogen
peroxide treatment (SnP), and Figure 3.6c shows the spectrum of a hydrogen peroxide
treated, uncoated Si crystal. The spectra are normalized to the Si band at 1108 cm".
Comparison of the valerate ligand absorbance bands in the treated and untreated ﬁlm
spectra shows that hydrogen peroxide treatment removes most of the valerate ligands.
The hydroxyl absorbance intensity in the Sn? and SnV ﬁlm spectra are similar, however,
the maximum intensity is shifted to higher frequency in the SnP ﬁlm spectrum. Recall that
in the SnV ﬁlms, the lower frequency hydroxyl absorbances were attributed to water H-
bonded to monodentate valerate ligands. The frequency shift is therefore consistent with
the removal of the valerate ligands. CH stretch bands at about 2900 cm'1 may be due to
traces of a hydrocarbon residue remaining after removal of valerate ligands. An

absorbance band centered at 1620 cm'1 in the Sn? spectrum can be largely attributed to

 

55:3:—

Figu

83

 

 

 

 

g (C)
E
.5

(b)

(a)

10.10 550 0:0

Binding Energy (eV)

Figure 3.5 Valence band photoelectron spectra of (a) tin(IV) oxide, (b) SnC ﬁlm, and
(c) SnP ﬁlm.

 

 

 

 

 

 

 

 

oucsseCmn—xx

Flgu;

84

 

(a)

 

 

 

Absorbance
7"

 

a» l

(C) J

r ‘—

 

 

l

3600 2500 1600
Frequency (cm'l)

 

Figure 3.6 FTIR spectra of SnV ﬁlm on Si crystal (a) before and (b) after hydrogen
peroxide treatment (SnP), and (c) the uncoated Si crystal.

sorbed w

attributai

vibration
absorba
gels we
SnV fr
hydrog
the ch
hydro
peror
Hyd;
400

uncl

Sn

Spe
e\’
an

de

8‘

CC

85

sorbed water. As noted for the SnC ﬁlm, XRD analysis of the SnP ﬁlm showed no peaks
attributable to the ﬁlm.

The effect of hydrogen peroxide treatment on the low frequency tin-oxygen
vibrations cannot be determined using ﬁlms coated on Si crystals because Si has strong
absorbances below 1250 cm". For this reason hydrogen peroxide treatment of tin valerate
gels was also studied. The tin valerate gel spectrum (Figure 3.7a) is similar to that of the
SnV ﬁlm with an additional, strong contribution from sorbed valeric acid. The effects of
hydrogen peroxide treatment on the hydroxyl and valerate absorbance bands are similar to
the changes induced in the SnV ﬁlm: most valerate groups are removed and the maximum
hydroxyl absorbance is, shifted to higher frequency (Figure 3.7b). The spectrum of the
peroxide treated gel also contains an absorbance at 1620 cm'1 due in part to sorbed water.
Hydrogen peroxide treatment has little effect on the strong absorbance between 800 and
400 cm", suggesting that the structure of the tin oxide polymer network is largely
unchanged by carboxylate removal.

XPS binding energies of backbone O ls (530.7 eV), ligand 0 1s (531.9 eV), and
Sn 3d5n (486.7 eV) peaks have the same values as measured for SnV ﬁlms. The C Is
spectrum of an SnP ﬁlm contains three peaks with binding energies of 284.6 eV, 286.4
eV, and 288.7 eV. Since FTIR analysis detects only traces of residual carbon species, we
attribute the SnP carbon to a residue of hydrogen peroxide reaction products which have
deposited on the ﬁlm surface. The lowest binding energy peak is due to a combination of
alkyl and adventitious carbon. Carbon ls peaks with binding energies of 288.7 and 286.4
eV are typical of carbon-oxygen moieties with a variety of carbon to oxygen ratios. The

combined concentration of these species is about 0.2 per tin atom in the surface of the

 

Abs or'barr ‘c

86

 

 

 

 

Q)
Q
5 (b)
.D
‘5
' .‘8
<
(a)
3600 2500 1600 960 460

Frequency (cm'l)

Figure 3.7 FTIR spectra of tin valerate gel (a) before and (b) after hydrogen peroxide
treatment.

ﬁlms. Ft
each earl
determrn
associate
it results
expectet
the vacr
that no
establisi
metal t
earbox;

increas

valeric:
to the
resPeer
COngis.
ﬁlms a
xm,

lobes

87

ﬁlms. For our calculations, it was assumed that there was one oxygen associated with
each carbon of these binding energies. The hydroxyl concentration (Table 3.1) was then
determined from the intensity of the ligand oxygen peak after subtraction of oxygen
associated with surface carbon contamination. While this assumption is only approximate,
it results in a reasonable value for the hydroxyl concentration. Again, sorbed water is not
expected to make a signiﬁcant contribution to 0 Is spectra since it is largely removed by
the vacuum system of the XPS instrument prior to analysis. We should also emphasize
that no Si (quartz substrate) was detected in the XPS analysis of SnP ﬁlms. This further
establishes the belief that hydrogen peroxide treatment does not signiﬁcantly damage the
metal oxide polymer structure. However, as noted for SnC ﬁlms, removal of the
carboxylate ligands increases the backbone oxygen content (T able 3.1), presumably by
increasing the extent of polymer crosslinking.

Our analyses have suggested that SnP ﬁlms are tin oxide - like materials. The
valence band spectrum of an SnP ﬁlm (Figure 3.6c) is similar in shape and binding energy
to the valence band spectra of polycrystalline SnOz and SnC ﬁlms (Figures 3.6a and 3.6b,
respectively). This indicates that tin is in a +4 oxidation state in SnP ﬁlms . This is
consistent with the quantitative XPS charge balance (Table 3.1). Like SnV ﬁlms, SnP
ﬁlms are subject to X-ray photoreduction and thus analyses were performed using minimal

X-ray exposure times. Calcined tin oxide ﬁlms and polycrystalline SnOz were not found

to be susceptible to photoreduction under our analysis conditions.

3.5

l0.

ll.

Re‘

M
AI

3.5

10.

11.

12.

13.

88

References

Madou, M.J.; Morrison, S.R. Chemical Sensing with Solid State Devices;
Academic Press: San Diego, CA, 1989.

Fuller, M.J.; Warwick, M.E. J.Catal. 1973, 29, 441. McAleer, J. J. Chem. Soc.
Faraday Trans. I 1979, 75, 2768. Berry, F.J. Adv. Catal. 1981, 30, 97.

Shewchun, J.; Singh, R.; Green, M.A. J. Appl. Phys. 1977, 48, 765. Bucher, E.
Appl. Phys. 1978. I 7. 1.

Peaker, A.R.; Horsley, B. Rev. Sci. Instrum. 1971, 42, 1825.

Cocco, G.; Enzo, S.; Carturan, G.; Giordano Orsis, P.; Scardi, P. Mater. Chem.
Phys. 1987, 17, 541-551.

Von Ortenberg, M.; Link, 1.; Helbig, R. J. Opt. Soc. Am. 1977, 67, 968.

Kohl, D. Sens. Actuators 1989, 18, 71. Ippommatsu, M.; Sasaki, H. J.
Electrochem. Soc. 1989, 136(7), 2123. Capehart, T.W.; Chang, S.C. J. Vac. Sci.
T echnol., 1981, 18(2), 393. .

Pink, H.; Treitinger, L. Vite, L. Japan. J. Appl. Phys. 1980, 19(3), 513. Tomar,
M.S.; Garcia, F.J. Prog. Cryst. Growth Charact. 1981, 4, 221.

Lalauze, R.; Pijolat, C.; Vincent, 8.; Bruno, L. Sens. Actuators 1992, 8, 237. Kim,
K.; Finstad, T.G.; Chu, W.K.; Cox, X.B.; Linton, R.W. Solar Cells 1985, I3, 301.

Brinker, C.J.; Scherer, G. Sol-Gel Science; Academic Press: San Diego, 1990.

Roger, C.; Hampden Smith, M.J.; Brinker, C]. In Better Ceramics Through
Chemistry V, Hampden Smith. M.J., Klemperer, W.G., Brinker, C.J., Eds.; Mater.
Res. Soc. Symp. Proc. 271; MRS: Pittsburgh, PA, 1992; 51.

Yoldas, B.E. J. Non-Cryst. Solids 1980, 386239, 81. Thomas, I.M. In Sol-Gel
Technology for Thin F ilms, Fibers, Preforms, Electronics and Specialty Shapes,
Klein, L.C., Ed.; Noyes Publications: Park Ridge, N.J., 1988; 2. Wark, T.A.;
Gulliver, E.A.; Jones, L.C.; Hampden-Smith, M.J.; Rheingold, A.L.; Huffman, J.C.
in Better Ceramics Through Chemistry IV, Zelinsky, B.J.J.; Brinker, C.J.; Clark,
D.E.; Ulrich, D.R., Eds.; Mater. Res. Soc. Proc. 180: Pittsburgh, PA, 1990; 61.
Wellbrock, U.; Beier, W.; Frischat, G.H. J. Non-Cryst. Solids 1992, 14742148,
350.

Livage, J.; Henry, M.; Sanchez, C. Prog. Solid State Chem. 1989, 18, 259.

l4.

l5.

l6.

l7.

l8.

19.

r.)
(J!

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

89

Gagliardi, C.D.; Berglund, K.A. in Processing Science of Advanced C eramics,
Aksay, K.A.; McVay, G.L.; Ulrich, D.R., Eds.; Mater. Res.Soc. Proc. 155:
Pittsburgh, PA, 1986; 127. Gagliardi, C.D.; Dunuwila, D.; Berglund, K.A. in
Better Ceramics Through Chemistry IV, Zelinsky, BJ.J.; Brinker, C.J.; Clark,
D.E.; Ulrich, D.R., Eds.; Mater. Res. Soc. Proc. 180: Pittsburgh, AA, 1990; 801.
Dunawila, D.; Berglund, K.A. Chem. Mater. 1994, 6(9), 1556.

Severin, K.G.; Ledford, J.S.; Torgerson, B.A.; Berglund, K.A. Chem. Mater.
1994, 6(7). 890.

Hampden-Smith, M.J.; Wark, T.A.; Brinker, C.J. Coor. Chem. Rev. 1992, 112,
81.

Takahashi, Y.; Wada, Y. J. Electrochem. Soc. 1990, 137, 267.

Software provided by Dr. Andrew Proctor, University of Pittsburgh, Pittsburgh,
PA.

Colthup, N.B.; Daly, L.H.; Wiberley, S.lE. Introduction to Infrared 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.

Mehrotra, R.C.; Bohra, R. Metal Carboxylates; Academic Press: London, 1983.

Grigor’ev, A.I. Russ. J. Inorg. Chem. 1963, 8(4), 409.

Hampden-Smith, M.J.; Wark, T.A.; Rheingold, A.; Huffman, J.C. Can. J. Chem.
1991, 69, 121.

Deacon, G.B.; Phillips, RJ. Coord. Chem. Rev. 1980, 33, 227-250.
Garner, C.D.; Hughes, B.; King, T.J. J. Chem. Soc., Dalton Trans. 1975, 562.

Alcock, N.W.; Tracy, V.M.; Waddington, T.C. J. Chem. Soc., Dalton Trans.
1976, 2243.

Orel, B.; Lavrencic-Stangar, U.; ijak-Orel, 2.; Bukovec, P.; Kosec, M. J. Non-
Cryst. Solids 1994. 167. 272.

Giuntini, J.C.; Granier, W.; Zanchetta, J.V.; Taha, A. J. Mater. Sci. Lett. 1990, 9,
1383.

 

36.

29.

30.

31.

32.

33.

34.

35.

36.

90

Handbook of X -ray Photoelectron Spectroscopy; Wagner, C.D.; Riggs, W.M.;
Davis, L.H.; Moulder, J.F.; Muilenburg, G.E., Eds.; Perkin Elmer, 1979.

Lau, C.L.; Wertheim, G.K. J. Vac. Sci. Technol. 1978, 15(2), 622.
Noonan, G.O.; Ledford, J.S. Chem. Mater. 1995, 7(6), 1117.

Donaldson, J.D. Prog. Inorg. Chem. 1968, 8, 287.

Sherwood, P.M.A. Phys. Rev. B 1990, 41(14), 10 151.

Themlin, I.M.; Chtaib, M.; Henrard, L.; Lambin, P.; Darville, J.; Gilles, J.-M. Phys.
Rev. B 1992, 46(4), 2460.

Sanjinés, R.; .Coluzza, C.; Rosenfeld, F.; Gozzo, F.; Alméras, Ph.; Lévy, F.;
Margaritondo, G. J. Appl. Phys. 1993, 73(8), 3997.

Thorton, E.W.; Harrison, P.G. J.C.S. Faraday I 1975, 71, 461.

4.1 r

prepare
of 100
solutio
calcini
Stable
temps
Uansf
ultras
ﬁlms
ﬁling
Sn u
SUrf:
Com
at 4

Chapter 4

Synthesis and Characterization of Sol-Gel Derived
Titanium-Tin Oxide Films

4.1 Abstract

In this study the homogeneity and properties of titanium-tin oxide thin ﬁlms
prepared using a modiﬁed sol-gel technique are examined Films with Ti:Sn atomic ratios
of 100:0, 75:25, 50:50, 25:75, and 0:100 have been formed by spin casting hydrolyzed
solutions of tin(IV) and titanium(IV) isopropoxides and valeric acid onto quartz slides and
calcining in air at temperatures between 250 and 1000 °C. All ﬁlms are transparent,
stable, and exhibit uniform adhesion to quartz. The effect of composition and calcination
temperature has been determined using X-ray photoelectron spectroscopy (XPS), Fourier
transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), ellipsometry, and
ultraviolet-visible spectroscopy (UV-Vis). Additionally, the homogeneity of the mixed
ﬁlms has been explored-with angle resolved XPS (AR-XPS). It was found that “as cast”
ﬁlms prepared by mixing separately hydrolyzed alkoxides have a surface enhancement of
Sn which is not present if the alkoxides are hydrolyzed together. Calcination causes Ti to
surface segregate and the degree of surface segregation is a function of both ﬁlm
composition and calcination temperature. Mixing suppresses the formation of crystallites
at 400 °C and at higher temperatures only mixed TiOz and SnOz rutile phases are

detected. Depending upon the composition, “as cast” ﬁlms range in thickness from 400 -

91

300 nm and

150°C. Re

chemical j
titanium t
higher cc
l2. 3). '
strongly
candida
and im‘
Madde
was i

aPPﬁc

T102.

200(
DIS:
ion
inh
c0.

Ge

92

300 nm and collapse between 50% and 20% upon calcination at temperatures as low as

250 °C. Refractive indices were also found to be related to ﬁlm composition.

4.2 Introduction

Mixing metal oxides is a promising approach for tailoring the physical and
chemical properties of materials for use in a wide range of applications [1]. Tin and
titanium oxides have been mixed to produce catalysts with modiﬁed reactivity [2] and
higher concentrations of strong acid sites than present in either of the single metal oxides
[2, 3]. Takahashi and Wada [4] found that the conductivity of TiOz-Sn02 ﬁlms was
strongly dependent upon the components of the ambient gas, making them ideal
candidates for gas sensing materials. Indeed, gas sensors with enhanced sensitivity [5,6]
and improved response times [5, 7] have been produced from mixed TiOz-Snoz materials.
Maddelena et al.[8] also demonstrated that the electrical conductivity of tin oxide glasses
was increased by doping with titanium oxide. In addition to catalytic and sensor
applications, the ability to control refractive indices and conductivities may make mixed
TiOz-Snoz thin ﬁlms useful as new transparent coatings.

Classically, mixed metal oxides are prepared by repeated high temperature (1000-
2000 °C) heating and grinding of single metal oxides in the desired atomic ratios.
Disadvantages of this method include high processing temperatures, contamination from
ionic impurities in raw materials. irreproducible stoichiometry, and the production of
inhomogeneous materials [1, 9]. Other researchers have prepared TiOz-SnOz powders by
coprecipitation of alkoxides [3, 10] or chlorides [2,5] and ﬁlms using chemical vapor

deposition (CVD) [4] and sol-gel [5,8] techniques. Sol-gel synthesis from metal alkoxides

is especiall
Multicomr
high purit
and it is .
homo gen
of the so
sensing

sol-gel s

depend
phases
charac
ﬁlms.

02,13
eXpen'
depos
on [he
215 Car
there!

metht

Produ

93

is especially advantageous because it avoids the problems inherent in classical methods.
Multicomponent oxides can be synthesized at relatively low temperatures, reactants are of
high purity, there are no residual impurities introduced from the presence of counterions,
and it is a wet chemical method which allows better control over the stoichiometry and
homogeneity of the materials produced [1,9, 11]. Additionally, the rheological properties
of the sols make them ideal for the production of thin ﬁlms which are useful for both gas
sensing and coatings applications [9]. In the current study we explore the use a modiﬁed
sol-gel synthesis method to make TrOz-SnOp, thin ﬁlms from metal alkoxides.

The catalytic, physical, optical, and electronic properties of thin ﬁlms are strongly
dependent, not only upon the particular metal oxide(s) in the ﬁlms, but also upon the
phases present and in the case of crystalline materials, orientation of the crystallites. These
characteristics are, in turn, strongly dependent upon the method used to synthesize the
ﬁlms. For instance, TiOz ﬁlms have different characteristics when produced using CVD
[12,13]; sputter deposition [14,15]. sol-gel [16,17], or atomic layer epitaxy [18,19]. Also,
experimental conditions such as gas pressure and substrate temperature for sputter
deposition [14] and additives used in a sol-gel synthesis [16] can have a strong inﬂuence
on the phases present, phase transition temperatures, and preferred crystallite orientations,
as can impurities or dopants [20] and the type of substrate used [13]. It is important,
therefore, to determine the unique characteristics of ﬁlms produced using a particular
method.

In the ﬁrst part of this paper we evaluate variations of the sol-gel method for

producing homogeneously mixed ﬁlms. We have previously reported the synthesis of

titanium

would b
alkoxide
CXLCHI C
oligome
Alkoxic'

polyme

oligom
favor i
been (1
tOgeth
homo‘
alk0x
are q
the a
pGC;
and 1

Solu

SP6:

94

titanium valerate and tin valerate ﬁlms from Ti(OPr‘)a and Sn(0Pri)a , respectively [21,
22]. The simplest method for preparing mixed metal oxide ﬁlms from these alkoxides
would be to mix them together and then carry out the synthesis as for the individual
alkoxides. Factors which may inﬂuence the homogeneity of the resultant ﬁlms are the
extent of oligomerization of the alkoxides and their relative hydrolysis rates. Highly
oligomeric alkoxides would be expected to produce heterogeneous mixed materials.
Alkoxides with different hydrolysis rates have a kinetic incompatibility that favors self-
polymerization [23].

The alkoxides chosen for the synthesis: Ti(OPri)4 and Sn(OPri)4 have low levels of
oligomerization: they are monomeric [24] and dimeric [25], respectively, and this should
favor intimate mixing of the two alkoxides before polymerization. Additionally, it has
been demonstrated that double metal alkoxides can be formed by reﬂuxing the alkoxides
together [26]. This could potentially further enhance the likelihood of producing
homogeneously mixed materials. The relative hydrolysis and condensation rates of these
alkoxides after complexation with valerate ligands is unknown but if the hydrolysis rates
are quite dissimilar, the resultant material would resemble one prepared by polymerizing
the alkoxides separately, followed by mixing. In this study, three variations are used to
prepare 50:50 mixtures: mixing the alkoxides with and without refluxing before hydrolysis,
and polymerizing the alkoxides separately, followed by mixing. ﬁlms prepared from these
solutions are compared.

The methods used to evaluate these ﬁlms are Fourier transform infrared

spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), and angle resolved XPS

(AR-X
Ti-O-S
provid
or Sn

surfac

the st
(XRI

chars

Addi

m m:

4.3

titar
Che

W215

Itic]
We
the

Our

95

(AR-XPS). Identiﬁcation of a band in the FTIR spectrum which could be attributed to a
Ti-O-Sn stretch or a peak in the O ls XPS spectrum due to oxygen in this moiety would
provide direct evidence of mixing. AR-XPS is used to detect a surface enhancement of T1
or Sn. Commonly one component of an inhomogeneous material will segregate at the
surface and this could be identiﬁed and characterized with AR-XPS.

In the second part of the paper, the effect of calcination (250 °C to 1000 °C) on
the structure and properties of TiOz and SnOz ﬁlms is examined using X-ray diffraction
(XRD), XPS, ultraviolet-visible spectroscopy (UV-Vis), and ellipsometry. The
characteristics of these ﬁlms are compared with that of ﬁlms prepared with Ti:Sn ratios of
25:75, 50:50, and 75:25 and calcined at temperatures between 400 and 1000 °C.
Additionally, AR-XPS is used to characterize surface segregation of oxides in mixed oxide

ﬁlms and XRD to identify crystalline mixed oxide phases.

4.3 Experimental

Materials. Tin(IV) isopropoxide 10% w/v in isopropanol (Chemat Technologies),
titanium(IV) isopropoxide (Aldrich Chemical Company), and valeric acid (99+%, Aldrich
Chemical Company) were used without further puriﬁcation. Distilled, deionized water
was used for all syntheses.

Film Preparation. The methods used to synthesize ﬁlms were based on
techniques developed previously [21, 22]. Films synthesized from Sn(OPri)4 were
prepared using mixtures of 40 isopropanol : 9 valeric acid : 1.5 water : 1 Sn(OPri)4 and
those from Ti(OPr‘)4 from 9 valeric acid : 1.5 water: 1 Ti(OPri)a . Reactions were carried

out at room temperature in capped vials in a N2 purged glovebox. Valerie acid was added

to the alkc
following
Tl
an equim
was add:
1.5 '. l.
solution
the alkt
Metho<
previoi
Stirred

l with

Clllart

l—Cmp

ﬁlm;
were
all [‘1

and

96

to the alkoxide followed by water. A vortex mixer was used to vigorously stir solutions
following the addition of each reactant.

Three different methods were used to synthesize 50:50 mixed ﬁlms. In Method 1,
an equimolar mixture of Sn(OPr‘)a and Ti(OPri)4 was stirred vigorously and valeric acid
was added immediately, followed by water. The acid to water to alkoxide ratios were 9 :
1.5 : 1. ISOpropanol, from the Sn(OPri)a precursor solution, was also present in these
solutions in a 20 alcohol : l alkoxide ratio. Method 2 is the same as Method 1 except that
the alkoxide mixture was reﬂuxed for one hour before the addition of valeric acid. In
Method 3 separate solutions were prepared from the two alkoxides as described in the
previous paragraph. After standing for three hours, the two solutions were combined and
stirred with a vortex mixer. For the calcination study, ﬁlms were prepared using Method
1 with the appropriate relative molar concentrations of the alkoxides.

Freshly prepared solutions were dispersed on cleaned, acetone rinsed and dried
quartz slides, spun for ﬁve minutes, and air-dried overnight. Films calcined at
temperatures 5600 °C were prepared by heating ﬁlms in a tube furnace under a ﬂow of dry
air (100 cm3/min, medical grade, AGA Gas Co.) at the desired temperature for 24 h.
Films calcined at 800 °C or 1000 °C were heated in a mufﬂe furnace for 24 h. All ﬁlms

were stored in a desiccator before analysis or further treatment Unless otherwise noted,
all ﬁlms prepared in this study were visually transparent, colorless, and exhibited complete
and uniform adhesion to quartz subsu'ates.

Samples are designated with the following abbreviations: 100 % Ti ﬁlm = Ti, 75%

Ti 25% Sn = TS, 50% Ti 50% Sn = EM (equimolar), 25% Ti 75% Sn = ST, and 100% Sn

= Sr
and
valer
ﬁlms

75%

analy
XPS

(15 it
were
refer:
of i

Sensil
ﬁtted
curve
ﬁlms

inlht

97

= Sn. Films studied without further treatment are referred to as valerate or “as cast” ﬁlms
and designated with “Val” (i.e. 100 % Sn “as cast” ﬁlm = SnVal or a 75% Ti 25% Sn
valerate ﬁlm = TSVal). Films which were heated in air are referred to as calcined or oxide
ﬁlms and designated with a “C” followed by the calcination temperature (i.e. a 25% Ti
75% Sn ﬁlm which has been calcined at 400 °C = STC400).

X-ray Photoelectron Spectroscopy (XPS). Films cast on quartz slides were
analyzed with a VG Microtech spectrometer using a Clam2 hemispherical analyzer. All
XPS spectra were collected using a Mg anode (1253.6 eV) operated at a power of 300 W
(15 kV and 20 mA emission current) with an analer pass energy of 50 eV. Four regions
were scanned for each ﬁlm: C ls, 0 Is, Sn 3d. and Ti 2p. Binding energies were
referenced to adventitious carbon (C Is = 284.6 eV) and were measured with a precision
of i 0.1 eV. Quantitative XPS calculations were performed using empirically derived
sensitivity factors previously determined in this laboratory [21, 22]. XPS peaks were
ﬁtted with 20% Lorentzian-Gaussian mix Voigt functions using a non-linear least squares
curve ﬁtting program [27]. Unless otherwise noted, composition of mixed metal oxide
ﬁlms are reported as a molar ratio with respect to the total metal concentration (Ti + Sn)
in the film. Values reported have a standard deviation of < 10%.

Angle Resolved XPS (AR-XPS). Mixed ﬁlms were analyzed at angles between
0° (deepest) and 90° (shallowest) with respect to the surface normal at 15° intervals. Two
regions were scanned: the C ls region to provide a binding energy reference and a 100 eV
region (binding energy range: 545 eV to 445 eV) which contained the O ls, Sn 3d, and Ti

2p peaks. Over this narrow range of kinetic energies it is valid to assume that the

photoelecr
was deten
3).. Usin
A was ca
measoft
enuovn
(kmnnn
reported
prepare

<5‘7c.

each or‘:
Surface
Show“
In 0rd.

Sensitir

slides ,
L5418
COHecu
WuSte]

Peak “

98

photoelectrons have similar escape depths. The mean free path (it) of the photoelectrons
was detemrined from the universal curve [28] and the sampling depth was assumed to be
3h. Using a monolayer thickness of 3 A, a maximum sampling depth of approximately 50
A was calculated. Relative concentrations of Sn and T1 were determined by measuring the
areas of the Sn 3dm peak and the Ti 2p peaks and applying a sensitivity factor. The Ti/Sn
sensitivity factor was determined by taking a ratio of the O/Sn and O” i sensitivity factors
determined by analysis of SnO; and TD; polycrystalline powders, respectively. Values
reported are normalized by dividing by the relative concentration of Ti and Sn used to
prepare the ﬁlms. While all standard deviations of reported values are <10%, most are
<5%.

In our instrument conﬁguration the X-ray source and the analyzer are at 90° from
each other. . Since X-ray emission is parallel to the sample surface for a 0° analysis angle, a
surface sensitivity due to the source is created. In all cases, samples collected at 0°
showed a higher concentration of the surface enhanced species than was measured at 15°.
In order to avoid confusion due to the convolution of source and analyzer surface
sensitivity, data collected with a 0° angle is not reported in this study.

X-ray Diffraction (XRD). XRD diffraction patterns of calcined ﬁlms on quartz
slides were obtained with a Rigaku XRD diffractometer employing Cu Kcr radiation (it:
1.541838 A). The X-ray was operated at 45 kV and 100 mA. Diffraction patterns were
collected using DD and DS slit widths of 1°. The background due to the quartz substrate
was removed from the patterns by subtracting the XRD pattern of a blank quartz slide.

Peak widths and locations were determined using a non-linear least squares curve ﬁtting

pl’l

us?

Si g
M

1h:

99

program [27] and assuming Gaussian line shapes. Mean particle sizes were calculated
using the Scherrer equation [29] assuming no line broadening due to stress.

Ultraviolet-Visible Spectroscopy (UV-Vis). Absorbance spectra of ﬁlms cast on
quartz slides were collected with a Hitachi U4001 UV-Vis spectrophotometer over a
range of 190-750 nm using a scan speed of 300 nm/min and a 2 nm slit width. A blank
quartz slide was used in the reference beam.

Absorption coefﬁcient (or) data was ﬁtted to:

ahv = Mint-15,)2 4.1

where hv is the photon energy, E, is the optical band gap energy, and A is a constant [30].
Band gaps were then determined from plots of (ahv)m versus hv. Values reported are i
0.15 eV.

Ellipsometry. Refractive indices and ﬁlm thicknesses were determined using an
Auto EL 11 (Rudolph Research International, USA) ellipsometer using a He-Ne laser (7t =
632.8 nm) as the light source. The reported values are the result of measurements of at
least three ﬁlms and four or more measurements were made on each ﬁlm. There was no
signiﬁcant difference in values measured at different places on a single ﬁlm, indicating that
within experimental error, the ﬁlms were of uniform thickness and refractive index over
the entire surface. Standard deviations for thickness and refractive index measurements

are less than 5% and 3%, respectively.

4.4

4.4.1

com

Ugar

tWo
baht
due.
banc
Vale

Whit

100

4.4 Results and Discussion
4.4.1 Effect of preparation method on film homogeneity.

The FTIR spectra of valerate ﬁlms made from range of relative concentration of Ti
and Sn are shown in Figure 4.1. The mixed ﬁlms were prepared using Method 1.
Identiﬁcation of bands in tin valerate (Figure 4.1a) and titanium valerate (Figure 4.1e) ﬁlm
spectra have been previously reported [21, 22]. In general, the bands at frequencies
higher than 1000 cm’1 are due to the presence of the valerate ligands and hydroxyl groups,
while those at lower frequencies are primarily due to metal - oxygen stretches, both in the
polymer backbone and due to ligand coordination to the metal centers. The mixed ﬁlm
spectra (Figures 4.1b - 4.1d) appear to be largely a combination of the SnVal and TrVal
ﬁlmspectra and indicate that, like the single metal ﬁlms, the mixed valerate ﬁlms are
comprised of metal - oxygen polymer backbones coordinated with valerate and hydroxyl
ligands.

The low frequency region of all spectra shows strong, broad bands. Unless the
two metals alternate rigorously in the polymer networks both v(Ti-O-Ti) and v(Sn-O-Sn)
bands will be present in the mixed ﬁlm spectra, as will bands from metal-oxygen stretches
due to ligand coordination. Unfortunately, because of the breadth and intensity of these
bands, it is difﬁcult to identify the presence of a v(Ti-O-Sn) band in any of the mixed
valerate ﬁlm spectra. Additionally, the FTIR spectra of 50:50 ﬁlms made from alkoxides
which were heated together (Method 2) and the spectra of ﬁlms made by mixing
individually polymerized alkoxides (Method 3) are not signiﬁcantly different from that
shown in Figure 4.1c. FTIR spectra of the three 50:50 mixed ﬁlms calcined at 250 °C are

also indistinguishable from one another (not shown) and again, no bands could be

 

85553 < emf—5:5 Z

101

 

 

 

 

 

l l l l T
(e) M
3 ((0
G
E
0
§
3 (C)
.s
76‘
E
2
(b)
(a)
3600 2560 16100 300 460

Frequency (cm'l)

Figure 4.1 FTIR spectra of valerate ﬁlms: (a) SnVal, (b) STVal, (c) EMVal, (d)
TSVal, and (e) TiVal.

identil
al [3
syntht
stretc
not S
they

resul'

thee

O 15
529.
IWO
sepa
(this

have

OCCI

inhC

102

identiﬁed that were not present in the TiC250 and SnC250 ﬁlm spectra. Nakabayashi et
al. [3] used Raman spectroscopy to study the formation of mixed TiOz-Snoz powders
synthesized by c0precipitation of metal alkoxides and reported that bands due to Ti-O-Sn
stretches were present between 770 and 430 cm". These bands were very broad so it is
not surprising that we are unable to isolate these stretches in the FTIR spectra, though
they occur in the frequency range we are scanning and should be infrared active. These
results are inconclusive and indicate that this is not a promising approach for evaluating
the extent of mixing.
XPS analysis was also explored as a method for evaluating ﬁlm homogeneity. The
0 Is binding energies for backbone oxygen in SnVal and TiVal ﬁlms are 530.6 eV and
529.7 eV, respectively. This separation should be adequate to distinguish between the
two peaks, if tin valerate and titanium valerate polymers have polymerized completely
separately. Assuming equal intensities and a FWHM of 2 eV for each of the 0 1s peaks
(this is the FWHM usually measured for single metal ﬁlms), the combined peak would
have a FWHM of 2.25 eV, an increase of >12 %. Curve ﬁtting the backbone oxygen 0 ls
peak for all 50:50 mixed ﬁlms shows a reasonable ﬁt with only one backbone 0 1s peak
(530.2 eV). In the mixed ﬁlms. ratios of the 0 Is FWHM to that of the Sn 3dm and Tr
2P3); peaks are similar (greater than but within 5%) to that measured for the pure Sn and
Ti valerate and calcined ﬁlms.
Several things are apparent from these results. First, some mixing must be
Occurring because the 0 Is peaks are narrower than expected for a completely

inhomogeneous mixture. Second. these results are true for 50:50 mixed ﬁlms prepared

with .

polyn

our p

miXt
titan?
three
mea:
Steel

enha

Sam]
With
Telat

[he

103

with any of the three methods. This means that even when the alkoxides have been
polymerized separately (Method 3) some mixing does occur. This is not surprising since
our previous study of titanium valerate ﬁlms indicated that polymerization occurs during
ﬁlm formation [21]. This technique is not useful for determining the extent of mixing.
The peaks are quite featureless and there is no unique curve ﬁtting solution to this
problem. There may be a range of bonding environments and therefore a range of 0 Is
binding energies present in these samples.

Ti-O-Sn moieties are undoubtedly present in these ﬁlms, but FTIR and XPS
cannot be used to determine the extent of mixing. AR-XPS is more indirect and relies on
the fact that frequently in heterogeneous mixtures, one component may surface segregate.
Those ﬁlms prepared by Method 3 are expected to produce the most inhomogeneous
mixtures and therefore, are the most likely to show a surface enhancement of either tin or
titanium. Figure 4.2 shows the result of AR-XPS study of 50:50 ﬁlms prepared using the
three different methods. Films prepared with Method 3 have a Ti/Sn ratio of 0.55 when
measured at the surface. This ratio increases to 0.80 as the sampling depth increases. The
steep gradient in the measurements suggests that there is a thin surface layer which is

enhanced in tin species.
AR-XPS depth proﬁles of ﬁlms prepared using Method 1 are also shown in Figure
4-2- First, the Ti/Sn ratio is relatively constant. ranging from 0.85 to 0.90 over the
Samng depth, indicating that these ﬁlms are uniform perpendicular to the ﬁlm surface
with only a slight surface enrichment of Sn. Second, the Ti/Sn ratio is close to 1, the
rclative concentration of the two alkoxides used to prepare the ﬁlms. Since the ﬁlms have

the expected bulk composition throughout the near surface region, this is another

Flgure t

indicati
{ht (let
measur
XPS rt
are mt

This St

and T;

104

 

 

Ti/Sn Atomic Ratio

 

 

 

 

Sampling Depth (A)

AR-XPS analysis of 50%Ti 50% Sn valerate ﬁlms prepared using Methods
1 and 2 I and using Method 3 O. The curves represent the best ﬁt to the

model.

Figure 4.2

indication that the ﬁlms are well mixed. In addition, there is no signiﬁcant difference in
the depth proﬁles of ﬁlms prepared from Methods 1 and 2. This shows that there is no
"measurable beneﬁt to reﬂuxing the alkoxides together before hydrolysis. Overall, AR-
XPS results indicate that ﬁlms prepared by mixing the alkoxides together before hydrolysis
are more homogeneous than those obtained from a mixture of the hydrolyzed solutions.
This suggests that the hydrolysis rates of the two carboxylated alkoxides may be similar.
These ﬁlms can be modeled most simply by considering that they are comprised of
two regions: an enriched surface layer and the bulk. The thickness of the surface layer ((1)

and Tr/Sn ratio for the surface layer (R,) can be estimated assuming exponential

105

attenuation of photoelectrons with depth [31]. The measured ratios (R) are related to the
analysis angle (0) by the following equation:
R = R.(1-e*‘"°°'°) + R,(e*’"‘°“’) 4.2

Estimates can be obtained by minimizing the sum of difference between the collected data
and values generated using this equation. R, is the bulk ratio and is assumed to be 1.
Results show that the Method 1 ﬁlms have a 15 A surface layer with R, = 0.85 (correlation
coefﬁcient: 0.97) and the Method 3 ﬁlms have a 10 A surface layer R. ='0.56 (correlation
coefﬁcient: 0.99).

The reason for the surface enhancement of Sn is not readily evident and the
identity of the Sn species at the surface is not clear, though the Sn 3d5,2 binding energy
(486.8 eV) is the same as that in the bulk and consistent with a Sn“ oxide. As determined
with AR-XPS, levels of hydroxyl groups and carboxylate groups are lower when
measured at ﬁlm surfaces than when a deeper cross-section of the ﬁlm is analyzed
Backbone oxygen (oxide) concentrations do not change with analysis depth. This means
that the surface is reduced but this may only reﬂect the sensitivity of the surface to x-ray
induced photoreduction [22] and not the true nature of the surface species. Possibly the
segregation is related to the relative concentration of uncondensed species in the ﬁlm
solutions. These species may rise and condense at the surface during the spin casting
process. In solutions prepared using Method 1, the alkoxides are intimately mixed and
hydrolyzed together so the relative concentration of uncondensed Sn and Ti species may
be similar. In the separately hydrolyzed solutions (Method 3) the concentrations may be

dissimilar, ultimately resulting in ﬁlms with an enhanced surface concentration of Sn.

106

Based upon these results, for the remainder of this study mixed oxide ﬁlms are
prepared using Method 1. Before examining the mixed TiOz-Snoz ﬁlms, the

characteristics of the single metal oxide ﬁlms will be established.

4.4.2 Effect of Calcination Temperature on Titanium Valerate Films.

Calcined titanium valerate ﬁlms contain approximately 1.85 backbone oxygen and
0.30 hydroxyl groups per Ti, as determined with XPS. Binding energies are 458.5 eV,
529.7 eV, and 531.6 eV for Ti 2pm, oxide 0 1s, and hydroxyl O ls peaks, respectively.
Film composition and binding energies do not change signiﬁcantly with calcination
temperature and are the same as measured for polycrystalline T102. Though ﬁlms are
visually uniform, traces of Si from the quartz substrate are evident in the XPS spectra of
ﬁlms calcined at 2600 °C. This indicates that the ﬁlms are cracked. The extent of
cracking, as reﬂected by the relative Si concentration, increases with calcination
temperature. Cracks may develop due to a difference in the expansion coefﬁcients of the
substrate and the ﬁlm or due to the formation of crystallites within the ﬁlm.

XRD analysis indicates that TiVal ﬁlms are amorphous and that crystallization
occurs as a result of calcination at temperatures 2400 °C. Figure 4.3 shows the XRD
patterns collected between 25° and 29° (20) for ﬁlms calcined between 250 °C and 1000
°C. Films calcined at 250 °C show no evidence of crystallinity. Those calcined at 400 °C
contain anatase crystallites with an average particle size of 10 nm. Anatase crystallites are
larger (average: 25 nm) after calcination at 600 °C and a weak rutile <110> peak indicates
that there are traces of rutile crystallites in this ﬁlm (average size: 8 nm). These peak

locations correspond to lattice constants “a” of 3.78 A (anatase) and 4.59 A (rutile).

107

 

 

 

 

r l l
Rutile
<1 10>

Anatase
<101>
(6)
”=7;
3
.33
5 (d)
E
(C)
(b)
a)
l l r ‘
24.0 26.0 28.0
Degrees (20)

Figure 4.3 “ XRD patterns of calcined TiVal ﬁlms: (a) TiC250, (b) TiC400, (c) riccoo,
(d) TiC800, and (e) TiClOOO.

108

These values are consistent with literature values [32]. Calcination at 800 °C results in a
ﬁlm which contains both anatase and rutile crystallites with average particle sizes of 25 nm
and 30 nm. Calcination at 1000 °C dramatically increases the intensity of the rutile <110>
peak by ~ 7 times and decreases that of the anatase <101> peak by one half, but particle
sizes are approximately the same as in the TiC800 ﬁlm.

Figure 4.4e shows a wider range of the XRD pattern for the TiClOOO ﬁlm. The
relative intensity of the <101>/<110> in polycrystalline rutile is 0.50 [33], however, in
these ﬁlms the ratio is 0.04. Additionally, there are no other peaks discernible in the ﬁlm
XRD pattern collected between 20° and 80° (29). This indicates that there is a strong
preferential orientation of the rutile crystallites with the <110> planes parallel to the ﬁlm
surface. In ﬁlms calcined at lower temperatures, the <110> rutile peak and the <101>
anatase peaks are the only ones present in the XRD patterns indicating strong preferential
orientation for crystallites in these ﬁlms as well.

Overall, the XRD results indicate anatase crystallites, largely orientated in the
<101> direction, are formed when ﬁlms are calcined at 2400 °C. Rutile forms at
temperatures 2600 °C with a preferential <110> orientation and it is the predominant
phase after calcination at 1000 °C. In ﬁlms calcined at 1000 °C, the number of rutile
crystallites dramatically increases but crystallite growth is not enhanced.

Film thicknesses and refractive indices are reported in Table 4.1. TiVal ﬁlms are
386 nm thick with a refractive index of 1.66. This is similar to the refractive index of ﬁlms
spin cast from acidic ethanol solutions of titanium ethoxide (1.61) reported by Exarhos

and Hess [17]. Titanium valerate ﬁlms collapse upon calcination, however they are not

completely collapsed unless calcined at 2400 °C. TiC250 ﬁlms are 239 nm thick with a

109

 

 

 

 

 

r l r r
Rutile
Anatase <1 10>
<101> Rutile
l <101> (e)
<1 10>

<110> <101> - (d)
=5
8

as <110> <110> <101> (c)
C
B
.5

<1 10> '
<101> (b)
<200>
Cassiterite <101>
<110> <200> (a)
l l l I
25 30 35 40
Degrees (29)
Figure 4.4 XRD patterns of ﬁlms calcined at 1000 °C: (a) SnC1000, (b) STC1000, (c)

EMC1000, (d) TSC1000, and (e)TiC1000.

110

Table 4.1 Thickness “(1” (nm) and refractive index “n” of ﬁlms as a function of
composition and calcination temperature. Standard deviations are < 5%
for thicknesses and < 3% for refractive indices.

 

 

 

 

 

 

 

Ti TS EM ST Sn

Fm“ ll rt 51 it :1 a it a it a

Valerate 386 354 346 335 303
1.66 1.65 1.60 1.63 157

250°C 239 237
2.04 1.75

. 400 .C 209 232 251 278 247
2.17 1.96 1.85 1.72 1.72

600°C ' 205 231 253 252 234
220 1.95 1.81 1.80 1.78

800 .C 205 229 , 244 261 244
2.19 1.97 1.78 1.73 1.72
-- 2.00 1.80 2.00 1.71

 

 

 

 

 

 

 

 

111

refractive index of 2.00 whereas those calcined at temperatures between 400 °C and 800
°C are ~205 nm thick with refractive indices of approximately 2.2. Accurate
measurements of the thickness and refractive indices of ﬁlms calcined at 1000 °C could
not be obtained using ellipsometry, probably because the ﬁlms were extensively fractured.

These results show that calcination at 400 °C leads to a collapse in the ﬁlms of
almost 50%. Dramatic collapse in sol-gel materials at such low temperatures is not
uncommon, especially after such a extensive calcination time (24 h) [34]. Generally, gels
collapse after the organic portions of the material have been burned away and through in
situ FTIR studies, we found that calcination at 250 °C is adequate to do this [35]. In
addition, increasing the calcination temperature above 400 °C does not cause additional
collapse or change the refractive index. Brinker and Harrington [36] found that the
refractive index of mixed 90% T102 - 10% SiOz ﬁlms reached the maximum value of 2.1
when calcined at 400 °C and interpreted this as indication that the ﬁlms are fully densiﬁed
at this temperature. This reasoning suggests that TiVal ﬁlms calcined at 2400 °C are also
fully densiﬁed. This is an advantage for coating materials but a disadvantage for sensor
and catalysis applications where high surface area is desired.

The refractive indices for these ﬁlms are less than that reported for pure
polycrystalline anatase or rutile (2.57 and 2.95, respectively [37]). This is commonly the
case for sol-gel thin ﬁlms [16] and may be due to the fact that these ﬁlms are comprised of
a mixture of crystalline and amorphous phases [17]. These ﬁlms collapse upon calcination
at much lower temperatures than needed to induce crystallization. Ritala et al. [19] .

proposed that crystallization may be retarded in a dense material and this may account for

112

the limited extent of crystallization in these ﬁlms. Highly crystalline materials are not
obtained unless the ﬁlms are calcined at 1000 °C.

TiVal ﬁlms have a band gap of 3.5 eV and calcination narrows the gap. The band
gap steadily decreases with calcination temperature from 3.4 eV for ﬁlms calcined at 250
°C to 3.1 eV for ﬁlms calcined at 800 °C. The band gap of the TiC1000 ﬁlm is 2.8 eV
(assuming ﬁlm thickness of 200 nm). The trend is consistent with the formation
crystallites. Band gaps are reported to be 3.2 eV and 3.0 eV for anatase and rutile,
respectively [38]. Values of 3.3-3.4 are commonly reported for TiOz thin ﬁlms containing
crystallites of either anatase, rutile, or a combination of both [14, 15].

In the ﬁlms calcined at temperatures between 400 and 800 °C there is a maximum
in the absorption curve at 4.85 eV (Figure 4.6e). We speculate that this is due to the
presence of anatase in these ﬁlms because this peak is present only in those ﬁlms which are
known to contain a predominant anatase phase; however, we have found no literature
precedent for this. The transmittance of the ﬁlms in the visible region is nearly 100% for
the T‘iVal ﬁlms and decreases from approximately 91% to 83% for ﬁlms calcined at

temperatures between 250 °C to 1000 °C, respectively.

4.4.3 Effect of Calcination Temperature on Tin Valerate Films.

The XPS results indicate that all calcined ﬁlms have compositions and binding
energies similar to that of polycrystalline SnOz. The oxide and hydroxyl concentrations
are 1.85 and 0.30 per Sn and binding energies are 486.7 eV, 530.6 eV, and 532.0 eV for
Sn 3dm, oxide 0 Is, and hydroxyl 0 Is peaks, respectively. Traces of Si from the quartz

substrate due to ﬁlm cracking are evident in ﬁlms calcined at 600 °C and higher. Si 2p

113

peaks are more intense in the spectra of ﬁlms calcined at higher temperatures, indicating
that the extent of cracking increases with calcination temperature. Films calcined between
250 °C and 600 °C are conductive [35] which indicates that though SnC600 ﬁlms contain
cracks, cracking is not extensive enough to render them completely discontinuous.

As determined with XRD, SnVal ﬁlms are amorphous. Crystallites are evident in
XRD patterns of ﬁlms calcined at temperatures 2 400 °C (Figure 4.5). Peaks are most
intense in the pattern of the ﬁlm calcined at 1000 °C (Figure 4.4a) and are characteristic of
cassiterite, a rutile form of tin oxide. Only lines due to the <110>, <101>, and <200>
planes are present in the XRD patterns between 20° and 40° (20). Calculation of the
lattice constants results in values of 4.74 A and 3.19 A for “a” and “c”, respectively, which
is consistent with values reported in the literature for polycrystalline cassiterite [39].
Relative intensities are different, however, and <101>/<110> and d00>/<110> values are
0.35 and 0.15 for the ﬁlms instead of 0.80 and 0.25, respectively, reported for the
powders [39]. This indicates that there is a preferred orientation for the <110> planes
parallel to the surface of the ﬁlms. though it is not as pronounced as in the calcined
titanium ﬁlms. The intensity of the <110> line increases and narrows with calcination
temperature (Figure 4.5). Average particle size increases from 4 nm in ﬁlms calcined at
400 °C up to 13 nm in ﬁlms calcined at 1000 °C.

As reported in Table 4.1, SnVal ﬁlms are 303 nm thick and have a refractive index
of 1.57. Calcination at 250 °C leads to a collapse of ﬁlms to ~240 nm and an increase in
the refractive index to 1.74. Additional calcination at up to 1000 °C does not lead to a
signiﬁcant change in ﬁlm thickness or refractive index. This indicates that SnVal ﬁlms are

densiﬁed at a very low temperature. Other researchers have reported refractive indices

114

 

 

 

 

r 1
Cassiterite
<1 10>
(e)
3 (d)
A?
8
8
E. (C)
(b)
(a)
l l ' I
24.0 26.0 28.0
Degrees (20)

Figure 4.5 XRD patterns of calcined SnVal ﬁlms: (a) SnC250. (b) SnC400, (c)
SnC600, (d) SnC800, and (e) SnC1000.

115

between 1.8 and 2.2 for polycrystalline SnO; ﬁlms [40, 41]. The low value measured for
ﬁlms in this study may reﬂect their largely amorphous nature. As indicated in the
discussion of the calcined titanium ﬁlms, densiﬁcation of ﬁlms at temperatures below the
crystallization temperature may inhibit the formation of crystallites.

SnVal ﬁlms may be thinner than TiVal ﬁlms because the solutions used to prepare
the SnVal ﬁlms have nearly a factor of 10 higher alcohol concentration than the TiVal
solutions. It is well-known that the thickness of dipcoated sol-gel ﬁlms can be controlled
by the addition of a solvent [34] and it may also be true for spin-coated ﬁlms. TiVal ﬁlms
collapse almost 50% when calcined at 400 °C compared with a 20% for SnVal ﬁlms.
TiVal ﬁlms are more highly carboxylated (1.2 valerate/1‘ i [21]) than SnVal ﬁlms (0.26
valerate/Sn [22]) and therefore it is not surprising that the loss of the organic components
of the ﬁlm which occurs upon calcination would have a more dramatic effect on the TiVal
ﬁlm

Preliminary results indicate that both the thickness and extent of carboxylation can
be controlled by dilution with isopropanol. A TiVal ﬁlm prepared from a solution with an
isopropanol concentration equal to that of a SnVal solution has approximately the same
valerate concentration and thickness as a SnVal ﬁlm. Unlike the SnVal ﬁlms, however,
the extent of collapse of the TiVal ﬁlm is much greater, about 35%. This suggests that
while the extent of compaction is a function of the carboxylate concentration, it is also a
function of the metal in the ﬁlm.

Analysis of the UV-Vis spectra of these ﬁlms shows that the optical band gap of

the valerate ﬁlm is 3.8 eV while all calcined ﬁlms have a band gap of approximately 2.8

116

eV. The band gap for cassiterite is reported to be 3.6 eV [42]. Other researchers [43, 44]
have found that amorphous tin oxide ﬁlms have smaller band gap energies than crystalline
ﬁlms and thus our low value (2.8 eV) may be another indication of the largely amorphous
nature of these ﬁlms. Melsheimer and Ziegler [44] also reported band gap energies of
~2.8 eV for their amorphous tin dioxide thin ﬁlms. The transmittance of the SnVal ﬁlms
and all of the calcined Sn ﬁlms is >97% in the visible range.

In summary, calcination of tin valerate ﬁlms leads to ﬁlm collapse, but other than
an increase in the extent of crystallization and the size of the crystallites, there is little

change in ﬁlm characteristics with calcination temperature.

4.4.4 Effect of Mixing and Calcination at 400 °C on Film Properties.

All mixed valerate ﬁlms are amorphous and have ﬂat AR-XPS depth proﬁles with
normalized Ti/Sn ratios of approximately 0.9. XPS binding energies for Sn 3d5,2 and Tr
2pm peaks are 486.8 eV and 458.5 eV, respectively. These values are similar to those
measured for the single metal ﬁlms and consistent with +4 oxidation states. In the mixed
valerate ﬁlms, binding energies for ligand (hydroxyl and carboxylate) 0 1s (531.8 eV) and
carboxylate C 1s (288.7 eV) do not change with composition and again are similar to
those measured for the single metal ﬁlms. The oxide 0 Is binding energies are 529.7 eV
and 530.6 eV for TiVal and SnVal ﬁlms, respectively, and intermediate values are
measured for the mixed ﬁlms (T SVal: 530.1 eV, EMVal: 530.2 eV, and STVal: 530.3
eV). The 0 1s peaks were ﬁtted assuming only one type of backbone oxygen.

The composition of the valerate ﬁlms is shown in Table 4.2. It can be seen that the

valerate concentrations increase and hydroxyl concentrations decrease from SnVal to

117

Table 4.2 Composition of Valerate Films: Ratios of Ligands and Backbone Oxygen
to Metal and Estimated Average Metal Oxidation State as Determined by

 

 

Quantitative XPS Analysis
. Valerate Hydroxyl BAQKDQDLQ Metal Oxidation
T'=S“ Metal Metal Metal State
100:0 1.2 -- 1.4 4.0
75:25 0.65 0.36 1.6 4.2
50:50 0.56 0.41 1.5 4.0
25:75 0.36 0.65 1.5 4.0
0:100 0.26 0.55 1.6 4.0

 

 

 

 

 

 

 

TiVal ﬁlms. Note that these trends are consistent with the increase in is0propanol
concentrations in the ﬁlm solutions. Backbone oxygen concentrations are approximately
the same in all ﬁlms.

In calcined mixed ﬁlms, the average oxide/metal and hydroxyl/metal ratios are 1.85
and 0.29, respectively, the same as in the calcined single metal ﬁlms. 0 ls and Ti 2pm
binding energies are unchanged by calcination. The Sn 3d5,2 binding energy is slightly
lower in calcined ﬁlms (486.6 eV), but still consistent with a +4 oxidation state. Though
both SnC400 and TiC400 ﬁlms contain crystallites, there is no evidence of crystallization
in any of the mixed ﬁlms after calcination at 400 °C. This may be an indication that the
metals are well-mixed in these ﬁlms or it may be that they mutually inhibit crystallization.
It is not uncommon for additives to change, either inhibit or enhance, the crystallization
behavior of oxides [20].

As with the single metal ﬁlms, all mixed ﬁlms collapse with calcination at 400 °C

and increasing the calcination temperature up to 800 °C does not cause signiﬁcant

118

additional compaction (Table 4.1). The extent of collapse does seem to be roughly
correlated with the Ti (and valerate, see Table 4.2) concentration the ﬁlms. TSVal ﬁlms
collapse 35%, EMVal ﬁlms 28%, and STVal ﬁlms 21%.

The refractive indices of the mixed ﬁlms are intermediate between those of the
single metal ﬁlms (Table 4.1). As with ﬁlm thickness, the refractive indices of the ﬁlms
change dramatically upon calcination at 400 °C but there is no signiﬁcant additional
change with calcination up to 800 °C. Refractive indices of the mixed valerate ﬁlms are all
close to 1.60. Values for calcined ﬁlms are 2.17, 1.96, 1.85, 1.72, and 1.72 for TiC400
TSC400, EMC400, STC400, and SnC400 ﬁlms, respectively. Results show that ﬁlms
with refractive indices between approximately 1.7 and 2.2 can be prepared by controlling
the relative Ti/Sn concentration.

The band gaps of the mixed valerate ﬁlms are between those of the single metal
ﬁlms (3.5 - 3.8 eV) and the transmittance of all valerate ﬁlms is >97% in the visible region
of the spectrum. Like the Ti ﬁlms, the band gaps of the mixed ﬁlms tend to decrease with
calcination temperature (Table 4.3). Changes in the band gap are probably related to
phase changes and changes in crystallite concentration (vide infra).

Absorption coefﬁcients are higher for titanium ﬁlms than for tin ﬁlms and mixed
ﬁlms have intermediate values. This can be seen in the UV—Vis spectra of the ﬁlms
calcined at 400 °C presented in Figure 4.6. Calcination does not change absorption
coefﬁcients in the UV region signiﬁcantly, but the transmittance of the mixed ﬁlms at
visible wavelength tends to decrease as ﬁlms are calcined at higher temperatures. In

general, the transmittance decreases from approximately 95% to 85% as ﬁlms are calcined

119

Table 4.3 Band Gap Energies for Films as a Function of Calcination Temperature and

 

 

 

 

 

 

 

 

 

 

 

Composition.

Film Ti TS EM ST Sn
Valerate 3.5 3.5 3.6 3.8 3.8
250 °C 3.4 2.8
400 °C 3.3 3.3 3.3 3.4 2.8
600 °C 3.3 3.2 3.1 3.5 2.9
800 °C 3.1 3.0 2.7 3.0 2.8
1000 °C 2.8 3.0 3.0 2.5 3.0

50 (e)

l
40 - (d)

 

Absorption coefficient (um-‘1)

 

 

 

 

 

 

 

200 250 300 350 400 450 500
Wavelength (nm)

Figure 4.6 UV-Vis spectra of ﬁlms calcined at 400 °C: (a) SnC400, (b) STC400, (c)
EMC400, (d) TSC400, and (e) TiC400.

120

from 400 °C to 1000 °C. Reﬂections at phase interfaces, due to partial crystallization,
may be responsible for decreases in measured transmittance with increased calcination
temperature.

4.4.5 Effect of Mixing and Calcination at 600 °C to 1000 °C.

As with the single metal ﬁlms, all mixed ﬁlms calcined at 2600 °C contain cracks.
Si 2p peak intensity, as determined with XPS, is very low in the C600 ﬁlms and increases
with calcination temperature.

XRD patterns of mixed metal oxide ﬁlms which have been calcined at 600 °C have
broad, low intensity peaks due to small crystallites (Figures 4.7b, 4.9b, and 4.11b).
Average particle sizes are less than 11 nm in all mixtures. These lines become more
intense and narrower as the calcination temperature increases, but line positions do not
change appreciably. This indicates that though crystallite size increases with calcination
temperature and perhaps the number of crystallites also increases, no new crystalline
phases develop. In all ﬁlms there is no evidence of pure tin oxide (cassiterite) or titanium
oxide (anatase and rutile) phases seen in the single metal oxide ﬁlms.

Crystalline Mixed Oxide Phases. There are two distinct phases present in the
mixed oxide ﬁlms. These can be seen most clearly in the XRD patterns of ﬁlms calcined at
1000 °C (Figure 4.4). The ﬁrst is a Sn-rich rutile phase, similar in nature to cassiterite.
The same lines are present in the STC1000 ﬁlm pattern (Figure 4.4b) as in that of the
SnC1000 ﬁlm (Figure 4.4a) but the lines are shifted to slightly larger angles (20). The
lattice constants “a” and “c” are 4.70 A and 3.13 A, respectively for the mixed phase,

which are slightly smaller than those measured for the cassiterite crystallites (a = 4.74 A

121

and c = 3.19 A). This contraction of the unit cell is consistent with the substitution of Ti
for Sn in the cassiterite lattice. Considering that lattice constants for the TiOz rutile phase
are 4.59 A and 2.96 A, the approximate relative concentration of Ti in the lattice can be
estimated assuming a linear relationship between the composition and the lattice constants.
Results indicate a 25% Ti - 75% Sn mixture in these crystallites, the same relative
concentration used to prepare the ﬁlms.

The relative intensities of the lines indicate that the Sn-rich rutile phase has an even
stronger <110> preferential orientation than the cassiterite crystallites in pure tin oxide
ﬁlms. Also crystallites are larger in the STC1000 ﬁlm (20 run) than in the SnClOOO ﬁlm
(13 nm) and the <110> peak is more than twice as intense. This suggests that the Sn-rich
rutile phase is very stable.

The second phase seen in the mixed oxide ﬁlms is a Ti-rich phase similar to rutile
and is most apparent in the XRD pattern of the TSC1000 ﬁlm (Figure 4.4d). There are
only two lines present due to this phase, an intense <110> line and a very low intensity
<101> line, at slightly smaller angles than seen in the TrC1000 ﬁlm pattern (Figure 4.4e).
Like the rutile crystallites in the titanium oxide ﬁlms, the Tr-rich rutile phase has a very .
strong preferential orientation in the <110> direction. Calculated lattice constants for this
phase are a = 4.61 A and c = 2.98 A, slightly larger than that for rutile T101. This is
consistent with the incorporation of Sn in the lattice and this phase is estimated to have a
composition which is approximately 90% Ti and10% Sn.

Sn-rich rutile crystallites develop upon calcination of ST, EM, and TS mixed oxide

ﬁlms and the lattice constants are the same in all ﬁlms. The Ti-rich rutile phase is present

122

in TS and EM mixtures but is not evident in the ST ﬁlms, even after calcination at 1000
°C. The lattice constant “a” for this phase is the same in T8 and EM ﬁlms. The intensity
of the <101> line (if present) is too low to be measured in the latter ﬁlms, so lattice
constant “c” in these ﬁlms cannot be calculated.

Our results do not suggest the formation of a continuous series of crystalline solid
solutions with compositions related to the Ti/Sn ratio used to prepare the ﬁlms. We see
instead, the presence of two distinct phases which have the same composition in all of the
mixed ﬁlms. This result is different than that of other researchers who have prepared
Tioz-SnO; mixtures. Nakabashi et al.[3] studied TiOz-Snoz powders produced by
coprecipitation of alkoxides and found that solid solutions with variable composition were
formed. As evidence of this, they show a nearly linear decrease in the rutile lattice
constant “c” with increasing TiO; concentration. Values drop from 3.19 for the pure
SnOz powder down to 3.05 for powders which contain only 25 % SnOz.

Takahashi and Wada [4] prepared ﬁlms with CVD and also showed that the lattice -
constants for rutile TiOz and 81102 crystallites are a function of composition, but they
found that lattice constants for each phase reached limiting values as the concentration of
each oxide decreased. For the TiO; phase, lattice constant “a” ranged between 4.61 A and
4.62 A for ﬁlms with TiOz molar percents between 75% and 25%, respectively, and “c”
was ~2.98 A in this concentration range. These values are the same as those of our Ti-
rich phase (4.61 A and 2.98 A) as calculated from the TS ﬁlms. Lattice constants for
Takahashi and Wada’s SnOz phase range from 4.73 A to 4.72 A for “a” in ﬁlms prepared

with 75% to 25% 81102 and have a constant “c” of 3.17 A. These values are larger than

123

those calculated for our Sn-rich rutile phase (a = 4.70 A and c = 3.13 A) over this
concentration range. Takahashi and Wada’s ﬁlms do not attain the high level of Ti
incorporation in the SnOz phase that is present in our ﬁlms.

Finally, Yoshimura et al. [5] studied XRD patterns of 70%Ti02-30%Sn02 ﬁlms
calcined up to 600 °C. They found that SnOz was not a solid solution in TiOz. These
ﬁlms were prepared from alkoxides using sol-gel techniques; however, Ti(OPr‘)4 was
hydrolyzed before the addition of the tin alkoxide. This procedure may lead to a more
heterogeneous mixture than the methods of other researchers.

There is no evidence of an anatase phase in any of the mixed ﬁlms, though it was
the predominant phase in the TiVal ﬁlms calcined between 400 and 800 °C. Suwa [10]
found that anatase formation was suppressed when the SnOz concentration was 220%.
Other researchers have also noted that incorporation of SRO: suppressed the formation of
anatase [2, 4]. Yoshimura et al. [5] reported the presence of anatase in their mixed oxide
ﬁlms. This may be further evidence of the heterogeneous nature of their materials.

Now that the identity of the phases in the mixed ﬁlms has been established, the
evolutiOn of the ﬁlms with increasing calcination temperature can be examined.
Calcination of mixtures often results in surface segregation of one component. Surface
segregation would have an impact on catalysis and gas sensing applications since surface
reactions are fundamental to their functioning. In this ﬁnal part of this paper, phase
changes are correlated with ﬁlm properties and with surface composition as determined

with AR-XPS.

124

Surface Segregation and Crystallization. In the simplest scenarios, there are
two ways in which the surface may become enriched in one of the metals. A surface
enhancement may be seen if crystals form with a different Ti/Sn ratio than in the starting
amorphous phase. This is expected to lead to ﬂat depth proﬁles, elevated in either Sn or
Ti, if the crystals are either larger than 50 A and form at the surface, or if they are not
within the sampling depth. Alternatively, if one of the metals migrates to the surface
forming a surface layer, measured ratios would change with sampling depth, provided that
the surface layer is much less than 50 A. Estimates of layer thickness and composition
are made using equation 4.2 as previously described. All ﬁts reported have correlation
coefﬁcients of 2 0.97.

As indicated previously, mixed valerate ﬁlms and ﬁlms calcined at 400 °C are
amorphous. In all three mixtures, valerate ﬁlms have a normalized Ti/Sn atomic ratio of
~0.9 and 400 °C calcination moves this ratio close to 1. Ratio changes with sampling
depth (Figures 4.8, 4.10 and 4.12) may indicate that the 400 °C calcined ﬁlms have a slight
surface segregation of Ti but overall these ﬁlms and the valerate ﬁlms appear to be quite
homogeneously mixed. All mixed ﬁlms develop crystallites when calcined at 2600 °C.
The crystallinity and surface composition of these ﬁlms depends upon the initial Ti/Sn
ratio and the calcination temperature.

25% Ti 75% Sn (ST) Films. The Sn-rich rutile phase is the only crystalline phase
evident in the ST ﬁlms (Figure 4.7). Average particle sizes increase with calcination
temperature with values of 7 nm, 9 nm, and 20 nm for ﬁlms calcined at 600 °C, 800 °C,

and 1000 °C, respectively. Peak intensities double ﬁom 600 °C to 800 °C and more than

125

triple from 800 °C to 1000 °C. The dramatic changes in the crystallinity of the ﬁlms
between 800 °C and 1000 °C may be responsible for the dr0p in the band gap from 3.0 to
2.5 eV (Table 4.3) and the increase in the refractive index from 1.72 to 2.00 (Table 4.1).
The ﬁlms also collapse about 40% more when calcined at 1000 °C than at lower
temperatures (I' able 4.1).

Figure 4.8 shows the results of the AR-XPS analysis of the ST ﬁlms. Calcination
at 600 °C leads to an surface enhancement of Tl which is 1.7 times the ﬁlm ratio. The
value steadily decreases almost 20% over the range of sampling depths. This decrease and
the low level of crystallization in these ﬁlms suggests that Tr has migrated to the surface
forming a Ti rich surface layer. The thickness of the surface layer is 11 A as calculated
using Equation 4.2. The normalized Ti/Sn ratio is 1.65 for this layer which corresponds to
an actual Ti/Sn ratio of 0.55. The model ﬁts this data quite well which suggests that there
is not a depletion region below the surface layer. Ti mobility and calcination time may be
adequate to eliminate any depletion which my have been created by the surface
segregation.

Depth proﬁles for the STC800 ﬁlms are ﬂatter, with average normalized Tl/Sn
ratios of about 1.3. The drop in Ti concentration at the surface and the ﬂat proﬁle
suggests that there is no Ti-rich surface layer on the STC800 films. The elevated Ti level is
consistent with a depletion of Sn in the amorphous phase due to crystallization, assuming
that the crystals are not forming at the surface. This also suggests that crystallites actually
have a Sn concentration >75%, the concentration estimated from the lattice constants.

Calcination at 1000 °C leads to a jump in the surface Ti concentration up to over

two times the expected ratio but also to ratios which drop 25% as the sampling depth is

126

 

 

 

 

l I I
<1 10>
32‘
S d)
a ‘
E:
(C)
(b)
a)
l l I
24.0 26.0 28.0
Degrees (20)

Figure 4.7 XRD patterns of calcined STVal ﬁlms: (a) STC400, (b) STC600, (c)
STC800, and (d) STC1000.

127

 

 

 

 

2.2
P o
g 2.0 -
a: ’ 0
o 1.8 -
' O
E a o
< 1.6 - o
5, . ‘ ‘ o
F A
E" 1.4 r ‘
B . 8 e e A
.E O Q .
To 1.2 -
E
z 10 - o ......... e ......... g ........ ennui--.
’ I I l I l I
0.8 e
L L l a I L J r l 4 l
0 10 20 30 40 50
Sampling Depth (A)

Figure 4.8 ' AR-XPS analysis of ISTVal, .STC400, 4STC600, OSTC800. and
OSTC1000 ﬁlms.

128

increased. These elevated and rapidly decreasing levels imply not only a Ti rich surface
layer, as in the STC600 ﬁlms, but perhaps also an amorphous phase rich in Ti as in the
STC800 ﬁlms. Applying the single layer model and allowing the bulk concentration to
vary indicates that the normalized surface and bulk Ti/Sn ratios are 2.0 (actual Ti/Sn:
0.67) and 1.2 (actual Tr/Sn: 0.40), respectively. The surface layer thickness is estimated to
be 9 A.

50%Ti 50% Sn (EM) Films. Only one low intensity peak is present in the XRD
patterns of the EMC600 ﬁlms (Figure 4.9). Whether this is a merging of the <110> peaks
of the two mixed rutile phases or a single broad peak with an intermediate position (a =
4.65 A) cannot be determined because the peak intensity is too low. Films calcined at 800
°C and 1000 °C contain both mixed phases. The average particles sizes are ~10 nm in the
EMC800 ﬁlms and 18 nm and 24 nm for the Sn-rich and Ti-rich rutile crystallites,
respectively, in the EMC1000 ﬁlms. As with the ST ﬁlms, the intensity of the <110> peak
of the Sn-rich rutile phase more than triples with the increase in calcination temperature
from 800 °C to 1000 °C. Despite the difference in crystallinity, the band gap energy, ﬁlm
thickness (250 nm), and refractive index (1.8) are not signiﬁcantly different between the
EMC800 and EMC1000 ﬁlms (Table 4.1).

As with the ST ﬁlms, all calcined EM ﬁlms have a surface enhancement of Ti and
like the STC600 ﬁlm, the EMC600 ﬁlm may have a Ti-rich surface layer. This can be seen
in Figure 4.10 where the Ti/Sn ratio is 1.5 times greater than the expected ratio and it
drops about 20% over the range of sampling depths. Calculations show that this could be

the result of a 6 A surface layer with a Ti/Sn ratio of 1.5.

129

 

 

 

 

l l V
<110>
<110>
1111 (d)
3
.2»
8
g (C)
(b)
(a)
L I I
24.0 26.0 28.0

Degrees (20)

Figure 4.9 XRD patterns of calcined EMVal ﬁlms: (a) EMC400, (b) EMC600, (c)
EMC800, and (d) EMC1000.

130

 

2.2

2.0 -

A
1.4:- . ‘ x .
1.2- 9 g g 0 8
O

1.0 - ....................................... 1...

Normalized Ti/Sn Atomic Ratio

0.8 -

 

)-
l-
l»

 

 

Sampling Depth (A)

Figure 4.10 AR-XPS analysis of IEMVal, .EMC400, AEMC600, OEMC800, and
OEMC1000 ﬁlms.

131

Films calcined at 800 °C and 1000 °C have similar concentration proﬁles. Both
are ﬂat with a relative Ti/Sn concentration of ~1.3 throughout the surface region. The
enhancement of Ti is not as great in these ﬁlms as in the ST ﬁlms, probably because both
phases are crystallizing, leading to less of a Sn depletion in the amorphous phase. Ti is
enhanced because the predominant crystalline phase is the Sn-rich one. As with the ST
ﬁlms, this implies that crystallites are not forming at the ﬁlm surface.

75% Ti 25% Sn (TS) Films. XRD pattems' for the TS ﬁlms are shown in Figure
4.11. Only the Ti-rich rutile phase is evident in the ﬁlms calcined at 600 °C and 800 °C.
Average particle sizes increase from 11 nm up to 14 nm with a tripling of the <110> peak
intensity over this temperature range. Films calcined at 1000 °C contain both mixed rutile
phases. Both particle size and intensity of the <110> peak of the Ti-rich rutile phase
double upon increasing the calcination temperature from 800 °C to 1000 °C. Sn—rich
rutile crystallites average about 21 nm. Like the EM ﬁlms, the band gap energy, ﬁlm
thickness (230 nm), and refractive index (2.0) are not signiﬁcantly different between the
TSC800 and TSC1000 ﬁlms, despite the difference in crystallinity.

None of the ﬁlms calcined at _>. 600 °C show any evidence of a Ti rich surface layer
(Figure 4.12). The depth proﬁles are ﬂat and the relative concentration of Ti/Sn is the
same as it is in the uncalcined ﬁlms. There may not be a strong driving force for T1 to
form a surface layer, since it is the majority component of the ﬁlms.

Crystallization. depletes the amorphous phase of the predominant component of the
crystallites. In the case of the TSC600 and TSC800 ﬁlms, Ti should be depleted in the

amorphous phase and this may be reﬂected in the Ti/Sn ratios. Apparently the crystallites

132

 

 

 

 

l I I
<1 10>
<1 10>
((0
”=3
3
£9
8
E (C)
(b) '
(a)
l l 1
24.0 26.0 28.0
Degrees (20)

Figure 4.11 XRD patterns of calcined TSVal films: (a) TSC400, (b) TSC600, (c)
TSC800, and (d) TSC1000.

133

 

 

 

 

2.2

,2 2.0 -

e

c:

.2 1.8 -

s

8 P

<3 1.6-

t:

g

E- 1.4-

8

.s

"g 1.2-

C " ’ o e

Z 1.0- 2 ------------------------------- o ----- em.

.. i. 2 i r a
J r J 4 l r l 1 l r l
0 10 20 30 40 50
Sampling Depth (A)

Figure 4.12 AR-XPS analysis of ITSVal, OTSC400, ATSC600, OTSC800, and
OTSC1000 ﬁlms.

134

are fairly homogeneously distributed in the ﬁlms since there is no signiﬁcant shift in the
vertical distribution of Sn and Ti in the near surface region of the ﬁlms upon
crystallization. Both phases are present in the TSC1000 ﬁlms and these samples also

show little surface enrichment.

4.5 Concluding Remarks

In all calcined ﬁlms, if a surface layer is formed, it is always enriched in titanium
oxide. Our results show that titanium is mobile in ﬁlms even at 400 °C and tends to
surface segregate when ﬁlms are calcined. According to Seah [45], surface segregation of
a component in a mixture is the rule rather than the exception. Lowering the surface free
energy is the driving force for the segregation and mobility generally increases with
temperature. In relation to this and as a ﬁnal comment on the study of methods presented
at the beginning of this paper, EM ﬁlms prepared using either Method 1 or Method 3 are
indistinguishable with our analysis techniques when they are calcined at temperatures 2
600 °C. This indicates that at these temperatures, Ti is mobile enough for the ﬁlms to
reach the same (apparently) thermodynamically stable conﬁguration, despite the initial
surface enhancement of Sn present in ﬁlms prepared with Method 3.

We have shown through this study that homogeneously mixed metal oxide ﬁlms
can be prepared using our modiﬁed sol-gel approach. These materials can be calcined at
temperatures up to 400 °C without signiﬁcant crystallization, cracking or surface
segregation occurring. The ﬁlms are highly transparent to visible light, are completely

densiﬁed, and can be prepared with a range of refractive indices.

135

Other researchers have found that Sn-Ti oxide materials have interesting catalytic
and gas sensing properties [2-8]. Unfortunately for semiconductor gas sensing
applications, preliminary results indicate that while tin oxide ﬁlms calcined between 250
°C and 600 °C are conductive, titanium oxide and mixed oxide ﬁlms are not. This is
surprising since other researchers have shown that TiOz and mixed TiOz-Snoz materials
can be conductive [4, 5]. It is known that conductivity in wide band gap semiconductors
is due to the presence of defects within the gap and preliminary results show that H;
reduced TiOz ﬁlms are conductive. Further research is needed to determine if the mixed

TiOz-Snoz ﬁlms can also be made conductive through a reductive treatment

4.6

10.

11.

12.

13.

14.

15.

16.

17.

136

References

Hubert-Pfalzgraf, L.G. New J. Chem. 1987, 11, 663.
Itoh, M.; Hattori, H.; Tanabe, K. J. Catal. 1976, 43, 192.

N akabayashi, H.; Nishiwaki, K.; Kakuta, N.; Ueno, A. Nippon K agaku K aishi
1991. 1, 13.

Takahashi, Y.; Wada, Y. Yogyo-Kyokai-Shi 1987, 95(9), 864.

Yoshimura, N.; Sato, S.; Itoi, M.; Taguchi, H. Sozai Busseigaku Zasshi 1990, 3,
47.

Chung, W.-Y.; Lee, D.-D.; Sohn, B.-K. Thin Solid Films 1992, 221, 304. Chung,
W.-Y.; Lee, D.-D.; Choi, D.-H. Sens. Actuators B 1993, 13-14, 517.

Nakajima, A. J. Mater. Sci. Lett. 1993, 12, 1778.

Maddalena, A.; Dal Maschio, R.; Dire, S.; Raccanelli, A. J. Non-Cryst. Solids
1990. 121, 365.

Hubert-Pfalzgraf, L. G. New J. Chem. 1987, 11(10), 663.

Suwa, Y.; Kato, Y.; Hirano, S.; Naka, S. 1982, 31, 955.

Mackenzie, J .D. in Ultrastructure Processing of Glasses, Ceramics, and
Composites, Hench, L.L.; Ulrich, D.R., Eds.; John Wiley & Sons: New York,
1984,p.15.

Lee, W. G.; Woo, S.I.; Kim, J. C.; Choi, S.H.; Oh, K. H. Thin Solid Films 1994,
237. 105.

Battiston, G. A.; Gerbasi, R.; Porchia, M.; Marigo, A. Thin Solid Films 1994, 239,
186.

Meng, L.; Andritschky, M.; dos Santos, M. P. Thin Solid Films 1993, 223, 242.

Radecka, M.; Zakrzewska, K.; Cztemastek, H.; Stapinski, T. Appl. Surf. Sci.
1993, 65/66, 227.

Langlet, M.; Walz, 0.; Marage, P.; Joubert, J. c. Thin Solid Films 1992, 221, 44.

Exarhos, G.J.; Hess, N. J. Thin Solid Films 1992, 220, 254.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

137

Ritala, M.; Leskela, M.; Nykanen, E.; Soininen, P.; Niinistb Thin Solid Films
1993, 225, 288.

Ritala, M.; Leskelii, M.; Niinisto, L.; Haussalo, P. Chem. Mater. 1993, 5, 1174.
Eskelinen, P. J. Solid State Chem. 1993, 106, 213.

Severin, K. G.; Ledford, J. 8.; Torgerson, B. A.; Berglund, K. A. Chem. Mater.
1994, 6, 890.

Severin, K.G.; Ledford, J. S. Langmuir 1995, 11, 2156.
Re, N. J. Non-Cryst. Solids 1992, 142, 1.

Babonneau, F.; Doeuff, S.; Leaustic, A.; Sanchez, C.; Cartier, C.; Verdaguer, M.
Inorg. Chem. 1988, 27, 3166.

Hampden-Smith, M.J.; Wark, T.A.; Rheingold, A.; Huffman, J. C. Can. J. Chem.
1991, 69, 121.

Mehrotra, R.C.; Mehrotra, A. Inorg. Chim. Acta Rev. 1971, 5, 127.

Software provided by Dr. Andrew Proctor, University of Pittsburgh, Pittsburgh,
PA.

Seah, M. P.; Dench, W.A. Surf. Interface Anal. 1979, 1, 2.
Scherrer, P. Go'tt. Nachr. 1918, 2, 98.

Mott, N. F.; Davis, E. A. Electronic Processes in Non-Crystalline Materials;
Clarendon Press: Oxford, 1 97 1.

Practical Surface Analysis, Vol. I , Briggs, D.; Seah, M.P., Eds.; John Wiley &
Sons, Inc.: New York, 1990.

Tang, H.; Levy, F.; Berger, H.; Schmid, P.E. Phys. Rev. B 1995, 52(11), 7771.
National Bureau of Standards, Mono. 25, Sec. 7, 1969, 83.

Brinker, C.J.; Scherrer, G. Sol-Gel Science, the Physics and Chemistry of Sol—Gel
Processing; Academic Press: San Diego, 1990.

Severin, K.G.; Ledford, J .S. unpublished results

 

 

   

4

t. .1 as} .. E... 2 3.1.9.2....313... 41...? n

 

36.
37.

38.
39.

40.

41.
42.

43.

45.

138

Brinker, C.J.; Harrington, M.S. Solar Energy Materials 1981, 5 , 159.
Samsonov, G. V. The Oxide Handbook; IFI-Plenum: New York, 1982.

Kruczynski, L.; Gesser, H. D.; Turner, C. W.; Speers, E. A. Nature 1981, 291,
399.

Swanson; Tatge NBS Circular 539(1), 1953, 54.

Lousa, A.; Gimeno, S.; Marti, J. Vacuum 1994, 45(10/11), 1143. Viirola, H.;
Niinistb, L. Thin Solid Films 1994, 249, 144. Melsheimer, J.; Ziegler, D. Thin
Solid Films 1983, 109, 71.

Meng, L.; dos Santos, M. P. Thin Solid Films 1994, 237, 112.

Agekyan, V.T. Phys. Status Solidi(a) 1975, 43, 11.

Park, S.-S.; Mackenzie, J.D. Thin Solid Films 1995, 258, 268.

Melsheimer, J.; Ziegler, D. Thin Solid Films 1985, 129, 35.

Seah, M. P. in Practical Surface Analysis, Vol. I, Briggs, D.; Seah, M.P., Eds.;
John Wiley & Sons, Inc.: New York, 1990, p.311.

Chapter 5

Effect of Preparation Temperature on Tin Oxide Film Reactivity

5.1 Abstract

Tin oxide ﬁlms are prepared by oxidizing sol-gel derived tin valerate ﬁlms at room
temperature, 200 °C, 400 °C, and 600 °C. Their structures, as determined with X-ray
diffraction and X-ray photoelectron spectroscopy (XPS), range from completely
amorphous (ﬁlms‘oxidized by a room temperature hydrogen peroxide treatment (SnP)) to
partially crystalline (films oxidized by calcination in air at 600 °c (SnC600)). The
reactivities of these ﬁlms and a polycrystalline tin oxide powder are evaluated using XPS
analyses before and after in situ H2 reduction. Lattice oxygen (oxide) and hydroxyl
groups are removed from these materials when they are heated under a ﬂow of H2. At
temperatures 5 300 °C, no metal is formed and the oxide/tin ratio decreases with
increasing reduction temperature. Reduction at temperatures >300 °C results in the
formation of tin metal (IS-tin). The degree to which ﬁlms are reduced by a 300 °C
reductive treatment is dependent upon the ﬁlm structure. SnP ﬁlms are highly reactive and
are reduced to largely Sn2+ oxide materials by a 300 °C H2 treatment, as indicated by
surface stoichiometry and XPS valence band spectra. SnC600 ﬁlms, at the other extreme,
are less reactive than the polycrystalline tin oxide powder. SnC200 and SnC400 ﬁlms

have intermediate structures and reactivities. Reoxidation at 200 °C for 1 h. restores the

139

140

original composition and reactivity of all except the Sn? ﬁlms, provided the ﬁlms have not

been reduced to metal. Once metallic tin is formed, the ﬁlms are irreversibly modiﬁed.

5.2 Introduction

Many semiconductor gas sensors respond to changes in conductivity that result
from reactions occurring at the surface of a metal oxide semiconductor. Tin oxide is
commonly used for the detection of reducing gases such as H2, CO, ethanol, and
hydrocarbons. One of the problems with these sensors is that reactions on tin oxide
surfaces are not particularly selective. Potentially useful strategies for improving sensor
selectivity can be found by analogy to selective oxidation catalysis. Like semiconductor
gas sensors, catalyst performance depends upon the selectivity of reactions occurring on
metal oxide surfaces [1]. It is known that lattice oxygen reactivity is one of the most
important factors in the selectivity of catalysts [2- 4]. It rmy therefore play an important
role in the selectivity of semiconductor gas sensors. In order to test this hypothesis,
materials with a wide range of lattice oxygen reactivities are needed. The synthesis,
characterization, and reduction properties of these materials is the focus of this study.

The approach used in this study to tailor lattice oxygen reactivity is suggested by
the work of Yamazoe et al. [5]. They found that when the temperature was adequate for
a metal oxide surface to catalytically oxidize a particular gas, it exhibited a maximum in its
sensitivity for that gas. Since a different temperature is required to oxidize different gases,
this results in a selectivity which is a function of the operating temperature. In our study,
instead of increasing the operating temperature to increase the reactivity of the metal

oxide, the reactivity of oxide materials is modiﬁed by changing their structure. A range of

141

structures can be obtained through oxidation of amorphous ﬁlms at different temperatures.
The ﬁndings of Yamazoe et al. [5] suggest that these materials may be sensitive to
different gases.

In this study we have chosen to modify the reactivity of tin oxide since it is the
semiconductor most commonly used for the detection of reducing gases. Both
compressed polycrystalline tin oxide powder and ﬁlm substrates are used commercially
[1]. Use of thin ﬁlms is advantageous since they are inexpensive to manufacture and
compatible with the fabrication of microsensors. Thin ﬁlms of tin oxide are conventionally
prepared by reactive sputtering [6], evaporation [7], spray pyrolysis [8], or chemical vapor
deposition [9]. Tin oxide ﬁlms can also be prepared from tin alkoxides using sol-gel
techniques [10]. Previously we have shown that this method can be used to make
amorphous tin oxide ﬁlms which are uniform, transparent, and stable [11]. Sol-gel
synthesis is particularly advantageous because it can be used to produce materials with
high surface areas and conu'olled porosity [12, 13], properties which are important for
sensor applications. In this study, tin valerate ﬁlms are synthesized using sol-gel
techniques and are then oxidized at temperatures between room temperature and 600 °C
to produce tin oxide ﬁlms With a range of structures.

A method is needed to evaluate the lattice oxygen reactivity of the materials.
Typically, temperature programmed reduction (TPR) is used to characterize the reactivity
of catalysts. The information obtained using TPR is primarily of a kinetic nature, though
some information can be determined about the chemical nature and environment of species

in the solid, the number of species present, as well as the relative concenu'ations of these

142

Species [14]. These are inferred from changes in the uptake of H2 or in the production of
H20 as the temperature is increased.

For evaluation of oxide reactivity in this study, we have also chosen to reduce our
materials at elevated temperatures under a ﬂow of H2, but the reactivity is evaluated using
X-ray photoelectron spectroscopy (XPS). Materials are reduced at a constant
temperature between 200 °C and 400 °C in a reactor attached to the surface science
instrument. XPS analysis is used to determine stoichiometry and oxidation state of tin
before and after H; reduction. In this way changes in concentration and oxidation state of
surface species can be measured directly. These surface changes may be different than
those that occur in the bulk. This is important because it is surface reactions which are
responsible for sensor response.

.The presence of tin metal in a sample is easily identiﬁed using XPS. The binding
energy of Sn 3dm peak of tin metal (484.6 eV) is signiﬁcantly lower than that of tin oxide
(486.6 eV). Whether there is a chemical shift between the Sn 3d peaks of Sn4+ and Sn2+
oxides is a matter of active debate. Researchers have proposed that there is either no
chemical shift [15,16] or a shift of 0.2 eV [17, 18] or a shift of 0.7 eV [19]. This
ambiguity makes it impossible to determine the relative concentrations of Sn2+ and SnM
oxides in a sample from the Sn 3d peak. It has been established, however, that the
oxidation state of tin can be determined from the valence band spectra of tin oxides,
provided no metallic tin is present in the sample [15, 19-22]. The lowest binding energy
feature in the valence band spectrum of a Sn“ oxide is a non-bonding 0 2p derived band
at approximately 5 eV. In the spectrum of a Sn2+ oxide the lowest binding energy feature

is at 2 eV and due to photoelectrons from antibonding Sn 5s - 0 2p orbitals [19]. An

143

increase in the relative intensity of the peak at 2 eV is an indication of the formation of a
Snz” oxide material.

In the ﬁrst part of this study, the reduction pr0perties of polycrystalline tin oxide
powder are established Then, the effect of calcination temperature on the structure and
reactivity of tin oxide ﬁlms is evaluated and ﬁlm reactivities are compared to that of the
powder. There are two important issues. First, do ﬁlms prepared at different
temperatures have signiﬁcantly different oxide reactivities? If not, this is not a promising
approach to making selective sensor materials. Second, do ﬁlms undergo irreversible
restructuring upon reduction? Materials which are suitable for sensors must be able to
operate for long periods of time with a reproducible response to reducing gases. If the
ﬁlms undergo signiﬁcant restructuring when reduced, this is an indication they will be
unstable and unsuitable as senSor substrates. Film stability is evaluated through oxidation-
reduction cycling. Potentially stable substrates are those whose reactivity is unchanged by

repeated treatments.

5.3 Experimental

Materials. Tin(IV) isopropoxide 10% w/v in isopropanol (Chemat Technologies),
hydrogen peroxide, 30% (Baker analyzed reagent, J.T. Baker), and valeric acid (99+%,
Aldrich Chemical Company) were used without further puriﬁcation. Distilled, deionized
water was used for all syntheses. Tin(IV) oxide (99.995+%, Aldrich Chemical Company)
was pressed into pellets or suspended in absolute methanol (Photrex reagent, J .T. Baker)
and spray coated with an airbrush onto quartz slides. Pellets and spray coated slides were

calcined in dry air (80 cm3/min, medical grade, AGA Gas Co.) at 600°C for 4 h.

144

Film Preparation. The methods used to synthesize ﬁlms were based on
techniques developed previously [11, 23]. Films synthesized from Sn(OPri)a were
prepared using mixtures of 40 isopropanol : 9 valeric acid : 1.5 water : 1 Sn(OPri)a .
Reactions were carried out at room temperature in capped vials in a N2 purged glovebox.
Valerie acid was added to the alkoxide followed by water. A vortex mixer was used to
vigorously stir solutions following the addition of each reactant.

Freshly prepared solutions were dispersed on cleaned, acetone rinsed and dried
quartz slides, spun for ﬁve minutes, and air—dried overnight. Films were oxidized using
two different methods. Room temperature oxidation was accomplished by immersing “as
cast” ﬁlms in H202 for 4h, followed by rinsing with deionized water, and air—drying.
Oxidation at temperatures between 200 °C and 600 °C was carried out by calcining ﬁlms
in a tube furnace under a ﬂow of dry air (100 cm3/min, medical grade, AGA Gas Co.) at
the desired temperature for 24 h. Because a calcination temperature of at least 250 °C
was necessary for removal of the organic components of the ﬁlm [24], those ﬁlms calcined
at 200 °C were pretreated with H202 before calcination. Films were stored in a desiccator
before analysis or further treatment Unless otherwise noted, all ﬁlms prepared in this
study were visually transparent, colorless, and exhibited complete and uniform adhesion to
quartz substrates.

Films studied without further treatment are referred to as valerate or “as cast”
ﬁlms and designated with “Val” (i.e.: “as cast” ﬁlm = SnVal). Films which were heated in

air are referred to as calcined or tin oxide ﬁlms and designated with a “C” followed by the

 

 

145

calcination temperature (i.e.: a ﬁlm which has been calcined at 400 °C = SnC400). H202
treated SnVal ﬁlms are designated as SnP ﬁlms.

In Situ Heat Treatments. Oxidation and reduction of materials were carried out
in a reactor attached to the surface science instrument. The reactor was equipped with a
heater cartridge, a thermocouple, and temperature controller (Omega). Materials were
heated for l h. under a ﬂow (100 cm3/min) of the desired gas at atmospheric pressure.
Unless otherwise noted, reactions were complete within 1 h. and heating for additional
time did not cause further changes to samples. Reductions were carried out under pure H;
(99.95 %, AGA Gas Co.) which had been passed through water and oxygen ﬁlters. Unless
otherwise speciﬁed, reductions were carried out sequentially with increasing temperature
from 200 °C to 400 °C. Oxidation was performed under a dried mixture of 20% 02
(U.H.P., Purity Gas Co.) in He (99.995%, AGA Gas Co.). Samples were allowed to cool
to room temperature before the reactor was evacuated. The sample was then transferred
to the main chamber for XPS analysis without intervening exposure to ambient air. XRD
measurements were made within 24 h. of treatment.

X-ray Photoelectron Spectroscopy (XPS). Materials were analyzed with a VG
Microtech spectrometer using a Clam2 hemispherical analyzer. All XPS spectra were
collected using a Mg anode (1253.6 eV) operated at a power of 300 W (15 kV and 20 mA
emission current) with an analyzer pass energy of 50 eV. Four regions were scanned for
each ﬁlm. C Is, 0 1s, Sn 3d, and the valence band. Binding energies were referenced to
adventitious carbon (C ls = 284.6 eV) and were measured with a precision of i 0.1 eV.

Quantitative XPS calculations were perfomred using empirically derived sensitivity factors

 

 

 

146

previously determined in this laboratory [11,23]. XPS peaks were ﬁtted with 20%
Lorentzian-Gaussian mix Voigt functions using a non-linear least squares curve ﬁtting
program [25]. Values reported have a standard deviation of < 10%.

X-ray Diffraction (XRD). XRD patterns of calcined ﬁlms on quartz slides and tin
oxide spray coated on quartz were obtained with a Rigaku XRD diffractometer employing
Cu Kot radiation (ll: 1.541838 A). The X-ray was operated at 45 kV and 100 mA.
Diffraction patterns were collected using DD and DS slit widths of 1°. Peak widths and
locations were determined using a non-linear least squares curve ﬁtting program [25] and
assuming Gaussian line shapes. Particle sizes were calculated using the Scherrer equation

[26] assuming no line broadening due to stress.

5.4 Results and Discussion
5.4.1 H2 Reduction of Polycrystalline Tin Oxide.

The XPS spectrum of the polycrystalline tin oxide powder contains a single Sn
3dm peak at 486.6 eV and O ls peaks due to lattice oxygen (530.5 eV) and hydroxyl
oxygen (532.2 eV). These binding energies are typical for tin oxides [15-22]. The surface
of the powder is slightly hydroxylated with approximately 1.85 lattice oxygen (oxide) and
0.30 hydroxyl groups per tin. The valence band spectrum (Figure 5.1a) collected for this
powder is characteristic of a Sn4+ oxide [15, 19 - 22]. TheXRD pattern indicates that the
powder consists of cassiterite crystallites [27] with an average particle size of
approximately 200 nm (Figure 5.2a). Results of all analyses (including reduction studies)

were similar for tin oxide pellets and tin oxide spray coated onto quartz slides.

147

 

(C)

(d)

(C)

(b)

 

(a)

t l

 

 

 

 

l 1
10.0 5.0 0.0
Binding Energy (eV)

 

Figure 5.1 XPS valence band spectra of polycrystalline tin oxide pellet (a) initially;
after sequential H2 reductions for 1 h. each at (b) 200 °C, (c) 250 °C, and
(d) 300 °C; and (c) after reoxidation in 20% 02 (He) for 1 h. at 200 °C.

148

 

 

 

 

 

 

 

 

 

 

 

 

 

l l l l
<200>
' . <101>
33?
ID
ﬂ
8
'5 <110>
<101>
) <211>
<200> (a)
l l l
20 40 60 80

Degrees (20)

Figure 5.2 XRD pattern of polycrystalline Shop (a) initially, with most intense
cassiterite peaks labelled, and (b) after H; reduction at 350 °C for 0.5 h.
Peaks due to B-tin are designated with a dot (0) and major B-tin peaks are
identiﬁed.

149

Hz reduction of the tin oxide powder at 200 °C leads to a reduction in the lattice
oxygen concentration to 1.6 per tin, a decrease of almost 15%. The hydroxyl
concentration is decreased by one half. There is no evidence of tin metal in the reduced
powder and XPS binding energies have not shifted signiﬁcantly with reduction. The
oxidation state of tin which is inferred from the oxide and hydroxyl concentrations using
charge balance calculations [11] is approximately +3.35, suggesting that the surface has
been partially reduced to a Sn“ oxide. Approximately one third of the tin centers are in a
+2 oxidation state. The valence band spectrum (Figure 5.1b) still resembles that of a Sn“
oxide material, but there is peak present at 2 eV which can be attributed to electrons in
antibonding Sn 5s - 0 2p orbitals.

Subsequent reduction at 250 °C leads to a slight additional decrease in the oxide
and hydroxyl concentrations to 1.55 and 0.10 per tin, respectively. Again, there is no
change in the binding energies but there is an increase in the relative intensity of the peak
at 2 eV due to the presence of Sn2+ oxide (Figure 5.1c). Additionally, it becomes apparent
that there is an increase in the intensity of a feature at 9 eV which is due to electrons in
bonding Sn 53 ,- 0 2p orbitals [19].

Reduction at 300 °C causes a decrease in the oxide concentration to 1.5 per tin
with no additional decrease in the hydroxyl concentration. The binding energies of the Sn
3d and 0 1s peaks decrease by 0.15 eV. If the shift in the Sn 3d binding energy is due to
the increase in the Sn2+ concentration, the 0 1s peak associated with it is also decreasing
in binding energy. It is commonly noted that the separation between the Sn 3d and 0 Is

peaks is the same for Sn“ and Sn“ oxides and is equal to 43.9 eV [15], as seen in this

150

study. The charge balance calculation indicates that the surface is comprized of a 50:50
mixture of the two oxides. The valence band spectrum reﬂects the increased Sn“
concentration in the increased intensity of the peaks at 2 eV and 9 eV (Figure 5.2d).

After the polycrystalline tin oxide powder has been reduced under H; at 300 °C it
can be reoxidized by exposure to atmospheric oxygen at room temperature. Heating
under a 20% O; in He mixture at 200 °C for 1 h. also restores the material to its original
composition as determined with XPS. The valence band of the reoxidized pellet is shown
in Figure 5.1e and is indistinguishable from the initial valence band spectrum (Figure 5.1a).
Initial concentrations of oxide and hydroxyl groups are also restored by this treatment.

Thermodynamic calculations indicate that 81102 can be reduced to Sn metal by H;
at 200 °C provided that the partial pressure of water is less than «-10’2 torr. The only
sources of water are sample and reactor outgassing, and reaction products. Water is being
continuously swept out of the reactor by the ﬂow of H2 gas and its partial pressure is
undoubtably maintained below this level. Despite this, only a fraction of the lattice oxygen
at the surface of the polycrystalline tin oxide is susceptible to H2 reduction at 200 °C and
only slightly more is removed by a 300 °C reductive treatment. Additionally, no tin metal
is formed. The reason that only a limited amount of the lattice oxygen is extractable under
these conditions must be due to kinetic factors.

The mobility of lattice oxygen vacancies may limit the extent of reduction. In
order for lattice oxygen to react with H2, it has to be available at the surface. If lattice
oxygen vacancies are not highly mobile at these temperatures, the reduction rate would

slow down as vacancies accumulate at the surface. Kohl [28] indicates that lattice oxygen

151

vacancies in single crystal SnO: are mobile at 275 °C. Fattore et al. [4] studied the
oxidation of propene by polycrystalline SnO; in the absence of gaseous oxygen at
temperatures between 450 °C and 550 °C. They found that oxygen at the surface rapidly
oxidized propene but once this surface oxide was depleted, the reaction slowed down
dramatically. The limiting factor was determined to be the rate of lattice oxygen migration
to the surface (alternatively, lattice oxygen vacancy migration to the bulk). From these
results, it is certainly possible that at our reduction temperatures, between 200 °C and 300
°C, migration of lattice oxygen vacancies could play a role in limiting the oxide accessible
for H2 reduction.

Alternatively, the activation energy may be too high for the reduction of bulk SnOz
to Sn metal or even to SnO. In that case, the oxygen which reacts may not be normal
lattice oxygen but oxygen associated with a defect or non-stoichiometric site, such as a
coordinatively unsaturated "surface site. Oxygen in these sites is known to be more
reactive than bulk lattice oxygen [2]. Additionally, reduction studies of SnOz single
crystal (110) surfaces by Cox et al. [29] demonstrated that there were two types of lattice
oxygen at the surface: “bridging” and “in plane” oxygen. These were found to require
dramatically different temperatures for their removal under vacuum Therefore, limited
reaction can also be explained in terms of lattice oxygen with different activation energy
requirements for removal.

It should be noted that increasing the reaction time does not change the oxide
concentration at the surface. A constant oxide concentration is only possible if vacancies

are mobile enough to maintain this concentration as the reduction continues or completely

152

immobile so that reduction stops. In the former case, the lattice oxygen concentration
which is maintained is activation energy limited In the latter, the oxide concentration is
vacancy mobility limited. These two possibilities could be distinguished by analysis for
H20 in the exhaust gas stream. This would determine if the reduction process is
continuing or stops after this oxide concentration is attained.

H2 reduction at temperatures above 300 °C leads to the formation of crystalline tin
metal (B-tin ). The XRD pattern of the the polycrystalline tin oxide powder after H;
reduction at 350 °C for 1/2 h. is shown in Figure 5.2b. Even after reduction, cassiterite
crystallites are still evident. Peaks due to B-tin are also prominent and these crystallites
have an average size of approximately 250 am There is no evidence of the presence of
any other crystalline phases, however, the sample was exposed to air before and during the
XRD analysis. The size of the B-tin crystallites suggests that lattice oxygen vacancy
mobility is not a limiting factor at this temperature.

Like reduction at lower temperatures, additional reaction time does not lead to a
signiﬁcant change in the oxide/tin ratio of the polycrystalline tin oxide. A pellet reduced
for 0.5 h at 325 °C has an oxide/tin ratio of 1.45. Increasing the reduction temperature
from 325 °C up to 350 °C does not change this ratio, nor does doubling the reaction time.
The relative concentration of metal, however, does increase with either increased
treatment time or temperature. Since the oxide/tin ratio does not change with increases in
metal concentration, this suggests that either a disproportionation reaction is occurring or
that lattice oxygen vacancies are migrating into the bulk, maintaining a constant oxide

concentration at the surface as metal forms. As stated previously, a temperature of

153

approximately 275 °C is necessary for diffusion of oxygen vacancies into the bulk, so the
temperature is high enough for this process to occur.

Egdell [30] studied reduction of single crystal SnOz (110) surfaces using Ar+ ion
bombardment (2 eV) and electron beam reduction in an attempt to create controlled lattice
oxygen vacancies. They found that ion bombardment led to a SnO material but it was
difﬁcult to achieve or control intermediate stoichiometries. Irradiation with an elecu'on
beam (160 eV) resulted in the formation of tin metal. We have found that with H;
reduction, some control over the concentration of lattice oxygen vacancies can be exerted
at temperatures 5 300 °C without the formation of metal. The minimum oxide/tin ratio
obtainable is only 1.5, not 1, as achieved with ion bombardment. H2 reduction may be a

more effective method of introducing a limited number of lattice oxygen vacancies.

5.4.2 H; Reduction of Tin Oxide Films

Characteristics of tin oxide ﬁlms have been extensively explored previously [11,
31] and ﬁndings relevant to this discussion are brieﬂy reiterated here. As determined with
XRD, SnP ﬁlms and those calcined at temperatures <400 °C are amorphous. Films
calcined at 400 °C and 600 °C contain traces of cassiterite. The concentration and
average size of cassiterite crystallites is greater in the SnC600 ﬁlms (8 nm) than in the
SnC400 ﬁlms (4 nm). Those calcined at 600 °C are slightly cracked. This is inferred from
the presence of a low intensity Si 2p peak in the XPS spectrum of the SnC600 ﬁlms. Like
the polycrystalline tin oxide powder, the surfaces of all ﬁlms are hydroxylated and results
of charge balance calculations show that tin in all ﬁlms is in a +4 oxidation state.

Calcination at 250 °C is adequate to remove all organic components of the valerate ﬁlms.

154

Films calcined at 250 °C or higher are fully densiﬁed (collapsed 21% from the valerate
ﬁlm) with thicknesses of ~240 nm, as determined with ellipsometry. Films prepared at
room temperature through hydrogen peroxide treatment are partially collapsed (12%) with
thicknesses of 286 nm. Subsequent heating in air at 200 °C causes complete collapse to
the same thickness as the ﬁlm calcined at higher temperatures.

These oxide ﬁlms are similar in composition to polycrystalline tin oxide and, as
indicated by their valence band spectra, are eleetronically similar as well (Figure 5.3). The
valence band has the same general shape for all materials, though the peaks are broader in
the less structured materials. With an increase in calcination temperature, the materials
become more uniformly structured and as a result, the features in the spectrum become
more well deﬁned.

Reduction at Temperatures S 300 °C. As with the tin oxide powder, changes
due to H2 reduction are completed within 1 h. Subsequent reduction at the same
temperature does not lead to signiﬁcant changes in ﬁlm composition.

SnP Films. Reduction of SnP ﬁlms at 200 °C leads to a decrease in the oxide
concentration from 1.80 to 1.38 per tin (Figure 5.4). Like the powder, changes in the
valence band spectra are consistent with the formation of a Sn2+ oxide. Increasing the
reduction temperature to 250 °C decreases the oxide concentration to 1.28 per tin and
reduction at 300 °C to a material with a oxide/tin ratio of 1.18. The valence band
spectrum of an SnP ﬁlm after reduction at 300 °C is shown in Figure 5.5a. It is similar to
valence band spectra of tin oxides with high Sn2+ concentrations reported by other

researchers [15 , 19].

155

 

(C)

(d)
(C)
(b)

(3)

WW

1 1
10.0 5.0 0.0
Binding Energy (eV)

1 l

 

 

 

 

Figure 5.3 XPS valence band spectra of (a) SnP, (b) SnC200, (c) SnC250, (d)
SnC400, and (e) SnC600 ﬁlms.

156

 

 

 

 

1.8
. I
1.7 -
o
'33 1.6 - g i
a: .
.0
g 1.5 r- g
2 - i
t:
-— 1.4 -
t E
g .
o 1.3 -
1.2 -
A r l r J
200 250 300
Temperature (°C)

Figure 5.4 Changes in oxide concentration with reduction temperature as determined
with XPS for ISnP, OSnC200, OSnC400. and OSnC600 ﬁlms.

157

 

(C)

(d)

(C)

M

(b)

M

(3)

MI

1

 

 

 

l 1
10.0 5.0 0.0
Binding Energy (eV)

 

Figure 5.5 XPS valence band spectra of tin oxide materials after H2 reduction at 300
°C: (a) SnP, (b) SnC200, (c) SnC250, (d) SnC400, and (e) SnC600 ﬁlms.

158

Changes in the oxide concentration with sequential reductive - oxidative
treatments are shown in Figure 5.6. This study was undertaken to detemrine the amount
of irreversible restructuring the ﬁlms undergo when they are reduced at different
temperatures. This is especially important for SnP ﬁlms since they are being reduced at
temperatures (up to 300 °C) much higher than used to prepare the ﬁlm (room
temperature).

Initially the ﬁlm has an oxide/tin ratio of 1.75 and H2 reduction at 200 °C
decreases this ratio to 1.37 (Treatment 1 (T 1) Figure 5.6). Reoxididation of the ﬁlm at
200 °C for 1 h. (T2) restores the ﬁlm stoichiometry and valence band spectrum to those
characteristic of a Sn“ oxide material, however, the oxide concentration has increased
slightly. With reduction, the hydroxyl concentration decreases and when the ﬁlm is
reoxidized, some of the hydroxyl groups are replaced with lattice oxygen.

After reoxidation (T 2), the H2 reactivity of the ﬁlm has been altered. Reduction at
200 °C only decreases the oxide/tin ratio to 1.63 (T3); a reduction temperature of 250 °C
is necessary to obtain a 1.37 oxide/tin ratio (1‘ 4). This suggests that signiﬁcant
restructuring has occurred with the initial reductive and oxidative treatments at 200 °C.
Though the ﬁlm was at 200 °C for only 2h. it is possible that this was adequate to cause
the ﬁlm to collapse to the fully densiﬁed state found for SnC200 ﬁlms. This may be
responsible for the decreased reactivity.

The ﬁlm was then reoxidized (T5) and reduced at 200 °C (T6) and 250 °C (T7)
which resulted in oxide/tin ratios similar to those measured in the previous redox cycle.
Reduction at 300 °C (T 8) leads to additional loss of oxide and reoxidation (T9) to a Sn“

oxide. Increased oxide concentration after these treatments (1.97 oxide/tin) is balanced

159

 

2.0

1.8 - o

Oxide/Tin Atomic Ratio

1.2 -

 

 

l l 1 l l 1 l L l l l l

0 1 2 3 4 5 6 7 8 9 10 11
Sequential Treatments

 

Figure 5.6 Oxidation - reduction cycling of SnP ﬁlm: Elinitial ﬁlm; I200 °C, 0250
°C, and 0300 °C H2 reduction; and 0200 °C 20% 02(He) oxidation.
Oxide/tin atomic ratio is determined with XPS analysis after each
sequential in situ treatment.

160

by a lower hydroxyl concentration. The calculated Sn oxidation state is approximately the
same (+4.1) after all three reoxidations (T2, T5, and T9). The ﬁnal two reductions at 200
°C (T 10) and 250 °C (T 1 1) result in oxide concentrations similar to those in the previous
cycle (T6 and T7, respectively), despite the intervening, higher temperature reduction at
300 °C (T8). Overall, these results suggest that after the initial restructuring, the ﬁlms are
quite stable with respect to oxidation and reduction at these temperatures. The ﬁlms
behave as if the oxide can be removed and replaced without causing irreversible changes in
the ﬁlm structure.

SnC200 Films. These ﬁlms are less reactive than the SnP ﬁlms. As seen in Figure
5.4, reductive treatments at 200 °C, 250 °C, and 300 °C lead to oxide/tin ratios of
approximately 1.6, 1.45, and 1.3. respectively. These oxide concentrations are
signiﬁcantly higher than those that remain in the SnP ﬁlms after reduction at the same
temperatures. The valence band spectrum of a SnC200 ﬁlm after reduction at 300 °C
(Figure 5.5b) is characteristic of a largely Sn2+ oxide material, as expected from the ﬁlm
stoichiometry.

Redox cycling was explored for these ﬁlms and the results are presented in Figure
5.7. The ﬁlm was subjected to four reduction cycles each consisting of H2 reduction at
200 °C, 250 °C, and 300 °C and reoxidation at 200 °C. In Figure 5.7a the results are
presented in terms of the oxide/tin ratio determined after each treatment. It can be seen
that at reduction temperatures of 200 °C and 250 °C, approximately the same amount of
oxide remains after each treatment in all four cycles. Unlike the SnP ﬁlms, little

restructuring occurs in the ﬁrst cycle. Some of the scatter in the data can be accounted for

161

 

2.0

Oxide/Tin Atomic Ratio
3‘. a;

2"
1:.

 

1.2

4.0

3.5

Tin Oxidation State

3.0»

2.5

 

 

 

 

 

 

(a)
O
O
O
I I 4 I r I L I a I I
2 4 6 8 10 12 14
Sequential Treatments
(b)
O
O
O
L I r I a I 4 J r L a I
2 4 6 8 10 l2 l4

Sequential Treatments

Figure 5.7 Oxidation - reduction cycling of SnC200 ﬁlm: Clinitial ﬁlm; I200 °C,
0250 °C, and 9300 °C H2 reduction; and 0200 °C 20% 02(He)
oxidation. (3) Oxide concentration and (b) tin oxidation state as calculated
using charge balance from XPS determined oxide and hydroxyl
concentrations after sequential treatments.

162

by replacement of hydroxyl groups with oxide groups as the cycling progresses. This is
evident from Figure 5.7b in which the calculated tin oxidation state is reported after each
treatment. There is very little change in the oxidation state from cycle to cycle. The tin
oxidation state is approximately +4 after oxidation and +3.3 and +3.0 after reductions at
200 °C and 250 °C, respectively.

There is a steady increase in the oxide concentration after reduction at 300 °C in
each subsequent cycle. This indicates that there is restructuring occurring at this
temperature. Interpretation can be based upon several assumptions. First, the oxide in the
ﬁlm can be in multiple bonding environments. These may include not only coordinatively
unsaturated surface sites and bulk sites but also, because the ﬁlm is amorphous, bond
angles and bond lengths are variable. Second, these differences in local structure lead to
lattice oxygen with different activation energy requirements for removal. It is possible that
very little restructuring is needed to raise the activation energy enough to make oxide in
some of these sites inaccessible to H2 reduction at 300 °C.

SnC400 Films. The reduction behavior of these ﬁlms is shown in Figure 5.4 and is
very similar to that of polycrystalline tin oxide. Reduction at 200 °C results in an oxide/tin
ratio of 1.6, subsequent reduction at 250 °C decreases this value to 1.48. Additional
reduction at 300 °C results in little additional oxide loss (1.45). The valence band
spectrum of the SnC400 ﬁlm after reduction at 300 °C (Figure 5.5d) resembles that of the
polycrystalline tin oxide after similar treatment (Figure 5.1d). Note that these results are

also very similar to those of the SnC200 ﬁlm after it has undergone several redox cycles.

163

Results of redox cycling of a SnC400 ﬁlm are shown in Figure 5.8. While there is
some scatter in the results, the oxide concentrations measured after the treatments are very
similar in all of the cycles and suggest that little restructuring occurs with reduction.

SnC600 Films. These ﬁlms are signiﬁcantly less reactive than the SnC400 ﬁlms
and polycrystalline tin oxide (Figure 5.5). A treatment at 300 °C is need to achieve the
same degree of reduction obtained with a 200 °C reduction of the other two materials. It
is not surprising that these ﬁlms are less reactive than the SnC400 ﬁlms: they are
signiﬁcantly more structured as indicated by the higher degree of crystallinity in the
SnC600 ﬁlms. It seems plausible that a densiﬁed, partially crystalline thin ﬁlm might have
a lower reactivity than a polycrystalline powder. Differences may be related to differences
in surface area. Recall that the SnP ﬁlm became less reactive after densiﬁcation. Lattice
oxygen vacancies may also be more mobile in the powder.

Overall, reactivity of the ﬁlms at temperatures 5300 °C is related to the
preparation temperature of the ﬁlm. The SnP (after the initial restructuring), SnC400, and
SnC600 have signiﬁcantly different reactivities. Additionally, these ﬁlms are subjected to
very harsh treatment conditions and very little restructuring is occurring, other than the
initial change in reactivity of the SnP ﬁlms. This indicates that they are quite stable with
respect to oxidation and reduction at these temperatures. Despite the densiﬁcation, ﬁlms
are reactive and densiﬁcation may contribute to the overall stability of these materials.
There is also no evidence of cracks developing the in materials with oxidative and
reductive treatments at these temperatures. This is another indication that these ﬁlms do

not undergo extensive restructuring.

2.0

1.9

1.8

1.7

1.6

1.5

Oxidefl' in Atomic Ratio

1.4

1.3

1.2

Figure 5.8

164

 

 

 

 

o . o
4 I r I r I r l r I r
2 4 . 6 8 10
Sequential Treatments

Oxidation - reduction cycling of SnC400 ﬁlm. Dinitial ﬁlm; I200 °C,
0250 °C, and 9300 °C H2 reduction; and 0200 °C 20% O2(He)
oxidation. Oxide/tin atomic ratio is detemrined with XPS analysis after
each sequential in situ treatment. Scale is adjusted to that of Figure 5.7a to
avoid an artiﬁcal over emphasis of differences.

165

Reduction at Temperatures > 300 °C. As with the polycrystalline tin oxide,
reduction for extensive periods of time at 300 °C does not lead to the formation of metal
in any ﬁlm, however, B-tin is present in all ﬁlms after H2 reduction at 325 °C. Once metal
crystallites form, the ﬁlms become discontinuous and the original structure cannot be
restored by an oxidative treatment. The reduction mechanism for SnC400 and SnC600
ﬁlms may be similar to that of the powder: in these ﬁlms the oxide/tin concentration also
remains at about 1.40 - 1.45 for reduction at 325 °C and 350 °C, despite an increase in

metal concentration with reduction time.

5.5 Concluding Remarks

The reactivity of tin oxide ﬁlms when exposed to pure H2 at atmospheric pressure
was found to be strongly correlated with degree of structure in these ﬁlms. This would be
expected thennodyamically since the driving force for crystallization must be an increase
in the stability of the metal oxygen bonds. However, the reactivity was determined in a
ﬂow system where products are constantly removed. Therefore, the complete reduction
of all ﬁlms to Sn metal (and certainly to SnO) is thermodynamically allowed even at 200
°C. Films are not reduced to metal unless the temperature is above 300 °C even after
extensive reaction times. The reduction must be limited by kinetic factors such as lattice
oxygen vacancy mobility or high activation energies. Vacancy mobility alone does not
account for the limited reacrivity since it is expected to be greater in more structured
materials and these have been shown to be less reactive than the amorphous ﬁlms.

In this study, the reduced oxide is referred to as a Sn2+ oxide. It may be equally

(or more) valid to view these materials as defective Sn“ oxides since no new crystalline

166

phases are detected and the Sn“ oxide structure and composition are apparently easily
restored. A distinction between a defective Sn“ oxide and a Sn2+ oxide is not often made
and may account for the controversy over the magnitude of the chemical shift between
Sn“ and Sn2+ oxides.

One of the problems with determining the binding energy of Sn 3d peaks for SnO
is that its surface is easily oxidized to SnO2. Commonly the surface is ion sputtered to
remove the Sn02 overlayer but this can alter the surface due to preferential sputtering of
oxygen [19, 22]. It is therefore questionable whether the material being analyzed is in
fact SnO at the surface. A second difﬁculty is created because materials may charge in the
spectrometer and there is not a reliable binding energy reference. The separation between
the 0 Is and Sn 3d5n, peaks is 43.9 eV for tin oxide in b0th forms [15], so the 0 Is peak
cannot be used as a reference. .

The lack of a reliable binding energy standard made it impossible to conclusively
determine the binding energy of the Sn 3d5,2 peak of the ﬁlms reduced at temperatures S
300 °C in this study. Films reduced at higher temperatures contain tin metal. Chemical
shifts between the metal and oxide Sn 3d5r2 peaks can be measured and correlated with the
oxide/tin ratios. If samples have dramatically different oxide/(tin oxide) ratios, this would
provide a way of measuring the chemical shift between Sn“ and Sn2+ (defective Sn“ )
oxides.

Angle resolved XPS, as described previously [31], was used to study the formation
of metal on the surface of an SnC400 ﬁlm after it was reduced at 350 °C. The measured

metal concentration decreased with increasing sampling depth (Figure 5.9) indicating that

167

 

 

 

 

0.60
’ I
0.55 -
u
:5 0.50 '-
8
2 I
g: 0.45 -
O
.8 b
a I
E 0.40 1-
+ I
0.35 '-
s I
‘ I
0.30 l r 4 i . i i r . r
0 10 20 30 40 50
Sampling Depth (A)

Figure 5.9 AR-XPS analysis of the formation of metal in an SnC400 ﬁlm after H2
reduction for 4h at 350 °C. the metal crystallites are forming at the surface
of the ﬁlm.

The oxide/(tin oxide) ratio was determined for all sampling depths (Figure 5.10). It was

found that this ratio was approximately 1.25 at the surface and increased to approximately

1.7 over the maximum sampling depth. These concentrations suggest that tin is largely in

a +2 oxidation state at the surface and in a +4 oxidation state in the bulk. This correlates

with an increase in the separation between the metal and oxide Sn 3d5,2 peaks. At the

surface, the chemical shift is 1.5 eV and it increases to a value of 2.0 eV as the sampling
depth increases. This suggests that there is a separation between the the Sn“ and Sn2+

oxide Sn 3d peaks of at least 0.5 eV. This result is consistent with Themlin et al. [19]

who found that the chemical shift was approximately 0.7 eV.

Lau and Wertheim [15] found that there was no chemical shift between Sn“ and

Sn“ oxides. Their SnO sample was deposited within the spectrometer prechamber and

did not have an Sn02 overlayer. They reason that there is no difference in binding energy

168

 

 

 

 

 

2.0 - , 0 . - 2.0
0
a: O
a - 2:-
1.8 - 4 1.8 o
'3 R
E . 3
o ' e ' ' e.
5 16 - - 16 8'
Q ' ' >
a" ' --~ 8
c . e a.
V] O
14 - I 7 14 9’?
5.
I
1.2 I 4 I a I n I a I I I 1.2
o 10 20 30 40 50

Sampling Depth (A)

Figure 5.10 Chemical shift between Sn 3d5n, metal and oxide peaks and the oxide/tin
oxide atomic ratio as a function of sampling depth for SnC400 ﬁlm after H2
reduction for 4h at 350 °C.

because the difference in free ion potential between Sn2+ and Sn“ is balanced by the

difference in the Madelung potential between the SnO and Sn02 lattices. While this may

be true for crystalline SnO and SnO2, the Madelung potential in an unrestructured

defective Sn“ oxide may be different than that in SnO. In a defective Sn“ oxide material,

the chemical shift may indeed be 0.7 eV from that of SnO2.

5.6

10.

11.

12.

13.

14.

15.

169

References
Madou, M.J.; Morrison, S.R. Chemical Sensing with Solid State Devices;
Academic Press, Inc.: San Diego, CA, 1989.

Kung, H.H. Transition Metal Oxides: Surface Chemistry and Catalysis; Elsevier:
Amsterdam, 1989.

Aso, I.; Nakao, M.; Yamazoe, N.; Seiyama, T. J. Catal. 1979, 5 7, 287.
Fattore, V.; Fuhrman, Z.A.; Manara, G.; Notari, B. J. Catal. 1975, 37, 215.
Yamazoe, N .; Kurokawa, Y.; Seiyama, T. Sens. Actuators 1983, 4, 283.

Ippommatsu, M.; Sasaki, H. J. Electrochem. Soc. 1989, 136(7), 2123. Capehart,
T.W.; Chang, 8.0 J. Vac. Sci. Technol., 1981, 18(2), 393.

Mokwa, W.; Kohl, D.; Heiland, G. Sens. Actuators 1985, 8, 101.

Pink, H.; Treitinger, L. Vite, L. Japan. J. Appl. Phys. 1980, 19(3), 513. Tomar,
M.S.; Garcia, R]. Prog. Cryst. Growth Charact. 1981, 4, 221.

Lalauze, R.; Pijolat, 0; Vincent, 8.; Bruno, L. Sens. Actuators 1992, 8, 237. Kim,
K.;~Finstad, T.G.; Chu, W.K.; Cox, X.B.; Linton, R.W. Solar Cells 1985, I3, 301.

Park, S.-S.; Mackenzie, J. D. Thin Solid Films 1996, 274, 154. Hampden-Smith,
M.J.; Wark, T.A.; Brinker, C.J. Coor. Chem. Rev. 1992, 112, 81. Takahashi, Y.;

'Wada, Y. J. Electrochem. Soc. 1990, 137,267.

Severin, K.G.; Ledford, J.S. Langmuir 1995, 11(6), 2156.
Brinker, C.J.; Scherer, G. Sol-Gel Science; Academic Press: San Diego, 1990.

Roger, C.; Hampden Smith, M.J.; Brinker, C.J. In Better Ceramics Through

Chemistry V, Hampden Smith, M.J., Klemperer, W.G., Brinker, C.J., Eds.; Mater.

Res. Soc. Symp. Proc. 271; MRS: Pittsburgh, PA, 1992; 51.

Jones, A.; McNicol, B.D. Temperature-Programmed Reduction for Solid
Materials Characterization; Marcel Dekker, Inc.: New York, 1986. LeMaitre, J.
L. in Characterization of Heterogeneous Catalysts, Delanney, F., Ed; Marcel
Dekker, Inc.: New York, 1984.

Lau, C.L.; Wertheim, G.K. J. Vac. Sci. Technol. 1978, 15(2), 622.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

170

Morgan, W. E.; Van Wazer, J. R.; J. Phys. Chem. 1973, 77, 964.
Hoﬂund, G. B. Chem. Mater. 1994, 6, 562.

Paparazzo, E.; Fierro, G.; Ingo, G. M.; Zacchetti, N. Surf. Interface Anal. 1988,
12, 438.

Themlin, J .M.; Chtaib, M.; Henrard, L.; Lambin, P.; Darville, J .; Gilles, J .-M. Phys.
Rev. B 1992, 46(4), 2460.

Kover, L.; Moretti, G.; Kovacs, Zs.; Sanjines, R.; Csemy, 1.; Margaritondo, G.;
Palinkas, J.; Adachi, H. J. Vac. Sci. Technol. A 1995, 13(3), 1382.

Sherwood, P.M.A. Phys. Rev. B 1990, 41(14), 10 151.

Sanjines, R.; Coluzza, C.; Rosenfeld, F.; Gozzo, F.; Alméras, Ph.; Lévy, F.;
Margaritondo, G. J. Appl. Phys. 1993, 73(8), 3997.

Severin, K. G.; Ledford, J. S.; Torgerson, B. A.; Berglund, K. A. Chem. Mater.
1994. 6. 890.

Severin, K. G.; Ledford, J. S. unpublished results

Software provided by Dr. Andrew Proctor, University of Pittsburgh, Pittsburgh,
PA.

Scherrer, P. Go'tt. Nachr. 1918, 2, 98.

Swanson; Tatge NBS Circular 539(1), 1953, 54.

Kohl, D. Sens. Actuators 1989, 18, 71.

Cox, D. F.; Fryberger, T. B.; Semancik, S. Phys. Rev. B 1988, 38(3), 2072.
Egdell, R. G.; Eriksen, S.; Flavel, W.R. Surf. Sci. 1987, I92, 265.

Severin, K. G.; Ledford, J. S. to be submitted to Langmuir

Chapter 6

Conclusions

6.1 Project Summary

Tin, titanium, and zirconium valerate thin ﬁlms have been prepared by spin casting
solutions produced from the hydrolysis of metal alkoxides in excess valeric acid at room
temperature. These ﬁlms are amorphous, visually transparent, uniform in thickness and
composition, adhere well to quartz substrates, are Stable and crackfree. Film thicknesses
are ~300 nm for SnVal ﬁlms and ~400 nm for TiVal and ZrVal ﬁlms. All ﬁlms consist of
a network of metal oxygen backbones coordinated with valerate ligands. Signiﬁcant
concentrations of hydroxyl ligands were found only in the tin valerate ﬁlms. In no ﬁlm
was there any evidence of unreacted alkoxide ligands.

Titanium valerate (T iVal) ﬁlms which are spin cast from aged solutions have
higher valerate levels (1.4/T i) than ﬁlms made from freshly prepared solutions (1.2/T i).
FTIR analysis showed that alkoxide ligands are present in freshly prepared solutions but
absent in 24 h. old solutions. Apparently, as the solutions age valerate ligands replace
alkoxide ligands. FTIR spectra indicate that the additional valerates may be bidentate
bridging ligands. Bridging ligands probably contribute signiﬁcantly to structural integrity
of the ﬁlms. This would account for the observations that aged solutions make higher
quality ﬁlms and that TiVal ﬁlms dissolve in H202. Hydrolysis and condensation reactions
must be completed during the spin casting and drying processes, since no alkoxide or

hydroxyl ligands are present in valerate ﬁlms.

171

172

The concentration of valerate ligands in the zirconium valerate (ZrVal) ﬁlms is
~1.9/Zr regardless of the solution age. Hydrolysis must occur more rapidly than in the
TiVal solutions since no alkoxide ligands are evident in the FTIR spectra of freshly
prepared ZrVal solutions. More rapid carboxylation and hydrolysis rates are consistent
with a nucleophilic substitution reaction mechanism since Zr is more electropositive than
Ti and is coordinated with less bulky alkoxide groups.

Tin valerate (SnVal) ﬁlms have much lower valerate concentrations (0.26/Sn) than
either the TiVal or ZrVal ﬁlms and solution aging does not increase it. Unlike the TiVal
and ZrVal solutions. alcohol not valeric acid, is the major component of these ﬁlm
solutions. It was found that if an equivalent concentration of isopropanol was added to a
TiVal ﬁlm solution, the valerate concentration in the ﬁlms was also 0.3/metal. Therefore,
the low valerate concentration in these ﬁlms is a result of the high alcohol level. Both
bidentate and monodentate valerate ligands are present in SnVal solutions and ﬁlms. The
monodentate ligands are stabilized by H-bonding to water adsorbed in the ﬁlms. The
hydroxyl concentration in the SnVal ﬁlms is 0.55/metal.

Tin oxide and titanium oxide ﬁlms were made from TiVal and SnVal ﬁlms by
calcination in air at temperatures up to 1000 °C. Calcination at 250 °C was found to be
adequate to remove the organic components of the ﬁlms. SnVal ﬁlms are fully collapsed
(20%) by calcination at this temperature. A temperature of 400 °C is required for
complete compaction of TiVal ﬁlms (50%). The higher percent compaction is consistent

with the higher concentration of valerate ligands in the TiVal ﬁlms than in the SnVal ﬁlms.

 

11.5 .911... . .......«,.r,......l.e.

173

As determined with XPS, the composition of all calcined TiVal and SnVal ﬁlms is similar
to that of polycrystalline Ti02 and SnO2 powders, respectively.

Crystallization occurs with calcination at 400 °C and TiC400 and SnC400 ﬁlms
contain traces of anatase and cassiterite, respectively. Particle sizes and crystallite
concentrations increase with calcination temperature. Rutile crystallites are present in the
Ti ﬁlms calcined at 600 °C or above and are the predominant crystalline phase in the
TiC1000 ﬁlms. All ﬁlms calcined at 600 °C are. slightly cracked, as indicated by the
presence of a Si 2p peak in the XPS spectrum. The ﬁlms become more fractured as the
calcination temperature increases.

A room temperature hydrogen peroxide treatment was used to remove the valerate
ligands from SnVal ﬁlms while leaving the tin oxygen backbone relatively untouched. It
was found that the ﬁlms collapsed only 12% with this method compared with 20% for
calcination. As with the calcined ﬁlms, their composition and XPS valence band spectra
are consistent with an Sn“ oxide. '

Mixed metal valerate ﬁlms were prepared from Ti(OPri)4 and Sn(0Pri)4. AR-XPS
analysis suggests that these ﬁlms are well mixed if the alkoxides are combined before the
addition of valeric acid. As with the single metal ﬁlms, all mixed ﬁlms are transparent,
stable, and exhibit uniform adhesion to quartz. Crystallization is inhibited by mixing: no
crystallites are apparent in the XRD patterns of mixed ﬁlms calcined at 400 °C. All mixed
ﬁlms calcined at this temperature are also crackfree and homogeneously mixed.

Refractive indices were found to be related to ﬁlm composition.

 

 

1. .2 i _. . a ti . .
. .. T».31f«—...Z‘C..O.n..flﬁ 1‘ I.d .1 ....-: .u.‘. C .V
. lit .

 

174

Mixed oxide ﬁlms calcined at 600 °C and higher are heterogeneous due to the
formation of crystallites. These crystallites are of mixed rutile phases only, no single metal
oxide crystalline phases are evident. Concentration of crystallites and crystallite size
increases with increasing calcination temperature. Mixed ﬁlms range from the highly
inhomogeneous 75:25 Sn:Ti ﬁlms which have a strong surface enhancement of Ti after
crystallization to the 25:75 Sn:Ti ﬁlm whose surface composition changes little with
calcination and crystallization.

The reactivity of tin oxide ﬁlms when exposed to pure H2 at atmospheric pressure
was found to be strongly correlated with degree of structure in these ﬁlms. Amorphous
SnP ﬁlms are highly reactive and are reduced to largely Sn“ oxide materials by a 300 °C
H2 treatment, as indicated by surface stoichiometry and XPS valence band spectra. At the
other extreme SnC600 ﬁlms, which contain cassiterite crystallites, are less reactive than
polycrystalline tin oxide powder. This latter result may be due to differences in surface
area and lattice oxygen mobility. Reduction at temperatures >300 °C results in the
formation of tin metal (B-tin) in all ﬁlms and the powder. Reoxidation at 200 °C for 1 h.
restores the original composition and reactivity of all materials except the SnP ﬁlms,
provided they have not been reduced to metal. Once metallic tin is formed, the ﬁlms are

irreversibly modiﬁed.

6.2 Future Work
Technology is currently available for the production of semiconductor gas sensors
from these sol-gel derived ﬁlms. Films can either be coated onto an electrode array or

electrodes can be attached to the surface of the ﬁlm. Several of the ﬁlms are promising as

175

sensor materials. Calcined SnVal ﬁlms are conductive at room temperature and can be
prepared with a range of reactivities. The selectivity of these ﬁlms should be investigated
to see if there is indeed a correlation between reactivity and selectivity, as suggested by
catalyst research. Additionally, sensors using highly reactive amorphous ﬁlms (i.e.
SnC250) may operate at lower temperatures than conventional polycrystalline tin oxide
sensors.

Titanium oxide ﬁlms and mixed TiO2-SnO2 ﬁlms are potentially useful sensing
materials. While they are not conductive at room temperature they may be at elevated
temperatures. More importantly, the titanium oxide ﬁlms become conductive at room
temperature after they are reduced at temperatures as low as 200 °C. In situ H2 reduction
studies were carried out on these materials. It was found that TiC400 ﬁlms were largely
inert with respect to H2 reduction and that no signiﬁcant change in oxide concentration
(1.80 - 1.85 oxide/titanium) was measured even after reduction at 400 °C. The stability of
these ﬁlms with respect to reduction suggests that they may have different sensing
properties than tin oxide ﬁlms.

As seen in Figure 6.1, mixed TiO2-SnO2 ﬁlms have intermediate reactivities. These
results suggest that the oxide reactivity of the thin ﬁlms can be tailored by making mixed
metal oxide ﬁlms. However, it is possible that these ﬁlms are composed of islands of the
two individual oxides and that only the tin oxide is being reduced. Sensor measurements
would indicate if the selectivity of mixed ﬁlms is different than that of tin oxide ﬁlms.

Ideally, these ﬁlms are homogeneous and contain a high concentration of Ti-O-Sn

176

 

 

 

 

- o 0
.0 1.8 0 0 O
E ' '
o 0
g 1.7 -
9.
<2 - I
(”t/E" O
+ 16 " El 0 I
E .
\
a Q
g 1.5 -
0 El
- LT
1.4 I r J 4 l 1 I r I
200 250 300 350 400

Reduction Temperature (°C)

Figure 6.1 Oxide concentration as a function of reduction temperature for 0TiC400,
DSnC400, OTSC400, IEMC400, and OSTC400 ﬁlms. The standard

deviation in the reported atomic ratios is i- 0.05.

177

moieties. If this is true, then mixed materials may have unique sensitivities and
selectivities when used as sensors.

Finally, Sn metal is not evident in the XPS spectrum of TSC400 ﬁlm after
reduction at 350 °C for 1 h. The tin may be dispersed well enough in the TSC400 ﬁlms to
prevent the formation of metal at this temperature. A sensor made from this material may

be more stable at high temperatures than one which uses a pure tin oxide ﬁlm.

  

ll

  

MlCHILr

 

l

  
  

N STAT

ll ‘

 
  

    
 

ll

NIVER lTV

l l

 

     

   
 

l

lHHAPIE’s

l l

 

 

1293 01762 6239