Post on 18-Jul-2020
transcript
University of California
Santa Barbara
Integration of Nanostructured Titania into Microsystems
A Dissertation submitted in partial satisfaction of the
requirements for the degree
Doctor of Philosophy in Materials
by
Zuruzi Abu Samah
Committee in charge:
Professor Noel C. MacDonald, Chair
Professor Kimberly L. Turner
Professor Cyrus R. Safinya
Professor Jacob Israelachvili
Professor Anthony G. Evans
June 2005
i
The dissertation of Zuruzi Abu Samah is approved.
Kimberly L. Turner
Cyrus R. Safinya
Jacob Israelachvili
Anthony G. Evans
Noel C. MacDonald, Committee Chair
June 2005
ii
Integration of Nanostructured Titania into Microsystems
Copyright © 2005
By
Zuruzi Abu Samah
iii
For
My Family
and
In Loving Memory of Mansur Omar
iv
ACKNOWLEDGMENTS
I thank Professor Noel MacDonald for the opportunity to come to Santa
Barbara for graduate studies and the freedom to pursue my research
interests and ideas. My education at UCSB has made me a better engineer
and, more importantly, a better person. Noel has enriched my life
tremendously and for that I am grateful.
I am fortunate to be able to interact with Professor Cyrus Safinya. I thank
Cyrus for access to cell-culture facilities and excellent X-ray tools. I had an
enriching experience interacting with the Safinya Lab, learning and
participating in biomaterials research. I thank Professor Anthony Evans for his
insights and helping me crack problems of my research. Tony has a way to
relieving stressful issues and I learn just by talking to him. I thank Professors
Jacob Israelachvili and Kimberly Turner for serving on my qualifying and
dissertation committees. I thank Professor David Clarke for proof reading the
chapter on oxidation in my dissertation.
Professors Martin Moskovits and Andrei Kolmakov are very much
appreciated for collaborating in sensor research, for sharing their insights and
educating me. Blaine Butler taught me cell-culture and collaborated with me
when I first started research. I enjoyed working with her and I wish her well. I
thank Marcus Ward for working together on nanocomposites and for delightful
conversations between symposiums in San Francisco. Diana DeRosa is
appreciated for spending a summer collaborating on Ti oxidation and sharing
v
her talents with me. She will make an outstanding engineer or medical doctor
one of these days, unless she decides to be a dancer.
There are numerous others who made my stay at UCSB meaningful and
contributed to my education. Listing them all is not possible. They include
friends, numerous clean room staff, administrative personnel in the Materials
and Mechanical Engineering departments and staff of the Microscopy and X-
Ray Labs of the MRL.
Relieving stress from a rough day at work is easy when one has great
folks at home. In this regard, I am particularly blessed. I thank Doris, Erland,
Jennifer, Kathy, Maki and Roy for their friendship. I am grateful to Amina and
family (especially lovely lovely Nancy!!) for their kindness and giving me a
home in theirs. I thank Marley for going on walks with me often. I can’t make
him understand how much I love him. There will be no dog more dear to me.
I am grateful to teachers and supervisors who nurtured my interest in
science and engineering especially Drs. R. S. Chandel, P. Cheang, W. T.
Chen, Z. Chen, D. Z. Chi, C.-h. Chiu, S. K. Lahiri, H. Li, D. Mangelinck, H. M.
Phillips and O. Prabhakar. I thank Rinus Lee, Jofelyn Lye, Kim Shyong and
Wang Weide for being great friends.
I thank my family for their love. My happiest times in the last few years
were those moments when I was home, with bro and sis. I love them all,
forever and always. For Mansur Omar, my ever lasting love and prayers.
vi
VITA OF ZURUZI ABU SAMAH
June 2005
EDUCATION: Bachelor of Applied Science (Materials Engineering) (Honours) Nanyang Technological University, Singapore, 1997 PROFESSIONAL EMPLOYMENT: Student Engineer, Murata Electronics, Singapore Site, 1996 Research Officer, Microelectronic Materials, Processes and Packaging Program, Institute of Materials Research and Engineering, Singapore, 1997-2000 (Dr. Syamal K. Lahiri, Director; Dr. Dominique Mangelinck, Supervisor) Graduate Student Researcher, Materials Department University of California, Santa Barbara, 2000-2005 (Professor Noel C. MacDonald, Advisor) HONORS: International Fellowship, National Science Scholars Program, Science and Engineering Research Council, Agency for Science Technology and Research, Singapore. (2002-2004) Graduate Student Silver Award, Materials Research Society (MRS). (2004)
vii
SELECTED PUBLICATIONS:
1. A. S. Zuruzi, N. C. MacDonald, M. S. Ward, A. Kolmakov, M. Moskovits, C. R. Safinya, “Nanostructured Titania”, University of California Disclosure, UC Case Number 2005-531-1 (2005).
2. A. S. Zuruzi, M. S. Ward, N. C. MacDonald, “Fabrication and
characterization of patterned micrometer scale interpenetrating Au-TiO2 network nanocomposites”, Nanotechnology, 16, 1029 (2005).
3. A. S. Zuruzi, N. C. MacDonald, “Facile fabrication and integration of
patterned nanostructured titania for microsystems applications”, Adv. Func. Mater., 15, 396 (2005).
4. N. F. Bouxsein, L. S. Hirst, Y. Li, C. R. Safinya, Z. Abu Samah, N. C.
MacDonald, R. Pynn, “Alignment of filamentous proteins and associated molecules through confinement in microchannels”, Appl. Phys. Lett. 85, 5775 (2004).
5. D.Z. Chi, D. Mangelinck, A. S. Zuruzi, A. S. W. Wong, S. K. Lahiri,
“Nickel silicide as a contact material for submicron CMOS devices”, J. Electron. Mater., 30, 1483 (2001).
6. A. S. Zuruzi, C. H-. Chiu, S. K. Lahiri. K. N. Tu, “Roughness evolution
of Cu6Sn5 intermetallic during soldering”, J. Appl. Phys., 86, 4916 (1999).
7. A. S. Zuruzi, C. H. Chiu, W. T. Chen, S. K. Lahiri. K. N. Tu,
“Interdiffusion of high-Sn/high-Pb (SnPb) solders in low-temperature flip chip joints under reflow process”, Appl. Phys. Lett., 75, 3635 (1999).
8. A. S. Zuruzi, G. Dong, H. Li, “Diffusion bonding of aluminium alloy
6061 in air using an interface treatment technique”, Mat. Sci. and Eng., A259, 145 (1999).
viii
ABSTRACT
Integration of Nanostructured Titania into Microsystems
by
Zuruzi Abu Samah
This thesis describes research on a novel process to fabricate
integrated nanostructured titania (NST) features as functional components in
microsystems devices. NST features were formed by oxidizing Ti films in
aqueous hydrogen peroxide followed by thermal annealing. The oxidation
kinetics and properties of NST formed were investigated. The process
developed is compatible with current microelectronics manufacturing
practices for Si and plastic substrates.
Amorphous hydrated titania gels form when hydrogen peroxide (H2O2)
reacts with Ti. Oxidation of a blanket (unpatterned) Ti surface with hydrogen
peroxide results in a titania layer with high crack density. In this study, NST
was formed by reacting pre-patterned Ti thin films with H2O2 solution. Crack
elimination was achieved when exposed Ti films were below a threshold
dimension. Hydrated titania gel crystallizes into anatase after annealing at
300 °C for 8 hr. Crack elimination is thought to result from stress reduction in
titania gels due to patterning.
ix
Oxidation of Ti films occurs by nucleation and growth mechanism.
During growth, oxidation of Ti films with thickness 50 nm and below proceeds
at a constant rate until films are fully consumed. For Ti films with thickness
100 nm or thicker oxidation rate reduces significantly after a period of growth.
This reduction is attributed to a change in mechanism controlling growth of
the hydrated titania gel layer.
Functionality of NST formed and compatibility of the process with
current microelectronics manufacturing practices were demonstrated by
exploring three applications. First, a prototype conductometric gas sensor
was fabricated that used micrometer-scale NST pad arrays as sensing
elements. This sensor is capable of detecting hydrogen and oxygen gas at
concentration of a few parts per million (ppm). Second, micrometer scale Au-
NST interpenetrating network nanocomposite contacts in micro-switches were
fabricated by infiltrating NST features with Au using electroless deposition.
Third, results of cell-culture studies showed that mouse fibroblast cells
exhibited enhanced initial attachment on NST relative to silicon dioxide which
is commonly used in microsystems devices for biological applications.
x
TABLE OF CONTENTS
1. General introduction 1.1 Motivation and objectives 11.2 Nanostructured TiO2
1.2.1 Titania polymorphs 1.2.2 Synthesis routes 1.2.3 Physical properties
557
111.3 Concluding remarks 131.4 References 15
2. Formation and characterization of integrated nanostructured TiO2
2.1 Introduction 182.2 Formation of nanostructured TiO2 202.3 Formation of nanostructured TiO2 by aqueous oxidation of
pre-patterned Ti films 21
2.4 Results and discussion 2.4.1 Morphological study using optical and scanning
electron microscopy 2.4.2 Formation and elimination of cracks in NST pads 2.4.3 Surface chemistry study using X-ray
photoelectron spectroscopy 2.4.4 Phase evolution study using X-ray diffraction 2.4.5 Structural study using transmission electron
microscopy
2324
3033
3438
2.5 Conclusions 452.6 References 46
3. Kinetics of reaction between Titanium and aqueous hydrogen
peroxide 3.1 Introduction 503.2 Experimental procedure 543.3 Results and discussion
3.3.1 Characterization of Ti thin films and calibration of apparatus
3.3.2 Effect of film thickness 3.3.3 Effect of grain size 3.3.4 Effect of temperature 3.3.5 Effect of hydrogen peroxide concentration
58
66747882
3.4 Phenomenological model of Ti oxidation in aqueous hydrogen peroxide
85
3.5 Conclusions 883.5 References 89
xi
4. Gas sensing using nanostructured TiO2 4.1 Introduction 914.2 Gas sensing using nanostructured metal oxides 944.3 Integration of nanostructured TiO2 as sensing elements
on silicon 95
4.4 Fabrication of nanostructured TiO2 on Kapton® 1014.5 Results and discussion
4.5.1 Oxygen sensing 4.5.2 Hydrogen sensing
103104111
4.6 Conclusions 1164.7 References 117
5. Fabrication of patterned micrometer scale interpenetrating Au–TiO2
network nanocomposites 5.1 Introduction 1205.2 Interpenetrating network composites 1225.3 Fabrication of interpenetrating Au–TiO2 network
nanocomposites 123
5.4 Results and discussion 1275.5 Integration of Au–TiO2 nanocomposites as contacts in
devices 140
5.6 Conclusions 1435.7 References 144
6. Attachment of mouse fibroblasts on nanostructured TiO2
6.1 Introduction 1486.2 Attachment of cells on surfaces 1496.3 Results and discussion
6.3.1 Seeding of fibroblast on various materials 6.3.2 Morphology of fibroblast on patterned
nanostructured titania
152152157
6.4 Conclusions 1596.7 References 160
7. Conclusions and future work
7.1 Conclusions 1627.2 Future work 163
xii
Chapter 1: General introduction
1.1 Motivation and objectives
Nanostructured materials are exciting due to their extraordinary physical
and chemical properties brought about by their small grain size (≤ 100
nm)1,2. Their high surface to volume ratio brings about unique properties that
are different from their bulk counterpart. Two examples of such unique
properties are enhanced sensitivity of electrical properties to chemical
species3-5 and enhanced biocompatibility of biological cells6-8. These two
properties alone make nanostructured materials attractive for integration into
future generation of microsystem (MEMS) devices as they render additional
functionality for chemical sensing and biocompatibility. With these enhanced
properties microsystem devices may find applications as electronic noses
and tongues for detection of chemicals9-11 as well as venture into the
biological milieu as implantable electronic (BioMEMS) devices such as drug
delivery12-14. Further motivation for integration of nanostructured materials
into microsystem devices is the enhancement of properties when
nanostructured materials are used in conjunction with other material
systems in the form of nanocomposites. Some of these nanocomposites
1
have lubricating and wear resistant properties15-17 and could be suitable as
contact materials in microsystem devices.
A large variety of nanostructured materials have been studied for
chemical sensing and enhancement of biocompatibility. In the former
application, nanostructured materials investigated have been of the discrete
type such as carbon nanotubes4, silicon nanowires5, palladium mesowire
arrays18, metal oxide nanowires19 and polymeric nanowires20. These
nanostructured materials have very high sensitivity to chemical species.
However, for implementation into practical devices, these nanostructures
need to be individually manipulated and positioned into place at specific
locations on a chip in a highly repeatable manner at the most cost effective
way. Presently, this is a major hurdle as there is no practical method to do
so reproducibly at low cost. Nevertheless, progress has been made towards
this goal.
In addition, any nanostructured material selected and the process used
for materials integration need to be compatible with Complementary Metal-
Oxide-Semiconductor (CMOS) processing. This is because, in most
applications, Si-based CMOS devices will be required to process information
generated. At present, most of the processes used to grow these
nanostructures require high temperatures. High processing temperatures
are undesirable because they result in high residual thermal stresses which
2
may lead to deformation of free standing structures ubiquitous in
microsystem devices. High processing temperatures also result in
interdiffusion and reactions at the interfaces of dissimilar materials which
may cause degradation of CMOS device characteristics.
For enhancement of biocompatibility, composites of nanostructured
materials such as carbon nanotubes and polymers are most commonly
used. However, unlike discrete nanostructures used in sensing applications,
nanostructured materials used to enhance biocompatibility are usually
applied over a larger area. Hence there is no real need for accurate
placement on surfaces of chips. In this aspect, implementation of
nanostructured materials for biocompatibility enhancement is, in relative
terms, more easily achieved.
Nanostructured titania has been widely investigated as a sensing
material due to its stability under adverse conditions and because its
specificity for various gases can be tailored by judicious use of surface
activation and dopants. In addition, it is widely accepted that titania
enhances biocompatibility of medical devices and implants. Therefore,
nanostructured titania is a suitable material for integration into microsystem
devices to render functionalities of gas sensing and biocompatibility.
The purpose of this research is to develop a technique of integrating
nanostructured titania into microsystem devices that is compatible with
3
material and tool sets used in conventional CMOS processing. Morphology
of nanostructured titania fabricated in the present research is porous and
sponge-like with walls consisting of nanocrystals. Nanostructured titania is
formed on rigid substrates such as Si, glass and Ti substrates as well as a
flexible organic (relatively) low-cost substrate; namely Kapton™ which is
commercially available. Emphasis will be placed on the evolution of titania
phases and morphology formed on these substrates. In addition, the
occurrence of cracks in nanostructured titania is closely studied as this
represents a potential source of reliability issue in microsystem device
applications. To demonstrate the functionality of nanostructured titania
fabricated and the compatibility of the process developed with Si-CMOS
processing, a prototype gas sensor is fabricated. In addition, the use of
integrated and patterned nanostructured titania as biocompatible cell
adhesion layers in microsystem is investigated. Also, a method of forming
integrated micrometer nanocomposites by infiltrating metal into the
nanostructured titania is explored. Having defined the scope and objectives,
the remainder of this chapter provides background information on
nanostructured titania, methods that are commonly used to synthesize
titania and selected properties of nanostructured titania that are relevant to
the present research.
4
1.2 Nanostructured TiO2
Titanium (IV) oxide (TiO2) commonly known as titania is widely used in a
number of industrial applications ranging from pigments in paints to coatings
on non-fogging surfaces. It has been recognized that properties of
nanostructured titania are different from the bulk form, which could lead to
new applications or provide better materials for existing ones. Currently,
there is tremendous effort to understand the energetics of various
nanostructured titania polymorphs as it controls size, morphology and
phases of titania formed.
1.2.1 Titania polymorphs
There are three known polymorphs of titania that exists in nature -
namely, rutile, anatase and brookite21. Rutile and anatase have tetragonal
crystal structure while brookite is orthorhombic. Rutile is the stable phase of
titania while anatase and brookite are metastable polymorphs. Hence under
ambient conditions anatase and brookite will transform to rutile when
kinetically permissible. Relative to bulk rutile, the enthalpies of formation of
bulk brookite and bulk anatase are higher by 0.71 ± 0.38 kJ/mol22 and 2.61 ±
0.41 kJ/mol23– indicating that in bulk form, rutile is the most stable phase
followed by brookite and anatase. It is noted that the anatase to rutile and
5
brookite to rutile polymorphic transformations do not occur reversibly, which
is in agreement with anatase and brookite being metastable phases.
The energetics and kinetics of nanostructured anatase, brookite and
rutile polymorphic transformations have recently been investigated by
Ranade et. al.22 as well as Zhang and Banfield24. It was shown that stability
of these TiO2 polymorphs is dependent on size. For spherical TiO2 particles
larger than about 204 nm, rutile is most stable. However between 204 nm to
38 nm TiO2 particles would exist as brookite. Below a critical size of about
38 nm, anatase is the most stable. The presence of these crossovers in
phase stability with variation in particle size is attributed to different surface
enthalpies of the various TiO2 polymorphs. The surface enthalpies of rutile,
brookite and anatase have been estimated to be 2.2 ± 0.2 J/m2, 1.0 ± 0.2
J/m2 and 0.4 ± 0.1 J/m2, respectively22. It is noted that phase stability is
governed by Gibbs free energy (ΔG=ΔH-TΔS) rather than the enthalpy (ΔH),
however data suggest that entropy of the various phases are similar.
Consequently, consideration of phase stability using enthalpies as
discussed above is valid. Because different methods of synthesizing titania
yield titania crystals with different size distribution, the phase evolution of the
various titania polymorphs in turn is dependent on the synthesis route.
6
1.2.2 Synthesis routes
Various methods for synthesizing nanostructured or nanoporous titania
have been reported in the literature. Here, only methods pertaining to
deposition of nanostructured titania films will be discussed as these are
relevant to the objective of the present work. Generally, these techniques
may be broadly classified as either chemical or physical methods. The main
chemical methods include anodization3,25, hydrolysis26,27, sol-gel15,28, and
spray pyrolysis29 while physical techniques include sputtering30,31,
supersonic cluster beam deposition32 and laser ablation33.
Anodization of Ti and its alloys to form porous nanostructured titania has
been used since the 1980’s to optimize adhesion of Ti alloy components in
joints. During anodization, the sample is immersed in an electrolyte solution
and oxidized. Recent effort includes the work of Varghese et. al.3 who
anodized Ti foils in dilute hydrofluoric acid to form nanotubes. One
parameter to control morphology of titania nanotubes was the voltage used.
It was found that with decreasing anodization voltage, the tube diameter,
wall thickness and length of the nanotubes decrease. At an anodization
voltage of 20 V, the average diameter of the tubes is 76 nm, wall thickness
is 27 nm and length is 400 nm while at 10 V the corresponding values are
22 nm, 13 nm and 200 nm, respectively. In general, porosity is increased by
increasing the voltage, electrolyte concentration and temperature. This
7
method has also been used to form nanostructured titania from Ti-6Al-4V
alloy. As formed titania nanotubes are amorphous but transformed to
anatase at about 300°C. Generally a dense amorphous TiO2 barrier layer a
few nanometers thick is formed between the porous nanostructured titania
and the unreacted Ti foil.
Another chemical route is hydrolysis. In this method, insoluble metal
hydroxides precipitate from aqueous solution, which are then subsequently
converted into its metal oxide by heat-assisted dehydration26,27. To form
nanostructured titania using this technique, titanium-containing precursors
such as an ethanol solution of tetrabutyl titanate (Ti(C4H9)4) are added to
deionized water resulting in the precipitation of titanium hydroxide gel which
is then annealed to form titania. Using this technique, coatings of
nanostructured titania with particle size ranging from a few nanometers to
hundreds of nanometers has been fabricated.
Sol-gel processing is a technique in which a sol is first formed followed
by formation of a gel15,28. Traditionally sol-gel synthesis uses either a
colloidal suspension or inorganic precursors as the starting material. The
latter approach is more often reported in the literature and is based on the
use of metal alkoxides which have the general formula M(OR)x. The
alkoxide can also be formed by reaction of a metal species such as a metal
hydroxide or metal halide with an alcohol. The metal alkoxide then
8
undergoes hydrolysis to form a colloidal sol, which further undergoes a
condensation reaction to form a gel. In essence the sol-gel technique
involves two types of reactions namely hydrolysis and condensation, as
described below:
Hydrolysis: -MOR + H2O → -MOH + ROH
Condensation: -MOR + ROM → -MOM- + ROH
OR-MOH + HOM → -MOM- + H2O
The usefulness of the sol-gel technique rests on the fact that parameters
that affect either of these reactions will impact the properties of the gel.
Hence by controlling these parameters, tremendous control of the properties
of the gel obtained is possible.
Another chemical technique to form nanostructured titania is spray
pyrolysis29. A major advantage of spray pyrolysis is that it allows the
synthesis of titania powders with a wide range of diameter, from a few to a
hundreds of nanometers, at relatively low cost. In spray pyrolysis, the vapor
of precursors, such as titanium tetrachloride (TiCl4), react with oxygen at
high temperature to form titania powder, usually in the form of aggregates.
The reaction may be adequately represented as:
TiCl4 + O2 → TiO2 + 2Cl2
9
The size of the titania particles is dictated by material and process
conditions. Since combustion of TiCl4 is exothermic, little fuel is needed to
sustain the flame and propagate the reaction, except that required to initiate
the reactor.
Among the physical techniques, cluster beam deposition is one of the
newest techniques developed. In this technique, clusters of titanium are
produced by a pulsed microplasma cluster source under high vacuum
conditions32. Titania is formed once these clusters are exposed to air, since
Ti is highly reactive to oxygen. Also, clusters of different sizes are separated
in the radial direction of the beam and nanostructured titania with various
cluster sizes can be obtained by intersecting substrates at different locations
of the beam. Clusters with largest mass are found nearest to the axis of the
beam while the smallest mass is farthest away from the beam axis. Using
this technique, nanostructured titania coatings with average particle size of
about 10 nm have been deposited.
Sputtering has been used in industrial settings to deposit titania for low
emissivity coatings and high reflection coatings30,31. Titania coatings were
formed by sputtering either Ti metal targets in an Ar plasma in the presence
of oxygen or titania targets in a pure Ar plasma. It was found that titania
formed by reactive sputtering adheres better to an organic (polyethylene
terepthalate) substrate due to formation of Ti-O-C bonds at the interface
10
between the titania layer and the substrate. Using a closed loop reactive
sputtering technique, coatings of nanostructured titania having average
grain sizes of about 10 nm have been deposited.
In laser ablation deposition of nanostructured titania, a laser source
ablates a Ti target in a chamber purged with a flow of oxygen33. The ablated
Ti species then reacts with oxygen to form titania nanoparticles that are
collected downstream. Average diameter of titania nanoparticles formed
increases with oxygen flow rate but is independent of the laser fluence.
Titania nanoparticles formed using this technique, however, tend to
coalesce on specific planes to form single crystals with dislocations at the
interface. Particles formed are about one order of magnitude larger than that
of the individual coalescing particles.
1.2.3 Physical properties
It is now widely accepted that the properties of nanostructure materials
are usually different than that of its bulk counterpart. In this section, only
properties with direct relevance to the present research will be briefly
discussed - namely electrical, chemical adsorption and wear resistant
properties.
11
As in other material systems, the electron transport properties of
nanostructured titania are different from its bulk counterpart. Because of
high surface to volume ratio in nanostructured titania, electron transport was
found to be significantly enhanced by surface effects. Rothschild et. al.34
recently showed that surface and grain boundary barriers, brought about by
electron trapping at interface states associated with chemisorbed oxygen
species, drastically reduced the conductance of nanocrystalline titania thin
films. By annealing in a reducing ambient however, the conductance can be
recovered.
The adsorption of chemical species such as oxygen, which may affect
the electrical properties as discussed above, in turn is affected by the size of
the titania nanocrystals. Recent work by Zhang et. al.35 showed that the
adsorption constant of nanostructured anatase for various organic acids
may show up to 70 fold increase when size of titania particles was
decreased from 16 to 6 nm. This observation is interesting in light of recent
work that showed enhanced cellular attachment on nanocrystal ceramics
compared to their microcrystal forms6. In physiological conditions, cellular
attachment on surfaces is always preceded by adsorption of proteins such
as fibronectin. Only after these proteins have been adsorbed will cellular
attachment and spreading occur36.
12
Another interesting property is the enhanced wear resistance of
composites containing nanostructured titania. Recent work has shown that
Au-titania nanocomposites produced using the sol-gel route exhibit excellent
wear resistance15. It was found that beyond about 5 mol% Au, the wear
resistance of the composite was significantly increased. In these
nanocomposites, Au was present as particles embedded in the titania
matrix. Wear resistance of the Au-titania composite is greater than that of Au
or its alloys which, at present, are widely used as contacts in many MEMS
devices.
1.3 Concluding remarks
The implementation of novel nanostructured materials promises
additional functionalities in microelectronic devices. Integration of
nanostructured TiO2 in microsystems is particularly important as titania is a
versatile material with applications in energy conversion, chemical sensing,
biocompatibility and drug delivery. However, there are a number of critical
issues that remain to be resolved before nanostructured TiO2 can be
implemented in devices for large volume manufacturing. First, the issue of
crack formation which is a common occurrence in nanostructured titania
structures produced using the commonly used sol-gel technique needs to be
13
resolved. Second, for successful implementation nanostructured titania must
be integrated using low-cost techniques that are compatible with current
microelectronics device manufacturing practices. In this context,
compatibility refers to both materials as well as process compatibility.
Techniques to integrate nanostructured titania that use existing process
tools and material sets would, of course, be ideal since no additional cost
would be incurred. It is the objective of the present research to realize such
a technique.
The following chapters describe a process developed that addresses the
above issues. In addition, investigations of integrated nanostructured titania
for gas sensing and biological cell attachment are presented. Also, using the
porous nanostructured titania as the ceramic component, patterned
micrometer scale interpenetrating metal-nanostructured titania
nanocomposites have been fabricated. Because the metal phase is
percolating throughout the structure, the composite is expected to have low
electrical resistance and possessing high wear resistance and hardness of
the titania phase.
14
1.4 References
1. R. Breslow, M. V. Tirrell, J. K. Barton, M. Barteau, C. R. Bertozzi, R. A. Brown, A. P. Gast, I. E. Grossmann, J. M. Meyer, R. W. Murray, P. J. Reider, W. R. Roush, M. L. Shuler, J. J. Siirola, G. M. Whitesides, P. G. Wolynes, — Beyond the Molecular Frontier: Challenges for Chemistry and Chemical Engineering, National Academies Press, Washington D. C, USA (2003)
2. A. Dowling, R, Clift, N. Grobert, D. Hutton, R. Oliver, O. O’neill, J.
Pethica, N. Pidgeon, J. Porritt, J. Ryan, A. Seaton, S. Tendler, M. Welland, R. Whatmore — Nanoscience and nanotechnologies: opportunities and uncertainties, The Royal Society and The Royal Academy of Engineering, London, UK (2004)
3. O. K. Varghese, D. Gong, M. Paulose, K. G. Ong, E. C. Dickey and C. A.
Grimes, Advanced Materials, 15, 624 (2003). 4. J. Kong, N. Franklin, C. Zhou, M. G. Chapline, S. Peng, K. Cho, H. Dai,
Science, 287, 622 (2000). 5. C. Y. Cui, Q. Wei, Q., H. Park, C. M. Lieber, Science, 293, 1289 (2001). 6. T. J. Webster, C. Ergun, R. H. Doremus, R. W. Siegel and R. Bizios,
Biomaterials, 22, 1327 (2001). 7. T. J. Webster, L. S. Schadler, R. W. Siegel and R. Bizios, Tissue
Engineering, 7, 291 (2001). 8. H-. H. Huang, S-. J. Pan, Y-. L. Lai, T-. H. Lee, C-. C. Chen and F-. H.
Lu, Scripta Materialia, 51, 1017 (2004). 9. M. Pardo and G. Sberveglieri, MRS Bulletin, 29, 703 (2004). 10. J. P Novak, E. S. Snow, E. J. Houser, D. Park, J. L. Stepnowski, R. A.
McGill, Applied Physics Letters, 83, 4026 (2003). 11. Q. Wan, Q. H. Li, Y. J. Chen, T. H. Wang, X. L. He, J. P. Li, C. L. Lin,
Applied Physics Letters, 84, 3654 (2004). 12. C. R. Martin, Science, 266, 1961(1994).
15
13. T. Paunesku, T. Rajh, G. Wiederrecht, J. Maser, S. Vogt, N. Stojićević, M. Protić, B. Lai, J. Oryhon, M. Thurnauer, G. Woloschak, Nature Materials, 2, 343 (2003).
14. L. Leoni and T. Desai, Advanced Drug Delivery Reviews, 56, 211 (2004). 15. W- M. Liu, Y-. X. Chen, G-. T. Kou, T. Xu and D. C. Sun, Wear, 254, 994
(2003). 16. M. C. Simmonds, A. Savan, E. Pfluger and H. V. Swygenhoven, Journal
of Vacuum Science and Technology, 19, 609 (2001). 17. C. Donnet and A. Erdemir, Tribology Letters, 17, 389 (2004). 18. F. Favier, E. C. Walter, M. P. Zach, T. Benter, R. M. Penner, Science,
293, 2227 (2001). 19. A. Kolmakov, Y. Zhang, G. Cheng, M. Moskovits, Advanced Materials,
15, 997 (2003). 20. Y. Wang, X. Jiang, Y. Xia, Journal of the American Chemical Society,
125, 16176 (2003). 21. A. Navrotsky and O. J. Kleppa, Journal of the American Ceramic
Society, 50, 626 (1967). 22. M. R. Ranade, A. Navrotsky. H. Z. Zhang, J. F. Banfield, S. H. Elder, A.
Zaban, P. H. Borse, S. K. Kulkarni, G. S. Doran and H. J. Whitfield, Proceedings of the National Academy of Sciences, 99, 6476 (2002).
23. T. Mitsuhashi and O. J. Kleppa, Journal of the American Ceramic
Society, 62, 356 (1979). 24. H. Z. Zhang and J. F. Banfield, Journal of Physical Chemistry B, 104,
3481 (2000). 25. J. P. Wightman and J. A. Skiles, SAMPE Journal, 24, 21 (1988). 26. H. Hirashima, H. Imai, M. Y. Miah, I. M. Bountseva, I. N. Beckman and V.
Balek, Journal of Non-crystalline Solids, 350, 266 (2004). 27. W. F. Zhang, M. S. Zhang, Z. Yin, Physica Status Solidi A-Applied
Research, 179, 319 (2000).
16
28. C. J. Brinker and G. W. Scherer, Sol-Gel Science: The Physics and
Chemistry of Sol-Gel Processing, Academic Press, San Diego (1990). 29. S. E. Pratsinis, Progress in Energy Combustion Science, 24, 197 (1998). 30. R. Dannenberg and P. Greene, Thin Solid Films, 360, 122 (2000). 31. H. Tang, K. Prasad, R. Sanjinés and F. Lévy, Sensor and Actuators B,
26-27, 71 (1995). 32. E. Barborini, I. N. Kholmanov, A. M. Conti, P. Piseri, S. Vinati, P. Milani,
C. Ducati, The European Physical Journal D, 24, 277 (2003). 33. H. Huang and X. Yao, Surface Coatings and Technology, 191, 54
(2005). 34. A. Rothschild, Y. Komem, A. Levakov, N. Ashkenasy and Y. Shapira,
Applied Physics Letters, 82, 574 (2003). 35. H. Z. Zhang, R. L. Penn, R. J. Hamers, J. Banfield, Journal of Physical
Chemistry B, 103, 4656 (1999). 36. B. D. Ratner, A. S. Hoffman, F. J. Schoen and J. E. Lemons,
Biomaterials Science: An Introduction to Materials in Medicine, Academic Press, San Diego (1996).
17
Chapter 2: Synthesis and characterization of
integrated and patterned nanostructured TiO2
2.1 Introduction
The excellent properties of nanostructured titania (NST) makes it the
material of choice in many applications. Porous NST has been used for
enhancing performance of implants1-3, gene delivery4, energy conversion5,
separation6, catalysis7 and gas sensing8-10. Many techniques have been
proposed for fabricating integrated and patterned micrometer scale NST
features. These techniques may be classified as either a reductive approach
in which features are etched from a continuous layer or the additive
approach where patterned features are directly deposited on the substrate.
Reductive approaches to forming patterned NST films have been
practiced for some time. A variety of techniques have been used to deposit
a continuous titania film11-14. The sol-gel technique, in particular, has
received tremendous attention as it renders molecular-level control, is
relatively low cost and allows incorporation of various metal dopants into
titania matrices15,16. These films are then patterned using techniques such
as reactive ion etching17 and embossing18,19 and laser trimming20. It is
noted that except for reactive ion etching, other techniques are not
18
compatible with high-volume semiconductor manufacturing processes.
Patterning of the TiO2 film could be done either before or after an annealing
step in which the film is heated at elevated temperature to convert the
amorphous titania gel into crystalline TiO2. In addition, precautions are
usually required to ensure crack formation21-23, does not occur and carbon,
from the organic precursors, is not incorporated24 in or on the NST features.
Additive approaches have been widely reported in the literature.
Because of their relatively low cost, screen printing23,25,26 and self-
assembled monolayers-assisted deposition of TiO2 have received particular
attention27-29. As the name implies, screen printing involves printing paste
containing titanium dioxide powders at desired locations on substrates using
a hard mask (the screen). Although this technique is low-cost, there are
several limitations to the process. First, the use of a hard mask means that
accurate alignment techniques are required to deposit the paste at desired
locations on the substrate. Second, the probability of adjacent
paste/features developing bridges increases as distance between adjacent
features decreases thus lowering process yield. Third, the dimension of
smallest features that can be deposited using a hard mask is about 100 μm,
which is larger than dimensions that can be deposited using
photolithography. For some microsystems applications these limitations
need to be resolved before screen printing can be implemented.
19
The microprinting of self-assembled monolayers (SAMs) using stamps
was first proposed by Whitesides et. al. 30. Selective deposition of TiO2
features occurs by interactions of functional group of the self-assembled
monolayers with TiO2 nuclei homogenously nucleated in solution. Because
surface coverage of SAMs is, in most cases, not high, this method results in
poor yield especially when large-area substrates are used. In addition,
relatively long time - up to a few hours - is required for deposition of TiO2
using this method28. Furthermore, the edge acuity of titania features
deposited using SAMs-assisted direct deposition is generally poor28.
2.2 Formation of nanostructured TiO2
Aqueous oxidation of Ti surfaces is an attractive technique for growing
NST. Using aqueous hydrogen peroxide (aq. H2O2) solution as an oxidant
Wu et. al. 31 reported the formation submicrometer porous titania layers from
thick sheets of Ti while Tengvall32-34 formed transparent bioactive titania gel
from Ti powder and unpatterned films. Similarly, Nishiguchi et. al.35 reported
the formation of a porous titania layer by reacting Ti with aqueous sodium
hydroxide. However, titania layers formed have high crack density and
delaminated extensively from the Ti substrate.
20
In this chapter, we present a technique that eliminates crack formation
and delamination in titania layers by oxidizing Ti thin films that have been
patterned below a threshold dimension. The issue of carbon incorporation
does not arise since no organic precursor is used. In addition, the patterning
technique developed allows the fabrication of miniaturized NST features and
formation of crystalline titania at relatively low temperatures, thus permitting
use of aluminium-based metallization. Hence, the technique described in
this chapter represents a practical route for fabricating and integrating NST
structures into Si-based microsystem devices.
2.3 Formation of nanostructured titania by aqueous
oxidation of pre-patterned Ti films
This research investigated the formation of NST on Ti bulk sheets and
thin films. Bulk Ti sheets (Goodfellow, 99.6% purity and 500 μm thick) were
first polished to a mirror finish with 0.3 μm colloidal silica and then rinsed
copiously with de-ionized (DI) water (>18.9 MΩ) with ultrasonic agitation. To
prepare Ti thin film samples, we used 2.5 cm square pieces of N-type
Si(100) (Mitsubishi Electronic Materials) as substrates. Si chips were
thermally oxidized at 1100 °C to grow 1 μm thick SiO2 layer (T-SiO2). These
chips were then cleaned with ultrasonic agitation for 5 min each in acetone,
21
2-propanol and de-ionized (DI) water (18.9 MΩ) and blown dry with nitrogen
prior to deposition of Ti films. A schematic flow diagram for formation of NST
on patterned Ti pad arrays is shown in Figure 2.1. Ti thin films were
patterned using either lift-off or selective masking process. In selective
masking, Ti film was electron beam evaporated on Si chips followed by SiO2
deposition. Silicon dioxide (PECVD-SiO2) was deposited using plasma-
enhanced chemical vapor deposition (PECVD) using silane (SiH4) and
nitrous oxide (N2O) precursors at 250 °C. Photo resist (PR) was deposited
on the SiO2 layer and patterned. The pattern on the PR layer was
transferred to the SiO2 layer by etching with CHF3 gas. After patterning, PR
was removed by soaking in acetone. In the lift-off technique, a PR layer was
deposited on Si chips and patterned. A Ti film was then evaporated. These
Si chips were then soaked in acetone for 24 hr, rinsed sequentially in 2-
propanol, DI water and blown dry. For depositing blanket Ti films, Si chips
were used as cleaned. The process pressure during evaporation of Ti films
was ~5.0 x 10-7 Torr. All Ti sources were of 99.995% purity or better and
cleaned by pre-evaporation to remove native oxide layer on surface.
22
Figure 2.1. Schematic of procedures for forming NST pad arrays. (a) Lift-off technique; (b) Selective masking technique.
2.4 Results and discussion
NST was formed by aging samples in aqueous H2O2 solution. Prior to
aging, Ti films were acid pickled in dilute hydrochloric acid for ~2 min to
remove native oxide layer, rinsed in DI water and then blown dry with
nitrogen. Aging was done in an oven at 80 ± 2 °C in air. Samples were
stored in a vacuum box prior to analysis. Crystal structure was analyzed by
23
X-ray diffraction (XRD) in Bragg-Brentano configuration using CuKα radiation
(1.5406Ǻ) (Phillips X’pert-MPD). Structural characterization was done using
an FEI dual beam focus ion beam (FIB) system equipped with Ga ion and
electron columns for high resolution machining and imaging, respectively.
Micromachining was done using a Ga ion current of 100 pA. Surface
chemical species were determined using a Kratos Axis Ultra X-ray
photoelectron spectroscopy (XPS) system. High resolution XPS scans were
obtained with monochromated Al Kα source (1486.6 eV) and 20 eV pass
energy with steps of 0.05 eV at a base pressure of 7.5 x 10-9 Torr. XPS
spectra collected were fitted to line shapes constructed from a linear
combination of Gaussian and Lorentzian profiles using a commercial
software (CasaXPS). Transmission electron microscopy (TEM) was done
using an FEI Sphera T20 machine operating at 200 kV. The following
sections present and discuss results of these investigations.
2.4.1 Morphological study using optical and scanning electron
microscopy
SEM micrographs of unpatterned bulk Ti sheets and Ti films after aging
in aq. H2O2 revealed formation of NST layers with high crack density -
Figures 2.2 (a) to (d). High resolution SEM shows that titania layers consist
of walls of pores having thicknesses and pore diameters ranging from 25 nm
– 50 nm and 50 nm – 200 nm, respectively. The high crack density results in
24
the formation of ‘grains’ about 5 μm – 7.5 μm average diameter. Cracks on
NST layers formed on thin films extend from the surface to the thermally
grown SiO2 layer and resulted in complete delamination NST layers
especially after prolonged oxidation times. The morphology of titania layers
formed on bulk Ti sheets and evaporated Ti films is similar. However,
delamination of NST layers formed on bulk Ti sheets is less extensive and
cracks are narrower. In addition, pores in the NST layer formed on bulk Ti
sheets are smaller.
Figure 2.2. Crack formation in NST layers formed on unpatterned Ti surfaces. SEM micrographs of unpatterned (a and b) bulk Ti sheets; and (c
and d) Ti thin films showing high crack density in NST formed on unpatterned Ti surfaces.
25
By using Ti films patterned below a threshold dimension, crack formation
on NST layers was eliminated. Figures 2.3 (a) to (e) are SEM micrographs
of NST layers formed from patterned Ti square pads of various dimensions
after aging for 2.5 hr at 80 °C in 10% aq. H2O2 solution - thickness of the Ti
layer is 2.0 μm. Cracking is most extensive on 100 μm pads and resulted in
the NST/unreacted Ti bilayer peeling off from the Si substrate – inset in
Figure 2.3 (a). Cracking is significantly reduced for 70 μm pads and for
arrays of 20 and 5 μm pads, crack formation is eliminated. In addition, gaps
developed between NST/unreacted Ti bilayer and the mask oxide. However,
NST pad arrays formed using lift-off technique have little adhesion to the
SiO2 substrate and delaminate easily during aging in aq. H2O2 solution.
Figure 2.3 (e) shows a 20 μm NST pad displaced from its original position.
In contrast, NST pads formed using selective masking have excellent
adhesion to the underlying SiO2 layer. Figure 2.3 (e) also shows that NST is
formed on sidewalls of pads formed using lift-off technique. In contrast, NST
is not observed on the sidewalls of pads formed using selective masking –
inset in Figure 2.3 (c) and (f). This observation suggests that gaps were
formed during the latter stages or possibly after aging.
Cracks were not formed on 20 μm and 5 μm pads even after annealing
at 300 °C for 8 hr, Figure 2.3 (f). However, the width of gaps between
NST/unreacted Ti bilayer and the mask oxide increased due to pad
26
shrinkage. For 20 μm pads, gap width is about 0.7 μm after drying in air but
increased to about 1.2 μm after annealing at 300 °C for 8 hr - compare
insets in Figure 2.3 (c) and (f). For arrays of 5 μm pads, width of gap before
and after annealing is estimated to be about 50 nm and 90 nm, respectively.
These observations suggest that compressive forces are created in NST
layers during aging as well as during oxidation. This hypothesis is further
supported by the curvature (concave upwards) of the peeled-off
titania/unreacted Ti bilayer - inset in Figure 2.3 (a).
Porosity of NST layers obtained could be due to morphology of the
intermediate gel layer formed during aging in aq. H2O2. Reaction of metallic
Ti with hydrogen peroxide had been investigated by Tengvall32-34 and was
shown to result in formation of a hydrated TiO2 gel layer. Recent studies by
Wu et. al. indicate that a submicron porous titania layer results when this gel
layer was annealed31. However, no high resolution microscopy images were
provided for comparison. The dark brown layer that appears after aging in
aq. H2O2 is a hydrated TiO2 gel layer. Gel layers observed by Tengvall were
yellowish. However, the color difference could be attributed to a difference in
concentration of aq. H2O2 solutions used. Focused ion beam milling was
used to investigate the structural properties of NST layers. Figures 2.4 (a
and b) are cross-section SEM micrographs obtained after milling NST layers
grown on 2.0 μm thick evaporated Ti pads after aging in 10% aq. H2O2 for
27
2.5 hrs at 80 °C. The NST/unreacted Ti interface is robust with no
delamination. In addition, the NST layers have uniform thickness with a
planar NST/unreacted Ti interface. Figures 2.4 (c) and (d) are cross-section
SEM micrographs of supported TiO2 membranes formed by oxidizing 0.35
μm thick evaporated Ti films by aging in 10 % aq. H2O2 for 3.5 hrs at 80 °C.
Although the Ti films were completely oxidized no cracks were observed on
the 20 μm pad arrays. In addition, gaps were not observed between NST
membranes and the mask oxide. This indicates that the thickness of Ti films
affects the extent of shrinkage of NST patterns.
28
Figu
re 2
.3. C
rack
elim
inat
ion
in N
ST
laye
rs fo
rmed
on
patte
rned
Ti f
ilms
belo
w a
thre
shol
d di
men
sion
. SE
M m
icro
grap
hs o
f NS
T la
yers
form
ed o
n (a
) 100
; (b)
70;
(c a
nd f)
20
and
(d) 5
μm
pad
s us
ing
sele
ctiv
e m
aski
ng; a
nd (e
) 20 μm
pad
s us
ing
lift-o
ff. C
rack
form
atio
n is
el
imin
ated
for 2
0 an
d 5 μm
pad
s. C
ompa
rison
of 2
0 μm
pad
arr
ay, (
c) b
efor
e an
d (f)
afte
r an
neal
ing
at 3
00 ºC
for 8
hr s
how
s an
incr
ease
of g
ap w
idth
afte
r ann
ealin
g (s
ee in
sets
).
29
Figure 2.4. Structure of NST layer interfaces. Cross-section SEM micrographs obtained after milling (a and b) NST layers formed after partial
oxidation of a 20 μm square pad Ti film, 2.0 μm thick; (c and d) NST membranes formed after complete oxidation of a 0.35 μm thick Ti film.
2.4.2 Formation and elimination of cracks in NST pads
Cracks in NST layers may start forming during oxidation, because of
stresses due to thermal mismatch between NST and residual titanium
layers, and/or during drying, as a result of stresses associated with
shrinkage. To elucidate the cause of crack formation, NST arrays were
observed while still in aq. H2O2 solution at room temperature, and again after
drying. Figure 2.5 (a) to (c) and (d) to (f) are optical images of 20 μm, 40 μm
30
and 50 μm pad arrays of NST before and after drying, respectively. In these
experiments, 0.35 μm thick Ti films were completely oxidized to form
supported TiO2 membranes. Prior to drying, cracks were observed only on
50 μm NST pads – it was found that 30% of pads were cracked (100 pads of
each size were studied). After drying, cracks were observed on all pads of
the 40 μm and 50 μm arrays - however, the 20 μm NST pad array remains
crack-free. These observations indicate that crack formation in NST is
primarily due to stresses associated with shrinkage during drying.
Crack elimination below a threshold dimension could be explained as a
result of load transfer between NST pads and the substrate at the edges.
Assuming NST pads are elastic membranes, load transfer at the edges is
described by the analysis of Freund and Suresh36. When NST pads are
dried, tensile stress is generated. However, points at the pad edges are
traction free as stress is relieved. However, for points at a distance away
from the edges stress is unrelaxed. The stress state approaches the
equibiaxial stress asymptotically with distance away from the edge. Hence
points away from the pad edge experience larger tensile force during drying.
As sizes of NST pads get larger a smaller volume of NST pads are nearer to
the edge. Consequently, a larger volume of the NST is experiencing a
higher tensile force which ultimately leads to crack formation when a
threshold NST pad size is exceeded.
31
Figu
re 2
.5. C
rack
form
atio
n in
NS
T pa
d ar
rays
dur
ing
oxid
atio
n an
d dr
ying
ste
ps. O
ptic
al
mic
rogr
aphs
of N
ST
arra
ys (a
-c) i
n aq
. H2O
2 afte
r oxi
datio
n; a
nd (d
-f) a
fter d
ryin
g. T
hese
imag
es
clea
rly s
how
that
cra
ck fo
rmat
ion
occu
rs p
rimar
ily d
urin
g th
e dr
ying
ste
p. T
he p
ads
in th
e ar
rays
ar
e 20
, 40
and
50 μ
m s
quar
es a
s in
dica
ted.
32
2.4.3 Surface chemical study using X-ray photoelectron spectroscopy
X-ray photoelectron spectroscopy of NST layers suggest that aging in
aq. H2O2 solution resulted in the formation of TiO2 species only. Figures 2.6
(a) and (b) are survey and high resolution XPS spectra, respectively, of NST
layer formed on evaporated Ti thin film. Similar results were obtained for
NST layers formed on bulk Ti sheets. All spectra are referenced to C1s peak
at 285.0 eV37. Assuming a Tougaard background, raw spectra were fitted
using Gaussian-Lorentzian components with appropriate constraints for
area, full-width-at-half-maximum and position parameters using a
commercial software (CasaXPS). From the analysis, binding energies for
the Ti 2p3/2 components were found to be 459.0 eV and 458.9 eV for titania
on bulk and evaporated Ti film, respectively. For NST on bulk and
evaporated thin film Ti, the Ti 2p1/2 components have a value of 464.8 eV.
These experimental values obtained are close to those reported in the
literature for TiO2 of 458.9 eV and 464.6 eV for Ti 2p3/2 and Ti 2p1/2
components, respectively38,39.
33
Figure 2.6. XPS spectra of TiO2 species after drying. (a) Survey and (b) high resolution XPS spectra of NST layer formed on evaporated thin Ti film.
2.4.4 Phase evolution study using X-ray diffraction
XRD studies show that amorphous TiO2 and nanocrystals of anatase
TiO2 polymorph were formed after aging and the amorphous phase
transforms to anatase upon annealing. Figure 2.7 shows XRD spectra of
evaporated Ti film, as-aged and annealed TiO2 layer formed from
evaporated blanket Ti films. Spectra of the sample aged in 10 % aq. H2O2
solution for 2.5 hr at 80 °C exhibit Ti peaks, which correspond to unreacted
Ti in aged films and three broad peaks at 2θ values of 25.20°, 47.97° and
62.68° which can be assigned to anatase 101, 200 and 204 planes. The
broadness and low intensity of these peaks suggest that as-formed titania
34
layer consists of anatase nanocrystals in a largely amorphous titania matrix.
Upon annealing at 300 °C for 8 hr, these peaks sharpened significantly and
increased in intensity. The peak sharpening and intensity increase upon
annealing are due to transformation of the amorphous phase to anatase
which also agrees with the appearance of additional peaks at 2θ values of
53.91°, 54.95° and 75.04°. These latter peaks correspond to 105, 211
and 215 reflections of anatase. All anatase peaks in the spectrum of
annealed TiO2 layer match perfectly to corresponding ones in spectrum
collected from reference TiO2 anatase powder (Alfa Aesar, 99.6 %). No
peaks from other TiO2 polymorphs are observed in spectra of annealed
samples. Hence, XRD and XPS data indicate that only nanocrystals of
anatase TiO2 and an amorphous titania phase are formed after aging.
Subsequent annealing transforms the amorphous phase to anatase.
Formation of single phase nanostructured anatase from amorphous TiO2 by
annealing in air as well as the coexistence of these phases had been
reported previously40,41.
35
Figure 2.7. XRD spectra of TiO2 species formed after aqueous oxidation. XRD spectra of (i) anatase reference sample, (ii) evaporated Ti film, (iii) as-
aged and (iv) annealed NST layer. ( g indicates anatase peak)
X-ray pole figure determination of bulk and thin film Ti surfaces indicates
orientation of anatase crystals formed during annealing is not influenced by
texture of the underlying Ti substrate. Samples in this experiment are
partially oxidized to prevent delamination during oxidation. Figures 2.8 (a)
and (b) are pole figures collected for Ti 0002 (2θ = 38.42 °) of bulk Ti sheet
and evaporated thin Ti film, respectively, before aging. These pole figures
agree with other studies and are typical of hot rolled Ti sheets and deposited
thin Ti film, respectively42,43. Figures 2.8 (c) and (d) are corresponding
scans for anatase 101 (2θ = 25.20°), collected after aging in aq. H2O2
solution and subsequent annealing at 300 °C for 8 hr. The similarity of
Figures 2.8 (c) and (d) and the narrow intensity range of these spectra
36
suggest that anatase crystals formed during annealing have a random
orientation for both bulk and thin film Ti and texture of the underlying Ti
substrate does not influence orientation of these anatase crystals. This is in
agreement with prior reports which showed that crystallization of anatase
during annealing of amorphous TiO2 layer takes place by recrystallization of
small anatase particles and solid-state aggregation of amorphous
particles40. These mechanisms would be expected to produce randomly
oriented anatase crystals.
Figure 2.8. XRD pole figure of parent Ti and TiO2 formed after aqueous oxidation. Pole figures shown are for Ti 0002 of (a) Ti sheet and (b)
evaporated thin Ti film, both before aging; and of anatase 101 collected after aqueous oxidation and annealing of (c) bulk and (d) thin Ti film
samples.
37
2.4.5 Structural study using Transmission Electron Microscopy
In this section we discuss in detail structural characterization of NST
using TEM. Cross-sectional TEM samples were prepared using a ‘lift-out
technique’, which involves micro-machining a thin slice of the NST layer
using a focus ion beam (FIB). The use of this technique for TEM preparation
of fragile and porous samples is relatively new44,45 and is described in the
following. Figure 2.9 shows important steps in the preparation of cross-
sectional TEM samples. First a membrane was made by milling a trench on
either side it. The membrane was thinned to 100 nm and then suspended by
milling its bottom and sides. A notch is then made at each end of the
supporting beam. Subsequently, the membrane was lifted out using a glass
needle attached to a micromanipulator with the help of microscopes and
placed on a carbon coated Cu grid. Three TEM samples were made from
different NST patterns to ensure that TEM observations were representative
of the whole specimen.
38
Figu
re 2
.9. S
teps
in c
ross
-sec
tion
sam
ple
prep
arat
ion
for T
EM
usi
ng F
IB m
illing
. (a)
Sam
ple
befo
re
mill
ing;
(b) t
renc
h m
illin
g on
eith
er s
ide
of m
embr
ane;
(c) m
embr
ane
afte
r fur
ther
thin
ning
; (d)
mem
bran
e su
spen
ded
by m
illing
the
side
s an
d bo
ttom
; (e)
one
end
of t
he m
embr
ane
befo
re n
otch
ing
and
(f) th
e ot
her e
nd a
fter n
otch
ing.
39
Figu
re 2
.10.
Typ
ical
cro
ss-s
ectio
nal T
EM
imag
es a
nd d
iffra
ctio
n pa
ttern
s. (a
) Low
mag
nific
atio
n im
age;
(b) a
nd (c
) hig
h m
agni
ficat
ion
imag
es o
f tw
o sa
mpl
es ta
ken
from
diff
eren
t NS
T pa
ttern
s; (d
), (e
) an
d(f)
are
SA
ED
patte
rns
corr
espo
ndin
gto
regi
ons
deno
ted
by(1
),(2
)and
(3)r
espe
ctiv
ely
in(c
).
40
Figures 2.10 (a) to (c) are TEM images of two samples from different
NST pads showing typical cross-sections. The difference in contrast
indicates that exposed Ti film is completely oxidized. Pores traversing the
entire thickness of NST layer are observed. The diameter of these pores is
about 150 nm. Also, the NST layer is considerably thinner than that of the
parent Ti film. From the TEM images a 350 nm thick NST layer was formed
from a 500 nm thick Ti film, in agreement with SEM imaging. It is postulated
that this reduction in thickness results from shrinkage of the hydrated titania
gel layer and dissolution due to finite solubility of Ti in the aqueous H2O2
solution. The NST/ T-SiO2 interface is sharp with no cracks detected. This
suggests that NST pads formed are strongly adhering to the T-SiO2
substrate. Since thickness of Ti films deposited using microelectronics
process tools can be controlled accurately to a few tens of nanometers, and
by taking into account shrinkage of titania gels, ultra-thin and porous NST
features could be integrated into devices using this technique.
Cross-sectional TEM studies also reveal features that are not apparent in
the SEM investigations. Figures 2.10 (b) and (c) show that lateral oxidation
of Ti under the P-SiO2 mask layer occurred. In both cases, oxidation
progressed ~580 nm into Ti films from the edge of the P-SiO2 mask. The
lateral oxidation rate under the experimental conditions used is estimated to
be ~2.8 nm/min. To investigate the crystal structure of the NST/Ti interface,
41
selected area electron diffraction (SAED) studies were carried out in regions
labeled (1), (2) and (3). The corresponding diffraction patterns are shown in
Figures 2.10 (d), (e) and (f), respectively. Electron diffraction pattern
obtained from region (1) suggests the presence of polycrystalline anatase.
The diffraction pattern acquired from region (2) is similar to that obtained
from (1). However, the intensity of ring patterns is noticeably reduced, which
suggests a higher proportion of amorphous phase in regions of NST nearer
to the unreacted Ti. The ring diffraction pattern indicates that titania formed
is crystalline and that nanocrystals of anatase are randomly oriented in the
walls of the pores. These observations are in agreement with XRD studies.
The diffraction patterns are indexed accordingly as labeled in Figures 2.10
(d) and (e). Energy dispersive X-ray spectroscopy studies at regions (1) and
(2) show that elemental compositions of these regions are similar. These
results confirm that TiO2 is formed in these regions after annealing. However
the proportion of the crystalline titania is lower in region (2). As expected,
the diffraction pattern of region (3) is typical for that of polycrystalline Ti
consisting of discrete rings with clustered elongated spots which suggests
small grains with multiple orientations46.
The presence of a completely amorphous titania phase had been
reported at the interface between Ti and porous titania formed by
anodization of thick Ti foils followed by annealing treatment. This amorphous
42
titania phase has been suggested to confer electrical isolation of the
semiconducting titania phases from Ti in some gas sensing devices47.
However a direct comparison with our results is not possible as we used a
different method to form the titania layer. Nevertheless, SAED results
indicate a fully amorphous phase is not present in our samples.
Figure 2.11. Crack formation near the Ti/NST interface. Cross-sectional image of sample (a) before and (b) after annealing.
Cross-sectional TEM also reveals cracks under the P-SiO2 mask layer
near the interface between the NST and unreacted Ti film. Such cracks were
observed on all three TEM samples studied. Two examples of crack
43
formation are shown in Figures 2.10 (b) and (c). Cracks are observed along
the P-SiO2/NST interface and continue near the NST/Ti interface before
being arrested ~150 nm from the NST/ T-SiO2 interface. It is noted that
cracks may be formed due to shrinkage during drying in air after oxidation or
during the annealing treatment. Figures 2.11 (a) and (b) show SEM images
of samples cross-sectioned by FIB milling before and after annealing. These
samples have been milled using a high beam current. Redeposition of
debris causes the pores of NST to be filled up. Cracks are observed in both
samples which indicate that crack formation occurs prior to thermal
annealing. It is postulated that cracks are formed due to shear stresses near
the NST/Ti interface as a result of shrinkage of the hydrated titania gel
during drying in air. However, no cracks were observed along the NST/T-
SiO2 interface. Also the NST under the P-SiO2 mask is dense with relatively
less porosity.
44
2.5 Conclusions
In conclusion, a method of forming integrated and patterned NST layers
for microsystems applications have been demonstrated. Using Ti thin films
patterned below a threshold area, crack formation on NST layers can be
eliminated. Crack formation in NST layers occurs primarily during drying and
is attributed to stresses associated with shrinkage. NST layers formed have
sponge-like morphology with pore diameter and wall thickness of about 50
nm - 200 nm and 25 nm - 50 nm, respectively. As-formed NST is largely
amorphous but transformed to anatase upon annealing at 300 °C. Pore size
of NST formed from bulk Ti foils is smaller that on Ti thin films. TiO2 grains in
NST on both Ti foils and Ti thin films have a random orientation.
45
2.6 References
1. F. Heidenau, F. Stenzel, G. Ziegler, Key Eng. Mater., 192-1, 87 (2000).
2. D. M. Brunette, P. Tengvall, M. Textor, P. Thomsen, Titanium in
medicine: materials science, surface science, engineering, biological responses and medical applications. Springer-Verlag, New York, (2001).
3. T. J. Webster, R. W. Siegel, R. Bizios, Biomaterials, 20, 1221 (1999).
4. T. Paunesku, T. Rajh, G. Wiederrecht, J. Maser, S. Vogt, N. Stojićević, M. Protić, B. Lai, J. Oryhon, M. Thurnauer, G. Woloschak, Nat. Mater., 2, 343 (2003).
5. M. Grätzel, Nature, 414, 338 (2001).
6. T. Sano, N. Iguchi, K. Lida, T. Sakamoto, M. Baba, H. Kawaura,
Applied Physics Letters, 83, 4468 (2003).
7. M. P. Harold, C. Lee, A. J. Burggraaf, K. Keizer, V. T. Zaspalis, R. S. A. Delange, Mater. Res. Soc. Bull., 19, 34 (1994).
8. G. Sberveglieri, L. E. Depero, M. Ferroni, V. Guidi, G. Martinelli, P.
Nelli, C. Perego, L. Sangaletti, Adv. Mater., 8, 334 (1996).
9. O. K. Varghese, D. W. Gong, M. Paulose, K. G. Ong, E. C. Dickey, C. A. Grimes, Adv. Mater., 15, 624 (2003).
10. P. I. Gouma, M. J. Mills, K. H. Sandhage, J. Am. Ceram. Soc., 83,
1007 (2000).
11. G. P. Burns, Journal of Applied Physics, 65, 2095 (1989).
12. C-. C. Ting, S-. Y. Chen and D-. M. Liu, Journal of Applied Physics, 88, 4628 (2000).
13. S. H. Oh, D. J. Kim, S. H. Hahn and E. J. Kim, Materials Letters, 57,
4151 (2003).
14. H. Tang, K. Prasad, R. Sanjinés and F. Lévy, Sensors and Actuators B, 26-27, 71 (1995).
46
15. C. J. Brinker and G. W. Scherer, Sol-Gel Science: The Physics and
Chemistry of Sol-Gel Processing, Academic Press, San Diego (1990).
16. J. Livage, Current Opinion in Solid State and Materials Science, 2, 32
(1997).
17. A. S. Holmes, R. R. A. Syms, M. Li and M. Green, Applied Optics, 32, 4916 (1993).
18. H. Krug, N. Merl and H. Schmidt, Journal of Non-Crystalline Solids,
147, 447 (1992).
19. R. L. Roncone, L. A. Wellerbrophy, L. Weisenbach and B. J. J. Zelinski, Journal of Non-Crystalline Solids, 128, 111 (1991).
20. T. Brylewski and K. Przybylski, Applied Superconductivity, 1, 737
(1993).
21. H. Kozuka, Journal of The Ceramic Society of Japan, 111, 624 (2003).
22. K. Kajihara, K. Nakanishi, K. Tanaka, K. Hirao and N. Soga, Journal
of The American Ceramic Society, 81, 2670 (1998).
23. M. C. Carotta, M. Ferroni, V. Guidi, G. Martinelli, Advanced Materials, 11, 943, (1999).
24. E. Halary, G. Benvenuti, F. Wagner ad P. Hoffmann, Applied Surface
Science, 154-155, 146 (2000).
25. L. Gao, Q. Li, Z. Song and J. Wang, Sensors and Actuators B, 71, 179 (2000).
26. C. J. Barbe, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V.
Shklover and M. Gratzel, Journal of The American Ceramic Society, 80, 3157 (1997).
27. R. J. Collins, H. Shin, M. R. Deguire, A. H. Heuer and C. N. Sukenik,
Applied Physics Letters, 69, 860 (1996).
47
28. K. Koumoto, S. Seo, T. Sugiyama and W. S. Seo, Chemistry of Materials, 11, 2305 (1999).
29. M. Bartz, A. Terfort, W. Knoll, W. Tremel, Chemistry – A European
Journal, 6, 4149 (2000).
30. A. Kumar, H. A. Biebuyck and G. M. Whitesides, Langmuir, 10, 1498 (1994).
31. J. M. Wu, S. Hayakawa, K. Tsuru, A. Osaka, Scripta Materialia, 46,
101 (2002).
32. P. Tengvall, “Titanium-hydrogen peroxide interaction with reference to biomaterial applications”, PhD thesis, Department of Physics and Measurement Technology, Linköping University, Sweden (1989).
33. P. Tengvall, I. Lundstrom, L. Sjoqvist, H. Elwing, L. M. Bjurstein,
Biomaterials, 10, 166 (1989).
34. P. Tengvall, H. Elwing, I. Lundstrom, Journal of Colloid and Interface Science, 130, 405 (1989).
35. S. Nishiguchi, S. Fujibayashi, H. M. Kim, T. Kokubo, T, Nakamura,
Journal of Biomedical Materials Research A, 67A, 26 (2003).
36. S. Suresh, L. B. Fruend, Thin Film Materials: Stress, Defect Formation and Surface Evolution, Cambridge University Press, Cambridge (2003), p. 222.
37. J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben, Handbook of
X Ray Photoelectron Spectroscopy: A Reference Book of Standard Spectra for Identification and Interpretation of XPS Data, Physical Electronics Inc., Minnesota (1995).
38. T. Choudhury, S. O. Saied, J. L. Sullivan, A. M. Abbott, Journal of
Physics D - Applied Physics, 22, 1185 (1989).
39. D. Gonbeau, C. Guimon, G. Pfisterguillouzo, A. Levasseur, G. Meunier and R. Dormoy, Surface Science, 254, 81 (1991).
40. H. Z. Zhang, M. Finnegan, J. F. Banfield, Nano Letters, 1, 81 (2001).
48
41. S. Z. Chu, K. Wada, S. Inoue, S. Todoroki, Chemisty of Materials, 14, 266 (2002).
42. J. P. Gueneau de Mussy, G. Langelaan, J. Decerf, J.-L. Delplancke,
Scripta Materialia, 48, 23 (2003).
43. B. Major, R. Ebner, P. Zieba, W. Wolczynski, Applied Physics A – Materials Processing, 69, S921 (1999).
44. A. J. Smith, P. R. Munroe, T. Tran, M. S. Wainright, Journal of
Materials Science, 36, 3519 (2001).
45. S. Shukla, S. Seal, P. Nguyen, H. T. Ng and M. Meyyappan, Sensor Letters, 1, 75 (2003).
46. B. D. Cullity, Elements of X-Ray-Diffraction, Addison-Wesley
Publishing Company, Massachusetts (1959).
47. O. K. Varghese, D. W. Gong, M. Paulose, K. G. Ong, E. C. Dickey, C. A. Grimes, Advanced Materials, 15, 624 (2003).
49
Chapter 3: Kinetics of reaction between Ti films and
aqueous hydrogen peroxide
3.1 Introduction
It is widely known that titanium has good corrosion resistance. This
property is attributed to the presence of a compact thin film of TiO2. In air, a
thin TiO2 layer readily forms on a pristine Ti surface due to the large
corresponding reduction in the free energy, ΔG=-203.8 kCal/mol1. The
overall reaction can be represented as follows:
Ti + O2 ⇒ TiO2 ………. Eqn. 3.1
In an aqueous environment, oxidation of a pristine Ti surface has been
explained on the basis that the Ti/TiO2 potential is more negative than the
potential of the hydrogen electrode2. The reaction between Ti surfaces with
aqueous H2O2 however, is more complicated and involves the formation
intermediate species such as Ti-peroxo compounds which takes the form of
gel-like structures3-5.
The oxidation of Ti surfaces in aqueous H2O2 solution had, over many
years, been studied by numerous research groups. This interest lies in the
fact that H2O2 released in biological systems significantly reduced the
50
corrosion resistance of titanium implants and results in the accumulation of
titanium ions in adjacent tissues. In contrast, titanium implants in H2O2-free
aqueous environment, in-vitro, have high corrosion resistance and,
correspondingly, release lower amounts of Ti ions3. From a technological
stand point, these studies have contributed to the development of better
performance implants.
Despite these studies, gaps remain in our understanding of Ti surface
oxidation in aqueous solution. It is now generally accepted that two layers of
surface oxide are formed when a Ti surface is exposed to an aqueous
environment. The oxide layer adjacent to the metal is dense and acts as a
barrier to further corrosion while the second layer, adjacent to the aqueous
solution, is porous and does not impede the corrosion process. Addition of
H2O2 results in a significant release of Ti ions into the solution and growth of
the porous titania layer. Healy and Ducheyne6,7 suggested that thickness of
these two layers is about the same. These researchers found that oxide
growth decreased significantly after the layer reached a thickness of ~7.5
nm and suggested that oxidation kinetics can be explained on the basis of a
limiting oxide thickness in which electric-field assisted transport of metal ions
into the oxide is the rate limiting step. Using results obtained from
electrochemical impedance spectroscopy techniques, Pan et. al.8,9
calculated that the dense-TiO2 layer is much thinner than the porous layer.
51
However, Pan et. al.8,9 was not able to determine the rate limiting step
controlling growth of the porous titania layer but alluded to the study of
Tengvall. Tengvall3 had proposed that the rate limiting step controlling
growth is dissolution of Ti ions into the solution. Recently Bearinger et.
al.10,11 used atomic force microscopy (AFM) coupled with electrochemical
techniques to elucidate the mechanism of oxide evolution. These
researchers concluded that growth of the titanium oxide layer occurs at the
metal/dense-TiO2 interface. Models proposed by Tengvall3 as well as
Ducheyne and Healy6,7, however, suggest growth of the total oxide layer
occurs at the oxide-solution interface – see Figure 3.1.
Another gap in the understanding of aqueous Ti oxidation is the effect of
the underlying Ti substrate. No systematic study has been carried out on the
effect of microstructure of Ti substrates on oxidation kinetics. It is widely
known that grain boundaries are regions of high energy and, therefore,
preferentially corrode. In addition, grain boundary diffusion is faster than
bulk diffusion. Hence, grain size of the Ti substrate would be expected to
have a large effect on the rate of oxidation. In addition, for thin Ti films,
kinetics of the oxidation process may be affected by thickness of the
underlying Ti layer. Healy and Ducheyne6,7 performed their experiments on
sputtered Ti thin films while Pan et. al.8,9 and Bearinger et. al.10,11 used bulk
52
Ti rods and discs, respectively. However, no prior work has investigated the
effect, if any, of the underlying Ti microstructure.
Figure 3.1. Schematic, based on current understanding, of processes taking place during reaction of Ti surfaces with aqueous hydrogen peroxide
solution. (a) In Bearinger’s model growth occurs at the interface between dense and hydrated Ti oxide layers due to mass transport of H2O2
molecules across the oxide layer. (b) In the model proposed by Tengvall as well as Healy and Ducheyne, the oxide growth occur at the hydrated Ti oxide layer/solution interface due to mass transport of Ti ions across the
hydrated Ti oxide layer. ( represents flux of Ti from the substrate) TiJ
53
The aim of this chapter is two-fold: First, to elucidate the mechanism
controlling growth of the porous titanium oxide layer and, second, to study
the effect of Ti film microstructure on oxidation kinetics. Since the overall
goal of this research is to integrate NST into microsystems, focus is placed
on Ti thin films rather than bulk Ti.
3.2 Experimental procedure
Oxidation kinetics of Ti films was investigated by monitoring change in
resistance of patterned Ti films deposited on glass substrates. In this
chapter, all Ti films have been deposited using electron beam evaporation.
The effect of thickness and average grain size of Ti films on the oxidation
kinetics was studied. Other parameters investigated include temperature
and concentration of hydrogen peroxide solution.
A schematic of the set-up used to monitor Ti oxidation kinetics is shown
in Figure 3.2 (a). Patterns of Ti lines were deposited on glass substrates
using the lift-off technique. Glass substrates were degreased by soaking
consecutively in acetone, isopropyl alcohol and deionized (DI) water, with
ultrasonic agitation for 5 min each. These substrates were then baked at
120 °C for 10 min. Photoresist was spun, exposed with patterns and
developed. A thin Ti film was deposited on substrates using electron-beam
54
evaporation. To reduce oxygen incorporation, Ti source was evaporated for
about 5 min prior to actual deposition on glass substrates. During this time,
Ti was evaporated on a shutter that lies between the glass substrate and the
source. After 5 min, the shutter was removed and Ti was evaporated on the
substrate. Pressure in the chamber was about 1 - 2 x 10-7 Torr during
deposition. Thickness of Ti films deposited was estimated using a quartz
crystal monitor. Chamber pressure at the end of a deposition run was lower
than that at start. After deposition, substrates were soaked in acetone
overnight. For oxidation experiments, the samples were connected as
shown in Figure 3.2 (a). Oxidation was done at various temperatures. The
apparatus was heated using a hot plate and temperature maintained to
within ±3 °C of that required. Samples were dipped into the reaction bath
only when the temperature stabilized. Resistance of Ti film was recorded
every 10 or 15 seconds intervals. For each condition, three samples were
oxidized and the average calculated.
55
Figure 3.2. Schematic of (a) apparatus for oxidation kinetics experiments and (b) Ti line on sample.
Phase transformation and chemical reactions in solids often result in
changes in measurable physical properties. Experimental quantification of
these changes enables a reliable and accurate estimate of the extent the
transformation has occurred or the progress of reaction. The use of
electrical resistance as a parameter to follow the extent of chemical
reactions involving thin films had been investigated by many researchers in
the microelectronics industry12-15. Tu et. al. used this technique to study
crystallization induced by thermal annealing of amorphous co-evaporated
transition metal-silicon13,14 thin films while Howard et. al studied kinetics of
56
reaction of transition metal-aluminum thin films15. As mentioned previously,
oxidation of Ti results in the consumption of conducting Ti and formation of a
two layered oxide that is, electrically, highly insulating. Hence by measuring
the electrical resistance of the residual Ti film, kinetics of the oxidation
reaction can be monitored. The leakage current across the solution is
expected to be small since deionized water was used. Nevertheless, the
resistance of DI water is expected to decrease during oxidation because of
dissolution of the hydrated titania gel layer.
Following the analysis of Tu and Howard et. al.12-15, a measure of
progress of the reaction called the extent of reaction, Xt, can be defined as:
( ) ( )( ) ( )0
0
FilmFilm
FilmFilmt RR
RtRX−∞−
= ………. Eqn. 3.2
In Eqn. 3.2, and ( )0FilmR ( )∞FilmR are resistance of films at start and the
end of the reaction while ( )tRFilm is resistance of films at time t. It follows
from Eqn. 3.2 that Xt at start and end of the reaction is 0 and 1, respectively.
Kinetics of the oxidation reaction can be followed by plotting Xt against time.
57
3.3 Results and discussion
3.3.1 Characterization of Ti films and calibration of apparatus
The measurement of resistance of the residual Ti film to monitor extent
of Ti oxidation in an aqueous environment has not been reported previously.
Therefore, careful calibration of the oxidation apparatus was carried out to
ensure resistance measured was due to current through the Ti film and not
from leakage current through the DI water. To calibrate the apparatus, Ti
films of various thickness were evaporated at 2.5 Å/s. Ti films used for
calibration were identical to those used to study the effect of film thickness
on oxidation kinetics. In the following, results of characterization of Ti films
used in the calibration are presented.
Characterization of Ti films for calibration
During evaporation, thickness of Ti films deposited was measured using
a quartz crystal monitor. The thickness of evaporated Ti film samples was
further verified using x-ray reflectivity experiments after evaporation. Incident
angle in reflectivity experiments was fixed at 3 degree. Reflectivity spectra
are shown in Figure 3.3 (a). The thickness, Δ , of a Ti film sample was
calculated using the formula:
( )( )ii
ii nnθθ
λsinsin2 1
1
−−
=Δ+
+ ………. Eqn. 3.3
58
where λ, and in iθ are the wavelength of radiation used (1.54056 Å for
CuKα radiation), the ith order of a reflection and the corresponding diffracted
angle at the peak of the ith order, respectively. Thicknesses of films
determined using X-ray reflectivity are close to those recorded using quartz
crystal monitor, Figure 3.3 (b). Digression between the two sets of thickness
values is greater as thickness increases. On average, film thickness
determined using X-ray reflectivity is 95% of that recorded using a quartz
crystal monitor. In the following, nominal film thickness recorded using the
quartz crystal monitor is used.
Figure 3.3. Plots of (a) reflectivity spectra of various films and (b) correlation between thickness determined using X-ray reflectivity and
quartz crystal monitor techniques.
59
Average grain size was determined using X-ray broadening. Figure 3.4
(a) shows x-ray diffraction spectra taken in Bragg-Brentano geometry of Ti
films of various thicknesses. Ti films are highly textured in the (0002) plane,
similar to a previous report22. To determine average grain size, reflections
from (0002) planes were used. Average grain size was determined using
Scherer’s relation22:
θλ
cosBCd = ………. Eqn. 3.4
where C, λ, B and θ are a constant with value of 0.9, the wavelength of
radiation used (1.54056 Å for CuKα radiation), true full width at half
maximum intensity and the diffracted angle, respectively. The technique
above gives average grain size perpendicular to the surface22. To obtain the
true full width at half maximum intensity B (in radians), Warren’s method
was employed22. In this method, B is determined from:
222SM BBB −= ………. Eqn. 3.5
where and are the measured full widths at half maximum
intensity of the sample and a standard, respectively. The standard used was
a bulk Ti sheet obtained from a commercial supplier that has been annealed
for a few hours at elevated temperatures and has an average grain size of a
few micrometers.
MB SB
60
Figure 3.4. Plots of (a) x-ray diffraction spectra taken in Bragg-Brentano geometry of Ti films and (b) Variation of average grain size with
thickness. (Evaporation rate 2.5 A/s)
Figure 3.4 (b) shows variation of average grain size with thickness.
Average grain size increases with thickness from 25 to 100 nm. For 100,
150 and 200 nm thick films, the average grain size is ~25 nm.
Calibration of apparatus
Calibration was done by measuring resistance of films, while immersed
in DI water, at various temperatures. The corresponding variations of
resistivity with thickness and temperature were then calculated. From the
61
variation of resistivity with temperature, the temperature coefficient of
resistivity (TCR) was calculated. Resistance values of samples in air and in
DI water were found to be similar, which suggest current leakage through DI
water is negligible. In the following, calibration results obtained from
samples immersed in DI water are presented and compared to those in the
literature.
The resistance measured using the apparatus shown in Figure 3.2 (a)
include resistance of the film, , and the contact resistances, . The
total resistance measured, R, can be written as:
FilmR RContacts
FilmContacts RRR += ………. Eqn. 3.6
By definition:
ALRFilmρ
= ………. Eqn. 3.7
where ρ, L and A are the resistivity, length and cross-sectional area
perpendicular to current flow, respectively. To estimate , R was
measured at a few L values. This was done by placing one terminal at A and
the other at D, and measuring R. The terminal at A was then moved to B
and R measured, and so on to C. A plot of R versus L was then made. As
L→ 0, from Eqn. 3.6 → 0. Hence, at L=0,
ContactsR
R RRFilm Contacts= . of 4 ContactsR
62
samples was measured and found to be similar. The average value is ~38 Ω
and is used as an estimation of the contact resistance in all samples.
Resistivity of Ti films deposited was calculated from measured in DI
water using Eqn. 3.8, which takes into account the effect of thermal
expansion
FilmR
16:
( ) (( RTl
wtTK
2931293
−+⎟⎠⎞
⎜⎝⎛= αρ )) ………. Eqn. 3.8
In Eqn. 3.8, w, l and t are the width, length and thickness of Ti film
patterns, respectively, and R is the resistance measured. The linear thermal
expansion coefficient, α, is taken as 9.7 μ°C-1 17. Figures 3.5 (a) and (b)
show variation of resistivity with thickness and temperature respectively.
Resistivity of Ti films deposited decreases with thickness. In addition, values
determined in the present research are comparable to those obtained from
similarly prepared evaporated Ti thin films18,19.
63
Figure 3.5. Variation of resistivity with (a) thickness and (b) temperature of Ti films. (Evaporation rate 2.5 A/s)
The reduction in resistivity with thickness in metal films has been
investigated by many researchers16,19-21. Mayadas and co-workers modeled
resistivity in polycrystalline films as a combination of 3 types of scattering –
(1) isotropic scattering due to point defects, (2) scattering due to external
surfaces and (3) scattering due to grain boundaries18,19 – and suggested the
last contribution to be dominant. Singh and Surplice observed that gettering
of gaseous impurities such as oxygen in thinner films results in increased
resistivity19. Igasaki and Mitsuhashi16 suggested that reduction in resistivity
with thickness increase is due to grain boundary scattering.
64
Resistivities of Ti films increase linearly with temperature and are almost
parallel for the range of temperature investigated. The rate of resistivity
increase with temperature, ⎟⎠⎞
⎜⎝⎛
dTdρ , ranges from 0.214 to 0.246 μΩcmK-1 and
agrees with values reported in prior studies16,23,24, see Figure 3.6. The
similarity between resistivity and ⎟⎠⎞
⎜⎝⎛
dTdρ values obtained in the present
research and others in the literature can be regarded as an indication of the
experimental validity and accuracy of resistance values measured from Ti
films immersed in DI water.
Figure 3.6. Comparison of TCR obtained from various studies. Data of Igasaki and Mitsuhashi, and Wasilewski were obtained from 300 nm
films and bulk Ti, respectively.
65
3.3.2 Effect of film thickness
Oxidation of Ti films of various thicknesses occurs through nucleation
and growth of the oxide phase. Depending on initial thickness of the Ti film,
oxidation behavior of films during growth can be divided into two sub-stages.
Plots of and XFilmR
R
t with time immersed in aqueous hydrogen peroxide
solution at 80 °C are shown in Figures 3.7 (a) and (b), respectively. Figure
3.7 (c) is a schematic of and XFilm t plots showing different oxidation
behavior for thick and thin films. In the first stage, resistance increases very
gradually while in the second, the rate of resistance increase is significantly
greater. Bearinger et. al.10,11 reported that oxidation of Ti surfaces occurs by
nucleation of titanium oxide domes followed by subsequent growth and
coalescence of domes to form a continuous porous layer. It is postulated
that the two stages observed in the present study correspond to nucleation
and growth processes observed by Bearinger et. al.10,11. Resistance of films
at the end of a reaction, ( )∞FilmR , decreased with increasing thickness. This
is because more Ti ions will dissolve into DI water at the end of a reaction
for thicker Ti films. However, the decrease is low; ( )∞FilmR of 25 and 200 nm
thick films are 1.7 and 1.0 MΩ, respectively.
66
Figure 3.7. Variation of (a) resistance and (b) extent of reaction, Xt, with time immersed in H2O2 solution, and (c) schematic of Xt for thin and thick
films.
During nucleation, little Ti is consumed (or, alternatively, hydrated
titanium oxide is formed) and the rate of change of Xt, dt
dX t , increases very
gradually. For 25 to 50 nm films, nucleation time differs by 110 s. In contrast,
duration of nucleation for 100, 150 and 200 nm is comparable – 362, 344
and 377 s, respectively. Grain size of 100, 150 and 200 nm films is similar,
Figure 3.4 (b), but decreased significantly as thickness decreased from 100
to 25 nm. These observations suggest that nucleation time increases with
grain size.
During the growth stage, variation of Xt with time for 25 and 50 nm Ti
films differs significantly from those of thicker ones. For 25 nm and 50 nm
films, Xt increases linearly with time until the Ti film is fully consumed. For
67
films with thickness 100 nm and greater, variation of Xt is more complicated.
Early in the growth stage before ttrans, Xt increases at a constant rate,
( )dt
transt ttdX <
( )
, which is dependent on grain size - in a similar manner as the
case of duration of nucleation. A plot of grain size and dt
transt ttdX <
( )
against
film thickness is given in Figure 3.8. From this figure, it is clear that structure
of Ti films affect oxidation kinetics. Films with smaller grains have greater
dttranst ttdX < . The effect of grain size on reaction kinetics will be investigated
in greater detail in Section 3.3.3. This dependence on grain size suggest
that dt
t ( )dt
transtdX of 25 and 50 nm films and ttdX < of thicker films are
controlled by a reaction at the interface between the dense-TiO2 to the
hydrated titanium oxide layer. Interface controlled reactions generally are
linear with time25. Dependence on grain size can be rationalized as follows.
Assuming, Ti species diffuse out through the dense-TiO2 layer for reaction to
proceed, flux of Ti species is greater for smaller grains due to grain
boundary diffusion. The increased flux subsequently results in increased
growth rate.
68
Figure 3.8. Variation of grain size and ( )
dtttdX transt < with thickness.
(Evaporation rate 2.5 A/s)
For 100, 150 and 200 nm films, the growth stage can be divided into sub-
stages I and II, see Figure 3.7 (c). Upon reaching a critical value, , at
the subsequent rate of increase of X
transtX
transt t, ( )
dtttdX transt >
trans
t
, decreases
appreciably. for 100, 150 and 200 nm films are 74, 65 and 46 %,
respectively, and corresponds to consumption of 74, 97 and 92 nm thick Ti.
These values as well as , time at which transition in rates took place, are
listed in Table 3.1. For 100, 150 and 200 nm films,
tX
trans
( )dt
ttdX transt > values are
69
similar – ~0.19 s-1. For these films, plots of Xt (and ) for t are
parallel. The similarity of
FilmR t> trans
( )dt
transt ttdX > values for thick films suggests a
common growth mechanism is occurring in these samples. According to the
models of Tengvall3, and Ducheyne and Healy6,7, Ti ions must diffuse
across the hydrated titanium oxide layer to grow. Alternatively, the model of
Bearinger et. al.10,11 requires that oxidants diffuse through the hydrated
titanium oxide layer for growth to take place. One possible mechanism
controlling growth in sub-stage II is mass transport across the growing
hydrated titanium oxide layer. As the layer increase in thickness, a critical
thickness is reached at time t after which mass transport of reactants
across the oxide layer is controlling growth rather than reaction at the
interface. Hence, rate change in sub-stages I and II is probably due to a
corresponding change in mechanism controlling growth.
trans
Table 3.1. Oxidation parameters for Ti films of various thicknesses.
Film thickness
(nm)
Duration of
nucleation (s)
transt (s)
transtX
(%)
(dt
ttdX trant <
(s-1)
( )dt
ttdX transt >
(s-1)
Thickness consumed
at transition
(nm) 25 110 None None 0.61 None None 50 220 None None 0.44 None None
100 362 538 74 0.42 0.18 74 150 344 548 65 0.32 0.18 97 200 377 515 46 0.31 0.20 106
70
Transition from sub-stages I and II occur with increasing thickness of Ti
consumed. Assuming that all Ti consumed goes into forming a hydrated Ti
oxide gel layer, this implies that thickness of the gel layer at which transition
from sub-stages I to II occurs increases with parent Ti thickness. One
possible explanation is porosity of the hydrated Ti oxide gel layer. It is
speculated that hydrated Ti oxide gel layer formed from 200 nm Ti film has a
more open porous structure resulting in diffusion controlled growth at greater
oxide thickness. To prove this hypothesis unambiguously, in-situ probing of
the hydrated Ti oxide gel layer is required. Nevertheless an estimation of
compactness of the hydrated Ti oxide gel layer can be obtained from its
structure after annealing. Figure 3.9 (a) and (b) show SEM images of NST
formed on 150 and 500 nm films after annealing. Clearly, NST formed from
the 500 nm films is more porous. No discernable difference can be observed
between corresponding images of 100 and 200 nm Ti films.
Figure 3.9. NST from (a) 150 and (b) 500 nm Ti films.
71
The nucleation and early growth of hydrated titanium oxide layer can be
described using the kinetic law in the theory of transformation kinetics25.
According to the kinetic law, the percentage of volume transformed (Xt) at
any time t ( ) at time t is given by( )
t
tX t25:
( ) )exp(100100 nt kttX −−= ………. Eqn. 3.9
where k and n are constants. The exponent n is a parameter related to the
mode of transformation13,14,25. Application of this law assumes growth is
linear with time and is valid during the early stage of diffusion controlled
growth. Bearinger et. al.10,11 reported that hydrated titanium oxide domes
grow linearly with time during nucleation. During the growth stage, rate of
increase of Xt is constant indicating that hydrated titanium oxide layer from
25 and 50 nm Ti films grows linearly, see Figure 3.7 (b). In contrast, growth
from 100, 150 and 200 nm films is parabolic overall and involves diffusion
after . Hence, the kinetic law in Eqn. 3.9 was applied only to oxidation
data of 25 and 50 nm films in Figure 3.5 (b). A good fit of the experimental
data to Eqn 3.9 is shown in Figure 3.10. The n values extracted for 25 and
50 nm films are 3.23 and 3.96, respectively. According to the theory of
transformation kinetics, for transformations constrained to two dimensions,
as the case in this study, these values suggest increasing nucleation rate.
trans
72
Figure 3.10. Fitting of kinetic law to experimental data for 25 and 50 nm films
In conclusion, the effects of Ti film thickness on oxidation kinetics can be
summarized as follows. Oxidation of Ti films of various thicknesses occurs
through nucleation and growth stages. dt
dX t is low during nucleation but
increased significantly during growth. For 25 and 50 nm thin films, dt
tdX is
constant during the growth stage. For thicker films, however, the growth
stage can be subdivided into sub-stages I and II. In sub-stage I, growth
occurs due to reaction at the hydrated titanium oxide/dense-TiO2 interface
73
( )dt
transt ttdXwith a constant rate < . Growth in sub-stage I is affected by grain
size of the underlying Ti film. After time t (or in sub-stage II), growth
occurs at a constant but significantly reduced rate
trans
( )dt
transt ttdX > and is
controlled by mass transport through the hydrated titanium oxide layer. Rate
of oxide growth in sub-stage II is independent of grain size but is affected by
porosity of the hydrated titania gel layer.
3.3.3 Effect of grain size
Grain size was found to have a significant effect on oxidation kinetics. To
study the effect of grain size, 50 nm thick Ti films were evaporated at
various deposition rates. Similarly, thickness was verified using X-ray
reflectivity and was found to be within 95% of the nominal values. Films are
highly textured in the (0002) plane as shown in Figure 3.11 (a). Average
grain size was calculated from (0002) peaks using the Scherer’s relation,
Eqn. 3.4, and was found to increase with deposition rate, Figure 3.11 (b).
This trend could be explained on the basis of adatom mobility. As
evaporation rates increase, the mobility of Ti adatoms arriving on the
surface correspondingly increases26,27. More mobile Ti adatoms are able to
migrate farther on a surface and consequently result in increased grain size.
74
Figure 3.11. Plots of (a) X-ray diffraction spectra in Bragg-Brentano geometry for 50 nm films deposited at various rates and (b) average grain
size calculated using Scherer’s relation.
As grain size increases, the rate of change of Xt decreases. Plot of Xt
with time immersed in aqueous hydrogen peroxide solution at 80 °C for the
various films is shown in Figure 3.12 (a). Distinct nucleation and growth
stages are observed. Duration of nucleation increases with grain size.
During growth, the oxidation of Ti proceeds with a constant dt
dX t until
completion. With increasing grain size, dt
dX t was found to decrease, Figure
75
3.12 (b). Rate of change of Xt, dttdX , of samples deposited at 0.2 A/s were
large and not enough data points could be recorded. Results are
summarized in Table 3.2.
Figure 3.12. Variation of (a) extent of reaction, Xt, with time immersed in
H2O2 solution and (b) grain size and dt
dX t with deposition rate.
Table 3.2. Oxidation parameters for 50 nm Ti films of various evaporation rates.
Evaporation
Rate (Å/s)
Grain Size (nm)
Duration of nucleation (s)
dtdX t
(s-1)
0.5 12.5 138 2.1 1.5 21.8 207 0.82 2.5 23.7 220 0.44 3.5 26.1 346 0.56 4.5 28.4 453 0.54
76
Results above indicate that grain size has a significant effect on kinetics
of reaction between hydrogen peroxide solution with 50 nm Ti films. The
effect of grain size could be the result of enhanced grain boundary diffusion
as discussed previously. For hydrated titanium oxide domes to nucleate, Ti
ions must be present on the surface of the dense-TiO2 layer. For the case of
films with small grain size, a higher concentration of Ti ions from the
substrate, due to enhanced grain boundary diffusion. The higher
concentration of Ti ions can affect the nucleation stage in two ways. First,
the nucleation density can be increased and second, the rate of growth of
nuclei will be increased. In both cases, duration of nucleation stage will be
increased.
Similarly, as grain size increases dt
dX t was found to increase. From the
preceding discussion in Section 3.3.2, it was deduced that growth rate for 50
nm films is controlled by reaction at the interface between the hydrated
titanium oxide layer and the dense-TiO2 layer. This assertion is further
supported here. Flux of Ti ions from the film would be expected to be higher
for smaller grain size as a result of grain boundary diffusion. Consequently,
the rate of reaction at the interface would be accelerated, resulting in
increased dt
tdX . This is in agreement with results from Section 3.3.3.
77
3.3.4 Effect of temperature
The effect of temperature can be modeled using the kinetic theory of
transformation. This theory is valid only during linear growth, which in the
present study, is strictly applicable to 25 and 50 nm thick films. Therefore, to
study the effect of temperature, 50 nm thick Ti samples evaporated at 2.5
Å/s were used. Thickness of samples was verified using X-ray reflectivity.
Plots of Xt with time immersed in aqueous hydrogen peroxide solution at
various temperatures are shown in Figure 3.13. As expected, the time to
reach completion of the reaction in shortened. With increasing temperature
duration of nucleation decreases while rate during growth, dt
tdX was found
to increase. Table 3.3. summarizes the effect of temperature on reaction
kinetics.
78
Figure 3.13. Variation of extent of reaction, Xt, with time immersed in H2O2 solution at various temperatures.
Table 3.3. Parameters for 50 nm Ti films oxidized at various temperatures.
Temperature
(°C) Duration of nucleation
(s) dtdX t (s-1)
51.5 795 0.25 60 600 0.29 70 375 0.29 80 220 0.44 90 177 0.68
79
The experimental data can be modeled by the kinetic law, Eqn. 3.9.
Fitting of the law to experimental data is shown in Figure 3.14 (a). n values
extracted range from about 4 to 7 which are in agreement with reported
values in the literature13,14,25. To obtain the temperature dependence of the
oxidation rate, a time τ can be defined at a particular reaction temperature at
which Xt(τ)=constant. Assuming an Arrhenius dependence, the relation
between τ and temperature T is given by:
⎟⎠⎞
⎜⎝⎛=
kTEaexp0ττ ………. Eqn. 3.10
In Eqn. 3.10, 0τ is a constant, k is the Boltzman’s constant, and Ea is
the apparent activation energy for the reaction. We follow the analysis of
Weiss et. al. and used ( ) 5.0=τtX to extract the apparent activation
energies14. Figure 13.4 (b) plots time, 5.0τ , (in logarithmic scale) at which
( ) 5.0=τTX against T1 - the exponent then gives
kTEa . The apparent
activation energy, , for the transformation was calculated to be 0.37 eV. aE
80
Figure 3.14. (a) Fitting of kinetic law to experimental data and (b) plot of 5.0τ
against T1 for 50 nm films oxidized at various temperatures.
The pathways for reaction between Ti(IV) ions with H2O2 are under
active investigation in the literature. Samuni suggested that the mechanism
involves reaction between more than one H2O2 molecules with one Ti (IV)
ion4. Recent theoretical work by Sever and Root postulated that energy
required to bind one H2O2 molecule to a Ti(IV) ion is about 0.17 eV5. As
discussed in Section 3.3.2, for Ti films of thickness less than 50 nm, kinetics
of oxidation is controlled by reaction of H2O2 molecule with Ti(IV) ions at the
hydrated titanium oxide/dense-TiO2 interface. The ratio of apparent
activation energy extracted to the energy required to bind one H2O2
81
molecule to one Ti(IV) ion is 2.2. Hence, it is very plausible that the
mechanism for the interface-reaction controlled kinetics during the growth
stage of 25 and 50 nm films as well as in sub-stage I of 100 – 200 nm films
involves reaction between two H2O2 molecules with one Ti (IV) ion. This
reaction will also occur in sub-stage II during oxidation of 100, 150 and 200
nm films. However, the growth rate in this stage will be controlled by mass
transport of species across the hydrated Ti oxide layer.
3.3.5 Effect of hydrogen peroxide concentration
The oxidation rate of Ti films increases with hydrogen peroxide
concentration as shown in Figure 3.15. For 30, 20 and 15 % (by volume)
hydrogen peroxide solutions, the reaction rate is large and only a few points
were able to be recorded. Nonetheless, it is clear from these plots that the
rate of reaction, dt
tdX, increases with hydrogen peroxide concentration.
Similarly the duration of nucleation is reduced with increasing hydrogen
peroxide concentration. Table 3.4. summarizes the effect of hydrogen
peroxide concentration on reaction kinetics.
82
Figure 3.15. Extent of reaction, Xt, with time immersed in H2O2 solution at 80 °C.
Table 3.4. Parameters for 50 nm Ti films oxidized at various H2O2 concentrations.
Hydrogen peroxide
concentration (volume %)
Duration of nucleation (s) dt
dX t (s-1)
10 220 0.44 15 97 2.57 20 61 3.21 30 42 5.54
83
The effect of hydrogen peroxide concentration suggests that the species
diffusing across the hydrated titanium oxide layer is hydrogen peroxide
molecules. This suggestion is borne out from Fick’s First Law which states
that:
xC
DJ OHOHOH ∂
∂−= 22
2222………. Eqn. 3.11
In Eqn. 3.11, , and denotes flux, diffusion coefficient
and concentration of hydrogen peroxide molecules, respectively. After
nucleation and after the oxide domes have coalesced to form a continuous
hydrated titanium oxide layer, a concentration gradient is set up,
22OHJ22OHD
22OHC
xOH
∂22
C∂,
across the hydrated titanium oxide layer. Assuming diffusion in the aqueous
solution is large, which is a reasonable in a solution, then concentration of
hydrogen peroxide molecules at the hydrated titanium oxide layer/solution
interface is equal to that in the bulk. Hence, the flux of hydrogen peroxide
molecules diffusing across the oxide layer increases with its bulk
concentration, in accordance with Eqn. 3.11. This explains the increase in
oxidation rate with increasing hydrogen peroxide concentration, Figure
13.15. This explanation also agree with the observation of growth occurring
at the hydrated titanium oxide layer/dense-TiO2 layer interface rather than at
84
the solution hydrated titanium oxide layer/solution interface as postulated by
Bearinger et. al. 10,11.
3.4 Phenomenological model of Ti oxidation in aqueous
hydrogen peroxide
From the above discussions, a phenomenological model of Ti oxidation
in aqueous hydrogen peroxide can be proposed, see Figure 3.16. Oxidation
of Ti surfaces occurs in two distinct stages, namely nucleation and growth.
Nucleation involves formation of hydrated titanium oxide domes on the
dense-TiO2 layer. These domes grow laterally and finally coalesce to form a
thin continuous hydrated titanium oxide layer. During the nucleation stage,
rate of consumption of the underlying Ti films is low. Growth of domes
occurs at the interface between the dense-TiO2 layer and the hydrated
titanium oxide layer by reaction between hydrogen peroxide molecules with
Ti ions.
Kinetics during the growth stage of thin (≤100 nm) Ti films is different
from that of thick ones. For thin films, growth occurs at a constant rate, dt
tdX ,
until the Ti substrate is fully consumed. In contrast, the growth of thick films
occurs in two distinct stages – sub-stages I and II. During sub-stage I, rate
85
( )dt
transtof growth, ttdX <
( )
, of the hydrated titanium oxide layer is controlled by
reaction between hydrogen peroxide molecules with Ti ions at the interface
between the dense and hydrated titanium oxide layers. Growth rate in sub-
stage I is sensitive to the structure of the underlying Ti substrate. As
reaction proceeds and thickness of the hydrated titanium oxide layer
increases, a transition thickness is reached at time t , at which mass
transport of hydrogen peroxide molecules across the hydrated titanium
oxide layer, rather than reaction at the interface, controls growth of the
hydrated titanium oxide layer. At , the rate of oxide growth,
trans
transtdt
transt ttdX > ,
decreases significantly. ( )
dttranst ttdX < is insensitive to parent Ti
microstructure but is affected by compactness of the hydrated titanium oxide
layer. A smaller transition thickness was observed for more compact of
hydrated titanium oxide layer formed from thinner Ti films.
86
Figure 3.16. (a) to (c) Schematics of processes occurring during oxidation of Ti thin films in aqueous hydrogen peroxide solution. (d)
Variations of extent of reaction, Xt, for thin and thick films; processes occurring at various stages of oxidation are labeled accordingly. ( represents Ti flux from the substrate due to grain boundary diffusion)
GBTiJ
87
3.5 Conclusions
In conclusion, the reaction kinetics of Ti thin films with hydrogen peroxide
solution has been investigated experimentally. Extent of reaction, Xt, was
monitored by measuring the resistance of the residual Ti film. It was found
that oxidation of Ti films occurs by nucleation and growth mechanism. Both
grain size and thickness of films affects the oxidation kinetics. Grain size
was found to affect both nucleation and growth stages. Films with finer grain
size have shorter nucleation period and higher growth rate.
During oxide growth, oxidation rate of films (25 and 50 nm) occurs at a
constant rate until the Ti films are completely consumed. For thicker films
(100, 150 and 200 nm films), growth rate decreases after a certain thickness
of porous titania has been formed. The oxide thickness at which the rate
reduction occurs is dependent on porosity of the oxide layer. Change in
oxidation rate is attributed to a change in the mechanism controlling growth
of the oxide layer. During the initial period of growth (sub-stage I) growth of
oxide is controlled by reaction of Ti species with hydrogen peroxide
molecules. During the later stage (sub-stage II), diffusion of hydrogen
peroxide molecules through the oxide layer is the rate controlling
mechanism. A phenomenological model of Ti oxidation in aqueous hydrogen
peroxide solution that takes into account the effect of grain size and
thickness of Ti films is proposed which explains the above observations and
is consistent with recent reports in the literature.
88
3.6 References
1. L. L. Sheir, Corrosion, Volume 1, Newnes-Butterworth, London (2000).
2. K. J. Vetter, Elektrochemische Kinetik, Springer, Berlin (1961), as cited by A. M. Shams el din, A. A. Hammoud, Thin Solid Films, 167, 269 (1988).
3. P. Tengvall, “Titanium-hydrogen peroxide interaction with reference
to biomaterial applications”, PhD thesis, Department of Physics and Measurement Technology, Linköping University, Sweden (1989).
4. A. Samuni, Journal of Physical Chemistry, 76, 634 (1972).
5. R. S. Sever, T. W. Root, Journal of Physical Chemistry B, 107, 4090
(2003).
6. K. E. Healy, P. Duchyene, Biomaterials, 13, 553 (1992).
7. K. E. Healy, P. Duchyene, Journal of Colloid and Interface Science, 150, 404 (1992).
8. J. Pan, D. Thierry, C. Leygraf, Journal of Biomedical Materials
Research, 107, 4090 (2003). 9. J. Pan, D. Thierry, C. Leygraf, Electrochimica Acta, 41, 1143 (1996).
10. J. P. Bearinger, C. A. Orme, J. L. Gilbert, Journal of Biomedical
Materials Research Part A, 67A, 702 (2004).
11. J. P. Bearinger, C. A. Orme, J. L. Gilbert, Surface Science, 490, 371 (2001).
12. S. P. Murarka, Metallization: Theory and Practice for VLSI and ULSI,
Butterworth-Heinemann, Boston (2001).
13. F. Nava, T. Tien, K. N. Tu, Journal of Applied Physics, 57, 2018 (1985).
14. B. Z. Weiss, K. N. Tu, D. A. Smith, Journal of Applied Physics, 59,
415 (1986).
89
15. J. K. Howard, R. F. Lever, D. A. Smith, P. S. Ho, Journal of Vacuum Science and Technology, 13, 68 (1976).
16. Y. Igasaki, H. Mitsuhashi, Thin Solid Films, 51, 33 (1978). 17. E. S. Greiner and W. C. Ellis, Transactions of the American Institute
of Mining and Metallurgical Engineers, 180, 657 (1949).
18. F. Huber, IEEE Transaction on Component Parts, CP-11, 38 (1964)
19. B. Singh and N. A. Surplice, Thin Solid Films, 10, 243 (1972).
20. A. F. Mayadas, M. Shatzkes, Physical Review B, 1, 1382 (1970).
21. A. F. Mayadas, R. Feder, R. Rosenberg, Journal of Vacuum Science and Technology, 6, 690 (1969).
22. B. D. Cullity, Elements of X-ray Diffraction, Addison-Wesley, Reading
(1959).
23. R. J. Wasilewski, Transactions of the American Institute of Mining and Metallurgical Engineers, 224, 5 (1962).
24. M. E. Day, M. Delfino, J. A. Fair, W. Tsai, Thin Solid Films, 254, 285
(1995).
25. J. W. Christian, The Theory of Transformations in Metals and Alloys, Pergamon Press, Oxford (1965).
26. C. R. M. Grovenor, H. T. G. Hentzell, D. A. Smith, Acta Metallurgica,
32, 773 (1984).
27. R. Messier, A. P. Giri, R. A. Roy, Journal of Vacuum Science and Technology A, 2, 500 (1984).
90
Chapter 4: Gas sensing using nanostructured TiO2
elements
4.1 Introduction
An electronic nose is, by definition, an intelligent chemical sensor array
system for odor classification. The concept of an electronic nose comes
about due to the increasing demand for electronic analogues of the human
olfactory system that are able to rapidly process, at relatively low-cost, odor
information for various applications ranging from food processing to
environmental monitoring and protection1. Similarly, electronic tongues are
counterparts of electronic noses that operate in liquid. In general, electronic
noses and tongues consist of sampling systems (for collecting samples),
arrays of chemical sensing elements (for detecting chemical species),
electronic circuitry (for transferring information collected by the sensing
elements and for heating sensing elements to the operating temperature)
and the data processing system. Among these components, chemical
sensing elements are undoubtedly the most critical as these largely dictate
the sensitivity and selectivity of the device — and which is the subject of this
chapter.
91
Chemical sensing elements may be categorized by the type of materials
used as either inorganic crystalline materials, organic materials or
biologically-inspired materials. Inorganic crystalline materials include
semiconductors such as metal oxides, silicon or carbon nanotubes.
Inorganic materials are particularly attractive as they hold the promise of
being easily integrated into microelectronic devices. Once material
integration has been achieved, it is then possible to leverage on the
enormous microelectronics fabrication infrastructure to produce electronic
noses and tongues in large volume at low-cost. Consequently, there has
been extensive effort towards integrating inorganic nanostructured materials
in devices. The mechanism of operation of a sensing element is the
transduction of interactions occurring between chemical species with its
surface to a measurable change in at least one property of the element. In
most cases, the measured property is resistance, capacitance or
temperature. Since nanostructured materials have large surface to volume
ratio, surface events result in a large change of a material property.
One approach to integrating nanostructured materials into devices is the
use of discrete nanostructures such as tubes or wires. Recent studies have
focused on single-walled carbon nanotubes2, silicon nanowires3, palladium
mesowire arrays4, metal oxide nanowires5,6 and polymeric nanowires7,8. As
a first step towards realization of electronic noses and tongues, the chemical
92
sensing properties of these materials have been characterized. However,
the challenge of locating and manipulating these nanostructures, fabricated
by “bottom-up” approaches, onto desired locations on chips impedes their
utilization in “real world” devices. To address this challenge, many
imaginative methods have been proposed such as electric9 and magnetic10
field induced or fluidic-assisted alignment together with surface patterning
techniques11-13, the fabrication of silicon nanowires from a macroscopic
silicon layer on an insulator14, self-organization or template synthesis of
nanostructures, and so on. Another approach involves the development of
processes to tailor, at the nanoscale, morphology of metal oxides, such as
titania15, zinc oxide16 and tin oxide17.
In this vein, we describe a new approach for fabricating micrometer-scale
sponge-like structures consisting of interconnected nanoscale walls or wires.
Configured as active elements of conductometric sensing devices, these
structures possess the ultra-high sensitivity of nanostructures while
benefiting from the enormous capabilities of micrometer-scale fabrication
that makes them amenable to integration into practical devices. We
demonstrate this approach by fabricating a gas sensor utilizing arrays of
NST pads as sensing elements on both Si and Kapton®. The latter is a
commercially available organic flexible substrate which is relatively low cost.
93
4.2 Gas sensing using nanostructured metal oxides
The use of metal oxides in electronic chemical sensing devices dates
back to the late 1950’s when Kiukkola and Wagner18 as well as Weissbart
and Ruka19 used ZrO2 as solid electrolytes. Since then, there have been
tremendous developments in metal oxide based sensors20. Titanium dioxide
(TiO2) has been widely studied for applications as sensing elements21-24. It
is attractive since its electrical conductivity and surface properties can be
modified by judicious use of surface activation and dopants25-28. TiO2 can
also be made porous and the pore structure controlled, allowing
functionalization with biomolecules29. In recent reports nanostructured
titania (NST) was used as a sensing element in ultra-sensitive sensors30,31.
Unlike conventional sensors using discrete nanotubes and nanowires, NST
has a sponge-like morphology with interconnected walls that act as
percolating channels for electron (hole) transport. This makes NST
particularly suitable as sensing elements. Finally, the approach described
has a high process yield and is compatible with current microchip
manufacturing practices and specifically photo-lithography which can place
an NST element at any predetermined locations on a chip.
94
4.3 Integration of nanostructured TiO2 as sensing elements
on silicon
Figure 4.1. Schematic process flow for fabrication of patterned integrated NST arrays.
Compared to other methods of forming ns-titania, the technique
developed for integrating NST is compatible with current microsystems
device manufacturing practices. As a demonstration of this compatibility, we
have fabricated gas sensor arrays with 20 μm ns-titania pads as sensing
elements. The detailed description for forming NST has been given in
Chapter 2. Here, only a brief description of the process to integrate NST as
95
sensing elements is given. The fabrication process of NST pad arrays is
summarized in Figure 4.1. A 1 μm thick SiO2 (T- SiO2) layer was thermally
grown on a 4 inch N-type Si(100) wafer at 1100 °C. A 500 nm thick Ti film
was then deposited using electron-beam evaporation. Subsequently, a 1 μm
thick SiO2 (P-SiO2) layer was deposited at 250 °C using plasma-enhanced
chemical vapor deposition (PECVD) using silane (SiH4) and nitrous oxide
(N2O) precursors. The patterns on the P-SiO2 layer were then etched using
CHF3 gas to expose Ti surfaces. These exposed Ti surfaces were then
oxidized by aging in aqueous 10 % hydrogen peroxide solution at 80 °C for
2.5 hrs. The samples were annealed at 300 °C for 8 hrs. Subsequently
Ti(10nm)/Pt(250nm) electrodes were then deposited using electron beam
evaporation.
Figures 4.2 (a) and (b) show the layout of a prototype gas sensor utilizing
NST pads as sensing elements consisting of a 5 by 5 array of square NST
pads, each 20 μm wide. In principle, each of these pads could be
individually addressed. For simplicity we have metallized only a few,
selected pads of the 5 by 5 array with 10 μm wide metal lines. Moreover,
further scaling down of the device is feasible with currently available
microelectronics process tools.
96
Figure 4.2. Optical micrographs of integrated NST arrays as sensing
elements in prototype gas sensors with 2 different configurations; (a) single pad metallized and (b) 3 pads metallized in series.
Although the NST pads are porous, the evaporated Ti/Pt electrodes have
good step coverage on the masking oxide, are continuous on the NST
elements and give satisfactory electrical contacts. Figures 4.3 (a) to (d) are
SEM micrographs of Ti/Pt electrodes on NST pads deposited before and
after NST annealing. Although Ti/Pt metallization has good step coverage
on the sidewalls of the PECVD masking oxide, as shown in Figure 4.3,
97
shrinkage of NST during annealing results in discontinuity of the
metallization lines at the corner of the masking oxide face with the NST
surface. However, this problem was not observed when Ti/Pt metallization
was deposited after NST pads have been annealed.
Figure 4.3. Continuity of Ti/Pt metallization layers deposited (a and b) before
and (c and d) after NST annealing.
98
The surface of Ti/Pt electrode layer is rough due to the underlying porous
NST morphology but is nevertheless continuous over the NST pad — see
Figures 4.4 (a) to (c). Electrode continuity is critical as it reduces electrical
contact resistance.
Grain size of Pt crystals on the NST layer is significantly larger than
those deposited on the P-SiO2 mask oxide. The average grain size of Pt
crystals on the NST and P-SiO2 mask oxide are ~35 nm and ~20 nm,
respectively. This smaller Pt grain size on NST pads could be the result of a
difference in surface density of Pt nuclei over the different materials during
evaporation. On P-SiO2 mask oxide, a larger nuclei surface density would be
expected. This is because, on the NST, nuclei can only be formed on the
surface of the pore walls. Because of the smaller nuclei density on the NST,
Pt crystals are able grow to over a wider distance before coalescing with an
adjacent grain, resulting in larger Pt grains on NST pads.
Ti/Pt electrodes deposited on the NST sensing elements do not
penetrate through the entire thickness of the NST layer as revealed by
cross-sectional SEM and TEM analysis as shown in Figures 4.5 (a) and (b).
The procedure for TEM sample preparation has been described in detail in
Chapter 2. In brief, TEM samples were micromachined using a focus ion
beam system using a Ga ion beam current of 100 pA. An FEI Sphera T20
machine operating at 200 kV was used for Transmission Electron
99
Microscopy analysis. Superficial penetration of metallization layers into the
NST pads is expected because electron beam deposition is a line of sight
process. Hence the metal layers deposited could only penetrate as deep as
the pores on the top surface of the NST. Hence the volume of NST under
the electrode is porous and remains unfilled. Similar observations of partial
penetration of metals evaporated on porous Si have been reported. In
contrast, complete infiltration of metal contacts has been obtained using
electroless deposition — this is the subject of a later chapter.
Figure 4.4. SEM micrographs of showing continuity of Ti/Pt electrodes
evaporated on single NST pads. Regions labeled in (a) represents location from which micrographs (b) to (d) were obtained.
100
Figure 4.5. Superficial penetration of Ti/Pt metallization into NST as
revealed by cross-sectional (a) SEM and (b) TEM analysis.
4.4 Fabrication of nanostructured TiO2 on Kapton®
Low temperature synthesis of NST was demonstrated by fabricating gas
sensing devices on Kapton® 300 HN foils (Kapton® foils were gifts from
DuPont). There are a few reasons why Kapton® was chosen. First, Kapton®
is suitable for use as a gas sensing substrate. Kapton® films are able to
101
withstand relatively high temperatures of up to 400 °C for at least 2 hrs. In
addition, Kapton® films are relatively inert to chemicals used during
processing, have low moisture absorption and high dielectric strength.
Secondly, Kapton® foils are widely used in industry and hence suitable for
large scale implementation.
Figure 4.6. Integration of NST on Kapton®: (a) Arrays of Ti window after SiO2 etch; (b) A complete prototype gas sensing device.
The processing steps described for integration of NST into devices on Si
are directly applicable to Kapton® substrates. However, as Kapton® films
are flexible, one difficulty encountered is the bending of substrates after
evaporation when Ti films are deposited. This issue however, could be
reduced by using sputtering where stresses in films are more readily
controlled. Stress in evaporated Ti is tensile while SiO2 films deposited by
102
plasma enhanced chemical vapor deposition are under compressive stress.
It was found that for 500 nm Ti films deposited on Kapton® 300 HN foils, an
overlying 100 nm thick SiO2 layer would produce a flat specimen. Figure 4.6
(a) shows an array of 20 μm square pads after etching of the SiO2 layer. As
expected, no delamination of Ti from the underlying Kapton® substrate was
observed. A complete gas sensing device with evaporated metallization is
shown in Figure 4.6 (b). A tip of a pen is included for scale.
4.5 Results and discussion
Sensing experiments were carried out in a vacuum chamber
equipped with microprobe contacts for current-voltage (I-V) measurements.
The gas sensing behavior of NST to oxygen and hydrogen gases were
studied. The partial pressure of the gas to be detected in the test chamber
was controlled by means of a pulsed needle valve and confirmed using an
ion gauge. A halogen lamp heater was used to raise the sensor assembly to
the desired temperature, which was measured using a K-type thermocouple.
Prior to sensing experiments, the device was exposed to UV for about 5 min
an elevated temperature in house vacuum and then flushed with nitrogen
gas for about 10 min. This cleaning cycle is done for about 45 min every
time a new device is put under test or after long periods of exposure to air.
103
4.5.1 Oxygen sensing
Figures 4.7 (a) and (b) show I-V characteristics of an NST pad in vacuum
and under oxygen exposure at various temperatures and O2 partial
pressures ranging from 0.3 to 0.65 mTorr. The linear I-V characteristics
obtained indicate that contacts between the Ti/Pt electrodes and the NST
pads were Ohmic at the temperatures utilized. The conductance of the NST
pads is very sensitive to the presence of oxygen and changes monotonically
with oxygen partial pressure. In the presence of O2, conductance of the pad
is approximately an order of magnitude less than that in vacuum. This high
sensitivity to O2 is also illustrated in Figure 4.7 (b) where O2 pressure
variations in the sub-mTorr range were easily distinguishable. Assuming O2
to be entrained in an unreactive gas such as nitrogen, these mTorr pressure
variations at 523 K correspond to ppm detectivity by an NST pad. By using
narrower and longer NST pads the detectivity could be enhanced to the ppb
level. Reducing the 20 μm square pads used in this prototype device is
feasible with current process tools.
104
Figure 4.7. I-V characteristics of an NST pad in vacuum and under oxygen exposure at various temperatures and O2 partial pressures ranging from 0.3
to 0.65 mTorr.
105
Figure 4.8. Sensing characteristics an NST pad element: (a) Response to oxygen cycles; (b) sensitivity to oxygen at various temperatures.
106
The response time and sensitivity of the sensor for oxygen detection was
obtained by measuring resistance changes when oxygen is introduced
cyclically in the chamber — results of these studies are shown in Figures 4.8
(a) and (b). Upon exposure to oxygen gas at 0.8 mTorr at 473 K the
resistance increased from 10 to 90% of its maximum value in 48 s. This
value, which will be referred to as the response time is comparable to those
of other oxygen sensing devices utilizing porous undoped-titania as a
sensing material. Sharma et. al. have fabricated porous undoped titania
films using various techniques. The resulting titania films have pore size
distribution ranging from ~0.5 to 1.5 μm and 0.75 to 2.5 μm35,36. However,
the response times they reported were obtained at higher temperatures. To
compare these response times with the values obtained in this study we
assumed that response times depend on temperature in an Arrhenius
fashion. For the titania films with ~0.5 to 1.5 μm and 0.75 to 2.5 μm pores,
the response times at 473 K were calculated to be 49 and 20 s, respectively
– see Figure 4.9. The sensitivity of the sensor is defined as the ratio of the
maximum resistance in the presence of oxygen to the resistance in the
absence of oxygen. The best sensitivity of the NST-based sensor, which
was measured to be 60 at 473 K, is superior to those of undoped-titania
based sensors reported in the literature. Sharma et. al. and Gao et. al.
reported sensitivity values of 1.43 and ~30 at their higher operating
temperatures35-37. The sensitivities expected for those systems at 473 K
107
would be much lower. Hence the oxygen sensitivity of sensors fabricated in
this study is at least twice that reported by Sharma et. al.. Another
advantage of the present approach to integrating NST into microsystems
devices is the low level of drift in the electrical properties of the NST sensor
with time, despite the low processing temperatures involved. This could be
due to the high crystallinity of the NST pads formed as indicated by electron
and X-ray diffraction data.
Figure 4.9. Comparison of response time for oxygen gas at 273 K of NST-based sensor to those of other devices based on porous undoped titania.
108
Undoped TiO2 shows n-type semiconducting behavior due to the
presence of shallow intrinsic donors22,38. Below 1273 K, the dominant donor
species are singly and doubly ionized oxygen vacancies, with associated
defect levels located just below the conduction band22,38. For TiO2-based
sensors, the resistance change upon O2 exposure is due to the interplay
between the ionic chemisorption of oxygen at the TiO2 surface and the
density of carriers in the material’s conduction band. Upon adsorption, an O2
molecule migrates to an oxygen vacancy site39, where it interacts with
surface vacancies eventually forming chemisorbed species such as O2- and
O- (depending on temperature). The process involves electron transfer
across the titania surface depleting the electron density of the bulk TiO2. The
net reduction in conduction is due to annihilation of the donor states. At
elevated temperatures the oxygen-induced vacancy annihilation process
could involve subsurface and even bulk donors35,36. Hence, for enhanced
sensing performance, materials with structures that allow permeation of
oxygen through a porous sensing element with a large surface area and thin
channels are desirable. For conventional TiO2 dense film-type sensors, the
compact nature of the particles means that only a thin layer of the film close
to the surface can effectively be used for sensing. NST on the other hand
has thin pore wall thickness of 25 – 75 nm, which means that a significantly
greater TiO2 surface area can be used for sensing as the open porous
109
structure allows enhanced oxygen gas permeation. As compared with
previously reported porous TiO2 films35, the pore size of NST used in this
study, ranging from 50 nm to 200 nm, is ~10 times lower and hence is
expected to have reduced oxygen gas permeation. However, the response
times at 523 K are similar. In addition, the NST-based O2 sensor exhibits
superior sensitivity. These observations suggest that enhanced oxygen
sensing performance of NST is due to a higher level of surface reactivity,
possibly due to presence of a greater number of surface defects.
Another possible mechanism that enhances the sensitivity of the sensors
is the sponge-like structure of NST. As discussed, for chemical sensing to
take place the presence of chemical species on the NST surface must be
transduced into an electronic signal. In the case of NST sensors, the
sponge-like morphology significantly facilitates electron percolation through
the TiO2 pad resulting in an increase in current. In the case of sensors
employing pads consisting of agglomerations of TiO2 nanoparticles, electron
percolation may not occur as well. This explanation is plausible as
analogous observations had been made in the case of solar cells using TiO2
electrodes40. In that case, it was found that current harvested from solar
cells was increased when TiO2 electrodes were treated with TiCl4. This
increase was due to the formation of interparticle neck growth that facilitates
electron conduction.
110
4.5.2 Hydrogen sensing
NST-based sensor was also used to detect hydrogen gas. The I-V
characteristics obtained at room temperature in the absence and presence
of hydrogen gas are shown in Figure 4.10 (a). The concentration of
hydrogen used was partial pressure 0.8 mTorr which is equivalent to ~1
ppm. In the presence of hydrogen, current is increased since hydrogen is a
reducing gas, as expected. The current-time response of the device to
hydrogen at a partial pressure of 0.8 mTorr at 300 °C is shown in Figure
4.10 (b) which indicates poisoning of the NST element. Upon introduction of
hydrogen, the resistance measured decreased to a ‘minimum’ value and,
when the hydrogen valve is switched off, increased to a ‘maximum’ value
and saturated. Upon further introduction of hydrogen, the resistance drops
to a ‘minimum’ which is lower than the ‘minimum’ value of the preceding
cycle. Similarly, when hydrogen gas flow was switched off, resistance
increased and saturated again at a ‘maximum’ value which is lower than the
‘maximum’ value of the preceding cycle.
Response time and sensitivity of the sensor towards hydrogen were
calculated from Figure 4.10 (b). Varghese et. al.41 and Patel et. al.42 have
recently reported hydrogen sensing studies using undoped titania. In these
studies, response time was defined as the time required for the sensor to
111
reach 90% of the maximum resistance and sensitivity is defined by the
formula:
gs
gs
RRR
S−
= 0 …….Eqn. 4.1
where and are the resistances in the absence and presence of
hydrogen gas, respectively. These definitions are different from that those
discussed in the section on oxygen sensing. However, for the sake of
making comparisons to the work of Varghese et. al.
0R gsR
41 and Patel et. al.42
(instead of sensitivity, this parameter is called % resistance change in the
work of Patel et. al.42) the definitions proposed by these researchers are
used in this section. From results shown in Figure 4.10 (b), it was found that
the response time and sensitivity of the NST based sensor to hydrogen are
78 s and 3.7%, respectively.
Using titania nanotubes as sensing elements, Varghese et. al.41
studied sensing down to 100 ppm hydrogen ambients. However, it was
found that sensitivity has a logarithmic relation with hydrogen concentration
and it was calculated from their data that the sensitivity for 1 ppm hydrogen
at 290 °C is 2 which is comparable to the present work. The corresponding
value reported by Patel et. al.42 at 300°C is 1. Varghese et. al.41 reported a
response time for ~350 s for 100 ppm hydrogen ambient – response time for
112
1 ppm hydrogen ambient was not reported. Patel et. al.42, did not report of
response time values in their work.
The mechanism of transduction of hydrogen adsorption on NST
surface to a change in electrical resistance is still an open question. There
are three possible mechanisms all of which can occur, independently of
each other. First, hydrogen molecules adsorb to the NST surface and
dissociate to atomic hydrogen which are subsequently chemisorbed. The
hydrogen atoms then donate an electron to the conduction band of the
titania causing the observed reduction the resistance. The protons can be
trapped as surface states or diffuse into the NST lattice. Second, an effect
known as spill over may occur. In spill over, molecular hydrogen is adsorbed
on the surface of the Pt electrodes and dissociates into atomic hydrogen.
These hydrogen atoms diffuse onto the NST surface and contribute to the
electrical conduction as mentioned above. Third, hydrogen molecules are
adsorbed onto the NST surface and dissociates into atomic hydrogen.
These atomic hydrogen then reacts with oxygen species such as O- or O-2
ions already adsorbed on the surface to form H2O. This reaction removes
electron traps from the surface causing a drop in the resistance observed.
Hydrogen sensing measurements were done in oxygen-free
environments, hence the increase in conductivity cannot be due to removal
of oxygen O- or O-2 ions by hydrogen. Also, the extremely fast response of
the NST sensor to changes in hydrogen concentration suggests that surface
113
effects are dominant. Upon dissociation of molecular hydrogen, hydrogen
atoms are chemisorbed on the titania surface. Chemisorbed atomic
hydrogen act as a surface state and charge transfer occurs from hydrogen
to the conduction band of titania resulting in an increase in conductivity. The
poisoning of titania by hydrogen, as indicated by the continuous drops in
‘minimum’ and ‘maximum’ values of different cycles could be explained by
trapping of hydrogen in bulk states. Chemisorbed hydrogen atoms may
diffuse into the bulk titania and form donor type OH-defects that join two
oxygen atoms43. These defects are not removed when the hydrogen gas
flow is switched off which results in the lowering of the ‘minimum’ resistance
from that in the preceding cycle.
114
Figure 4.10. Hydrogen sensing: (a) I-V characteristics of an NST pad in vacuum and under hydrogen exposure (8x10-4 Torr) at room temperature and their differential; (b) Response to hydrogen cycles at 300 °C (8x10-4
Torr).
115
4.6 Conclusions
In conclusion, we have described a method which is compatible with
current microelectronics manufacturing practices for forming an array of
crack-free patterned and integrated NST (NST) pads for sensor-on-a-chip
and other micro- and nanodevice applications. Using patterned Ti thin films
with lateral dimensions below a certain threshold, crack formation in NS-
titania layers can be eliminated. NST has sponge-like morphology with pore
diameter and wall thickness in the ranges of ~25 – 200 nm and 25 - 75 nm,
respectively. Using these integrated NST pad arrays as resistive
components, we successfully fabricated a prototype oxygen sensor capable
of detecting oxygen and differences in oxygen concentration at ppm levels.
The sensor operates at lower temperatures, has a fast response time and
superior sensitivity relative to oxygen sensors based on undoped-titania.
116
4.7 References
1. P. Gouma, G. Sberveglieri, R. Dutta, J. W. Gardner, E. L. Hines, MRS Bulletin, 29, 697 (2004). (see also articles cited therein)
2. J. Kong, N. Franklin, C. Zhou, M. G. Chapline, S. Peng, K. Cho, Dai, H., Science, 287, 622 (2000).
3. Y. Cui, Q. Wei, Q., H. Park, C. M. Lieber, Science, 293, 1289 (2001).
4. F. Favier, E. C. Walter, M. P. Zach, T. Benter, R. M. Penner, Science, 293, 2227 (2001).
5. A. Kolmakov, Y. Zhang, G. Cheng, M. Moskovits, Advanced Materials, 15, 997 (2003).
6. Y. Wang, X. Jiang, Y. Xia, Journal of the American Chemical Society, 125, 16176 (2003).
7. S. Boussaad, N. J. Tao, Nano Letters, 3, 1173 (2003).
8. H. Liu, J. Kameoka, D. A. Czaplewski H. G. Craighead, Nano Letters, 4, 671 (2004).
9. P. A. Smith, C. D. Nordquist, T. N. Jackson, T. S. Mayer, B. R. Martin, T. Mbindyo, T. E. Mallouk, Applied Physics Letters, 77, 1399 (2000).
10. M. Tanase, L. A. Bauer, A. Hultgren, D. M. Silevitch, L. Sun, D. H. Reich, P. C. Searson, G. J. Meyer, Nano Letters, 1, 155 (2001).
11. Y. Huang, X. Duan, Q. Wei, C. M. Lieber, Science, 291, 630 (2001).
12. Z. Deng, C. Mao, C., Nano Letters, 3, 1545 (2003).
13. S. Jin, D. Whang, M. C. McAlpine, R. S. Friedman, Y. Wu, C. M. Lieber, Nano Letters, 4, 915 (2004).
14. Z. Li, X. Chen, X., T. I. Kamins, K. Nauka, R. S. Williams, Nano Letters, 4, 245 (2004).
15. A. S. Zuruzi, N. C. MacDonald, Advanced Functional Materials, 15, 396 (2005).
16. X. L. Cheng, H. Zhao, L. H. Huo, S, Gao, J. G. Zhao, Sensors and Actuators B, 102, 248 (2004).
117
17. C. Nayral, T. Ould-Ely, A. Maisonnat, B. Chaudret, P. Fau, L. Lescouzeres, A. Peyre-Lavigne, Advanced Materials, 11, 61 (1999).
18. K. Kiukkola, C. Wagner, Journal of The Electrochemical Society, 104, 379 (1957)
19. J. Weissbart, R. Ruka, Review of Scientific Instruments, 32, 593 (1961).
20. C.O. Park, S. A. Akbar, Journal of Materials Science, 38, 4611 (2003).
21. T. Y. Tien, H. L. Stadler, E. F. Gibbons, P. J. Zacmanidis, American Ceramic Society Bulletin, 54, 280 (1975).
22. W. Gopel, G. Rocker, R. Feierabend, R., Physical Review B, 1983, 28, 3427.
23. A. Takami, American Ceramic Society Bulletin, 67, 1956 (1988).
24. K. D. Schierbaum, U. K. Kirner, J. F. Geiger, W. Gopel, Sensors and Actuators B, 4, 87 (1991).
25. A. Ruiz, J. Arbiol, A. Cirera, A. Cornet, J. R. Morante, Materials Science and Engineering C, 19, 105 (2002).
26. H. Lin, S. Kumon, H. Kozuka T. Yoko, Thin Solid Films, 315, 266 (1998).
27. N. Bonini, M. C. Carotta, A. Chiorino, V. Guidi, C. Malagu, G. Martinelli, L. Paglialonga, M. Sacerdoti, M., Sen. Actuators B, 68, 274 (2000).
28. M. Ferroni, M. C. Carotta, V. Guidi, G. Martinelli, F. Ranconi, M. Sacerdoti, E. Traversa, Sensors and Actuators B, 77, 274 (2001).
29. V. Faust, F. Heidenau, J. Schmidgall, F. Stenzel, G. Lipps, Ziegler, G., Key Engineering Materials, 206-2, 1547 (2002).
30. G. Sberveglieri, L. E. Depero, M. Ferroni, V. Guidi, G. Martinelli, P. Nelli, C. Perego, L. Sangaletti, Advanced Materials, 8, 334 (1996).
31. O. Varghese, D. Gong, M. Paulose, K .G. Ong, E. Dickey, C. G. Grimes, Advanced Materials,15, 624 (2003).
32. P. Tengvall, Titanium-Hydrogen Peroxide Interaction with Reference to Biomaterial Applications, PhD Thesis, Department of Physics and Measurement Technology, Linköping University, Sweden (1989).
33. J. M. Wu, S. Hayakawa, K. Tsuru, A. Osaka, A. Scripta Materialia, 46, 101 (2002).
118
34. J. W. Hutchinson, Z. Suo, Advances in Applied Mechanics, 29, 63 (1992).
35. R. K. Sharma, M. C. Bhatnagar, G. L. Sharma, Sensors and Actuators B, 46, 194 (1998).
36. R. K. Sharma, M. C. Bhatnagar, Sensors and Actuators B, 56, 215 (1999).
37. L. Gao, Q. Li, Z. Song, J. Wang, Sensor Actuat. B –Chem, 71, 179 (2000).
38. U. Diebold, Surf. Sci. Rep., 48, 53 (2003).
39. R. Schaub, E. Wahlstrom, A. Ronnau, E. Laegsgaard, I. Stensgaard, F. Besenbacher, Science., 299, 377 (2003).
40. C. J. Barbé, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V. Shklover, M. Grätzel, Journal of the American Ceramic Society., 80, 3157 (1997).
41. O. K. Varghese, D. W. Gong, M, Paulose, K. G. Ong, C. A. Grimes, Sensors and Actuators B, 93, 338 (2003).
42. S. V. Patel, K. D. Wise, J. L. Gland, M. Zanini-Fisher, J. W. Schwank, Sensors and Actuators B, 42, 205 (1997).
43. A. von Hippel, J. Kalnajs, W. B. Westphal, Journal of Physics and Chemistry of Solids, 23, 779, (1962).
119
Chapter 5: Fabrication of patterned micrometer scale
interpenetrating Au–TiO2 nanocomposites
5.1 Introduction
Nanocomposites are, by definition, materials in which at least one of the
phases has constituents less than 100 nm in size1. Metal-ceramic
nanocomposites thin films presently attract tremendous attention due to their
fascinating properties and potential applications in electronics,
optoelectronics, magnetics and catalysis2. For MOS-type storage
applications, metal-ceramic nanocomposites consisting of either Ag, Au, Pt
or Ni nanoclusters in ultra-thin SiO2 or HfO2 layers hold advantages over
those employing semiconductor nanoclusters 3-5. In optoelectronics, metal-
ceramic nanocomposites with large third order optical non-linearities are
promising candidates for use in all-optical switches and optical
computation6-10. Also, nanocomposites of Co and FePt particles dispersed
in a non-magnetic ceramic phase such as SiO2, Al2O3, B2O3 and Si3N4 are
being intensively investigated for high density magnetic storage
applications11-13. In catalysis, nanocomposites of catalytically active metal
nanoparticles of either Ni or Pd, for example, in either TiO2, ZrO2 or MgO
phases are being studied for applications such as decomposition of
120
methanol to produce hydrogen and dry reforming of methane to produce
syngas14-16.
Various methods have been proposed for fabricating metal-ceramic
nanocomposites. These include evaporation10,14, sputtering9,11,12, ion
implantation7,8, deposition-precipitation15,17 and sol-gel processing6,13,16,18-
21. The sol-gel process has received particular attention as it renders
molecular-level control, allows the incorporation of various metal dopants
into ceramic matrices and is relatively low cost. Although conventional sol-
gel processing produces a continuous gel film, in applications where
patterned features are required, additional process steps such as reactive
ion etching, laser trimming, embossing or through-mask UV irradiation of
chelate modified gels and subsequent dissolution are used22-28.
Alternatively, self-assembled monolayers have been used as masks to
direct formation of gel layers from precursor solutions29-31. However, all the
aforementioned methods produce dispersed, non-percolating metal
nanoclusters embedded in a ceramic phase.
121
5.2 Interpenetrating network nanocomposites
An interpenetrating network composite is one in which each constituent
phase is continuous and interpenetrating throughout the microstructure32.
Previous reports of interpenetrating network nanocomposites are largely
limited to organic-organic33 and organic-inorganic material systems34. Prior
work in metal-ceramic material systems involve micrometer-scale phases
and hence, by definition, may not be classified as nanocomposites35. Au-
TiO2 interpenetrating network nanocomposites are expected to have the
functionalities of its constituents such as the wear resistance of TiO2
together with the high electrical conductivity of Au32. Thin films of Au and its
alloy have been widely used as contact material in microswitch devices
because of their low-resistivity and oxidation resistance36. However, they
suffer from wear and stiction which shorten device lifetimes37. In contrast,
titania films formed using sol-gel processes have excellent wear resistant
properties19. Presently, there is no prior work on the fabrication of patterned
micrometer scale metal-TiO2 interpenetrating network nanocomposites. In
this chapter, we demonstrate a technique to form patterned features of an
interpenetrating network nanocomposite of Au and nanostructured TiO2
(NST). NST features were first fabricated by reacting pre-patterned Ti with
aqueous hydrogen peroxide, as described previously. Using electroless
122
deposition, Au was then infiltrated into the NST features. Although Au was
used in this study, the process can be generalized to other metals that can
be deposited using electroless deposition.
5.3 Fabrication of patterned interpenetrating Au–TiO2
network nanocomposites
Fabrication of interpenetrating Au-NST network nanocomposites involves
two main process steps. Firstly, porous NST patterns are formed and,
secondly, NST patterns are infiltrated with Au using electroless deposition. A
schematic of the process is shown in Figure 5.1. NST pad arrays with
porous sponge-like morphology were formed by reacting pre-patterned
arrays of exposed Ti surfaces with aqueous hydrogen peroxide (aq. H2O2)
solution followed by low temperature annealing. The process steps for
fabricating the NST has been discussed in detail in Chapter 2, here only a
brief description is given. To prepare Ti thin film samples, 2.5 cm square
pieces of either N-type Si(100) or glass pieces were used as substrates. Si
pieces were first thermally oxidized at 1100 °C to grow 1 μm thick SiO2
layer. For glass substrates, a 1 μm thick SiO2 layer was deposited by
plasma-enhanced chemical vapor deposition (PECVD). Substrates first
cleaned for 5 min each in acetone, 2-propanol and de-ionized (DI) water
123
(18.9 MΩ) and blown dry with nitrogen before Ti was deposited. Ti film was
deposited on substrates using electron beam evaporation. Photo resist (PR)
was then spun on the silicon dioxide layer and patterned. The pattern on the
PR layer was transferred to the silicon dioxide layer by etching with CHF3
gas. The exposed patterns of Ti surfaces were oxidized by aging in 10% (by
volume) aq. H2O2 at 80 °C. This is followed by annealing at 300 °C for a few
hours. Aging and annealing steps were done in ambient air.
Au was infiltrated into pores of the NST using selective electroless
deposition. Commercially available electroless gold plating solution, based
on alkaline gold cyanide complex with disodium ethylenediaminetetraacetate
(Na2EDTA) as chelating agent, was purchased in a ready to use condition.
Chelating agents are species that form multiple bonds to, and hence
sequesters, a metal ion. In this study, Na2EDTA is used to trap metal ions
that would otherwise initiate secondary reactions that reduce the lifetime of
the Au plating bath. Substrates were coated with Au by immersion in a
heated bath of the plating solution. Prior to immersion, temperature of the
solution was maintained and stabilized at 60 ± 2 °C for about 20 min.
Temperature of the sample was maintained at the set temperature using
automated electronic feedback control. Substrates were immersed using
Teflon holders to eliminate contamination of the electroless plating bath. For
each plating run, fresh plating solution was used to ensure similar Au
124
deposition rate was obtained for all runs. After Au deposition, samples were
soaked and rinsed thoroughly in deionized water and subsequently blown
dry with nitrogen.
Figure 5.1. Schematic process flow for forming patterned Au-NST nanocomposite. (T-SiO2 and P-SiO2 denote thermally grown SiO2 and
SiO2 deposited by PECVD, respectively.)
125
Various techniques were used to characterize the NST and Au-NST
nanocomposites fabricated. X-ray diffraction using CuKα radiation (1.5406Ǻ)
in Bragg-Brentano configuration was used to determine crystal structure.
Structural characterization was done using an FEI dual beam focus ion
beam system equipped with Ga ion and electron columns for
micromachining and imaging, respectively. To micromachine samples, a Ga
ion current of 100 pA was used. Surface chemical species was determined
using a Kratos Axis Ultra X-ray Photoelectron Spectroscopy (XPS) system.
High resolution and survey scans of various surfaces were obtained. XPS
scans were obtained using monochromated Al Kα source (1486.6 eV) and
20 eV pass energy with steps of 0.05 eV at a base pressure of 7.5 x 10-9
Torr. XPS spectra collected were fitted to line shapes constructed from a
linear combination of Gaussian and Lorentzian profiles using commercial
software (CasaXPS). Atomic force microscopy (AFM) was used to
investigate the surface morphology of parent Ti, NST and Au-NST
nanocomposite. A Digital Instruments (Veeco) Dimension D3000
microscope operating in tapping mode in air was used in the AFM studies.
126
5.4 Results and discussion
As discussed previously, reaction of metallic Ti with hydrogen peroxide
had been investigated previously and was shown to result in formation of a
hydrated titania gel layer38,39. Because of the high crack-density in TiO2
layers formed using this technique, they are generally not suitable for device
applications40,41. We have recently developed a technique that allows
integration of NST into microsystems and eliminates crack formation. This
method involves pre-patterning Ti surfaces prior to reaction with hydrogen
peroxide41. After aging in hydrogen peroxide and subsequent annealing,
NST patterns have a nanoporous sponge-like morphology with
interconnecting pore walls; see Figures 5.2 (a) to (c). The pore walls have
thickness ranging from about 75 – 125 nm while pore diameters range from
50 nm – 200 nm. Unannealed NST patterns are hydrated titania gel layer
consisting of peroxo compounds38. In some regions, walls of pores are long
and narrow enough to be described as wires. The porous structure of NST
could be due to morphology of the intermediate gel layer formed during
aging in aq. H2O2. Similar porous microstructures were observed when
unpatterned bulk Ti foil was reacted with aq. H2O2 41. Figure 5.2 (c) is a
cross-sectional SEM image of NST obtained after focus ion beam milling. To
reduce material redepositing on the cross-sectioned surface, ion beam
127
milling was done at 100 pA. The cross-sectional image reveals that the 350
nm thick Ti layer had been fully oxidized and the porous structure of the
NST layer formed extends to the underlying SiO2 surface. NST patterns
formed have excellent adhesion to the underlying SiO2.
Representative SEM images of Au-NST nanocomposites formed after
electroless deposition of Au on NST patterns are shown in Figures 5.3 (a) to
(c). Electroless plating is the result of a chemical reaction in which a metal
layer is formed by reduction of a metallic ion by a reducing agent in an
aqueous solution. Protrusions 25 to 50 nm high decorated the surface of the
Au-NST nanocomposites. These protrusions were formed when Au plates
on titania wall protruding from the NST layer. After infiltration, pores of the
NST patterns were completely filled with Au. Figure 5.3 (c) shows that Au
has deposited in the pores throughout the NST feature and not just on the
surface. In addition, no significant void formation was detected in the Au-
NST nanocomposites. This suggests that during the electroless plating
process, there is no significant difference in the rate of Au deposition in the
NST layer than that on the top surface of the NST which could be due to the
relatively large pore size of the NST after annealing.
128
Figure 5.2. SEM images of NST: (a-c) before Au infiltration, (a) top view of a 20 μm pad, (b) higher magnification image (c) cross-sectional images
after micromachining (tilt 30 degree).
129
Figure 5.3. SEM images of Au-NST interpenetrating network nanocomposite: (a) top view of a 20 μm pad, (b) higher magnification image
(c) cross-sectional images after micromachining (tilt 52 degree).
130
The phase evolution of NST after has been discussed in Chapter 241.
Here we provide only a brief description. We found that after aging in aq.
H2O2, NST consists, largely, of amorphous TiO2 and nanocrystals of
anatase and that the amorphous TiO2 phase transforms to anatase upon
annealing as discussed in previously. XRD studies of patterned NST pad
arrays indicate that Au was deposited on NST pads only and not on the SiO2
mask. Such selective deposition was observed on amorphous and
crystalline NST; corresponding to the as-aged and annealed conditions,
respectively. Figure 5.4 shows XRD spectra of amorphous NST after plating,
demonstrates the selective deposition of Au after various immersion times in
the electroless plating bath. A comparison of XRD spectra of the sample
after Au plating indicates the presence of three peaks at 2θ values of 44.18,
64.46 and 81.76 degree on the NST pad array. These peaks, however, are
not present in spectra collected from the SiO2 mask region. These
observations strongly indicate that Au has deposited only in pores of the
NST pads and not on the surface of SiO2 mask. The three peaks observed
can be assigned to the 200, 220 and 222 of Au. A prior study of
electroless Au deposition on Ti films, which has an amorphous native oxide,
has reported the presence of these three Au reflections42. It is noted that the
patterned NST sample has also been partially oxidized leaving a layer of
residual Ti below the NST layer. Ti films on the patterned and blanket
samples used for XRD studies were deposited in different evaporation
131
machines, which resulted in different textures as reflected in the as-
deposited XRD spectra. Prior studies indicate that Ti surface texture has
little effect on the morphology of NST formed41.
Figure 5.4. XRD of samples illustrating selective Au deposition on unannealed NST(i) As deposited Ti; (ii) Plate 40 min, SiO2 mask; (iii) Plate 30 min, NST array; (iv) Plate 50 min, NST array. (Peaks for Ti and Au are
denoted by and ∆, respectively).
132
Results of XPS studies are shown in Figures 5.5 and 5.6. All spectra
were referenced to C 1s peak at 285.0 eV. Figure 5.5 (a) is a high resolution
scan of the Ti 2p peaks of unpatterned NST film formed on PECVD SiO2 on
Si substrate after annealing at 8 hr for 300 °C. Similar results were obtained
from NST formed on a PECVD SiO2 coated glass substrate. Assuming a
Tougaard, background, fitting to the raw spectra was done using Gaussian-
Lorentzian components using a commercial XPS analysis software
(CasaXPS). From the analysis, binding energies of Ti 2p3/2 and Ti 2p1/2 were
found to be 458.9 and 464.8 eV, respectively. These values are close to
those reported in the literature for TiO2 of 458.9 and 464.6 eV for Ti 2p3/2
and Ti 2p1/2, respectively, and confirm the formation of TiO2 after
annealing43,44.
Figure 5.5 (b) shows XPS spectra for binding energies from 55 to 110 eV
of unpatterned NST film before and after 5 min Au plating. Before plating
only Si 2p signal with a peak at 101.3 eV was detected. After plating,
however, three additional peaks were detected. Assignment of these peaks
was similarly done using CasaXPS. Two peaks at binding energies of 84.60
and 88.16 eV are assigned to Au 4f7/2 and Au 4f5/2. These experimental
values of Au 4f7/2 and Au 4f3/2 obtained are consistently higher, by 0.38 and
0.22 eV respectively, but close to the corresponding values reported in the
literature; which range from 83.70 to 84.25 eV for Au 4f7/2 and 87.71 to 87.94
133
eV for Au 4f5/2 of elemental Au 45-47. Hence, it is concluded that
electrodeposited Au is in the elemental form, in agreement with XRD results.
The peak at 63.13 eV in Figure 5 (b) is assigned to Na 2s; another peak,
which is assigned to Na 1s, at 1071.58 eV was also observed after plating.
A prior study has reported entrapment of sodium in a metal layer deposited
from a plating bath that similarly used Na2EDTA as a chelating agent48. It is
known that Na when dissolved in Au, may exist as Au(Na) alloy and/or
AunNa (where n = 5, 2, 1) intermetallic compounds49. Solomun had
determined Na 1s binding energies of adsorbed sodium atoms, in the
presence of coadsorbed iodine atoms, on Au(100) surfaces. In the absence
of iodine and when coverage of Na atoms on the Au surface is 0.09 of a
monolayer, the Na 1s binding energy was determined to be about 1071.05
eV, which is close to that obtained in the present study50 and suggests that
a Au(Na) alloy has been formed. Unfortunately there are no reports in the
literature of Au 4f7/2 and Au 4f5/2 binding energies from AunNa intermetallic
compounds for comparison. However, the presence of AunNa compounds
after plating, if any, is not extensive because no XRD reflections detected
can be assigned to these compounds. From the preceding discussion it may
be concluded that Au has been deposited into the pores of NST pads and
Na, at a dilute concentration, was incorporated into the Au layer during
plating.
134
Figure 5.5. XPS results of: (a) Ti2p high resolution line scans of NST before Au plating; (b) line scans before and after Au plating for binding
energies from 55 to 110 eV.
135
Results of area-mode XPS studies confirm that Au deposits selectively
on annealed NST. Figures 5.6 (a) and (b) show, respectively, Si 2p and Au
4f signals of one region of an NST pad array after electroless Au plating.
The results of the area-mode XPS are presented in gray-scale in which
areas with higher concentration of a chemical species appear brighter. In
addition, it is noted that XPS is a technique sensitive to chemical species on
the first few nanometers from the surface only51. From Figure 5.6 (a), the
strong Si 2p signal implies that very little, if any, Au has deposited on the
SiO2 mask oxide surface. This is confirmed in Figure 5.6 (b) where Au 4f
signal is observed only in areas that correspond to those of NST pads which
strongly indicate that Au has been deposited only on the NST pads and not
on the SiO2 mask. Hence, XPS results are in agreement with those of XRD
studies and indicate that Au has been deposited selectively only on NST
features.
136
Figure 5.6. XPS area-mode scans, after Au plating, of (a) Si 2p and (b) Au 4f signals.
137
Surface morphologies of the parent Ti thin film, NST and Au-NST
nanocomposite formed were studied using atomic force microscopy (AFM).
Figures 5.7 (a) to (c) show typical images of these surfaces. The surface of
the parent Ti film is rough, consisting of distinctly faceted Ti nanocrystals
separated by gaps. Ti crystals are platelet-like with thickness and longest
width of 40 ± 10 nm and 124 ± 31nm, respectively. (All stated dimensions
are an average and one standard deviation of 30 measurements.) In some
regions of the Ti surface, gaps are large enough to be described as troughs
with average diameter of about 101 ± 37 nm. AFM images of surfaces of
NST formed after oxidation and subsequent annealing steps show a
distinctly different morphology from that of the as deposited Ti. The NST
surface is rough with ridge-like protrusions that surround holes. The ridge-
like protrusions correspond to titania walls and consists of spherical
protrusions about 93 ± 32 nm in diameter. Average diameter of holes was
measured to be about 126 ± 33 nm. After Au plating however, the surface of
the Au-NST nanocomposite formed consists of globular grains with average
diameter of about 52 ± 12 nm. This results in a smoother surface, as Au was
deposited in the troughs and on the walls of the NST layer
138
Figure 5.7. AFM images of (a) as-deposited Ti film; (b) as-formed NST and (c) Au-NST nanocomposite formed after Au plating. (Images are 1 x 1
μm squares)
139
5.5 Integration of Au–TiO2 nanocomposites as contacts in
devices
In this section we demonstrate the fabrication of electrical contacts made
from interpenetrating Au-NST network nanocomposites. The device, in
which the contacts were integrated, is a microswitch that operates by the
closure and opening of a gap between two electrodes which are Au-NST
contacts. A schematic illustrating the operation of a switch is shown in
Figure 5.8. The gap between the electrodes is closed when a cantilever
deflects down and makes contact with the two electrodes. In this state the
circuit consisting of the electrodes is closed. When the cantilever is not
deflected, the gap between the electrode is open. Hence by deflecting the
cantilever up and down using a drive signal, the circuit consisting of the
electrodes can be switched on and off. At present the contacts in
microswitch devices are made from Au or its alloys. However, these
materials present reliability issues of wear and stiction36,37. In this section,
we propose and demonstrate a process that integrates Au-NST network
nanocomposites as contact materials.
The process steps for fabricating the electrical contacts are similar to
those describe in the Figure 5.1. Here only the salient steps are described in
detail. Because of shrinkage, thickness of NST features after annealing is
less than that of the masking oxide PECVD. To make the contacts a Ti layer
140
is first sputtered and subsequently patterned by etching in fluorine
chemistry. Of course, the Ti pattern can be formed using the lift-off
technique just as well. Au patterns were then evaporated on the Ti layer
using a lift-off technique to protect area of where NST is not desired. The
device is then soaked in hydrogen peroxide solution to form NST and
subsequently annealed at 300 °C. As mentioned earlier, incorporation of
impurities such Na may increase the resistance of the Au-NST composite.
To alleviate this problem, a layer of gold is evaporated on the Au-NST
composite. Figure 5.9 shows the integrated Au-NST composite contacts.
Figure 5.8. Schematic of a micro-switch device structure and operation: (a) Top view of device; Device in (b) on and (c) off states. (Micro-switch
fabrication work, except Au-NST integration, was carried out by Chang Song Ding)
141
Figure 5.9. Integration of Au-NST nanocomposite contacts in devices: NST pad (a) before Au-infiltration and (c) after Au-infiltration followed by
flash Au evaporation. (Magnification: 200X)
142
5.6 Conclusions
In this chapter we have demonstrated a facile technique to form
integrated and patterned micrometer-scale features of interpenetrating Au-
nanostructured TiO2 (NST) network nanocomposites. First, NST pads were
formed by aging Ti surfaces in aqueous hydrogen peroxide (aq. H2O2)
solution. As aged NST is largely amorphous but transforms to single phase
anatase upon annealing at 300 °C. Second, pores of the NST are then filled
with Au using electroless deposition. Cross-sectional SEM images show that
complete infiltration of NST pores was achieved with little void formation. X-
ray diffraction and x-ray photoelectron spectroscopy studies indicate that Au
was deposited selectively on NST pads and not on surfaces of the SiO2
mask. It must be noted that although Au was used in this study, it is
postulated other metals that can be deposited using electroless deposition
would work just as well. Hence, this technique represents a general
technique for integrating interpenetrating metal-NST network
nanocomposites features in microsystems.
143
5.7 References
1. R. Roy, R. A. Roy and D. M. Roy, Materials Letters, 4, 323 (1986).
2. S. Seal and M-. I. Baraton, MRS Bulletin, 29, 9 (2004).
3. Z. T. Liu, C. Lee, V. Narayanan, G. Pei and E. C. Kan, IEEE Transactions on Electron Devices, 49, 1606 (2002).
4. Z. T. Liu, C. Lee, V. Narayanan, G. Pei and E. C. Kan, IEEE Transactions on Electron Devices, 49, 1614 (2002).
5. Tan Zerlinda, S. K. Samanta, K. Y. Won and S. Lee, Applied Physics Letters, 86, 013107 (2004).
6. D. Ricard, P. Roussignol and C. Flytzanis, Optics Letters, 10, 511(1985).
7. K. Fukumi, A. Chayahara, K. Kadono, T. Sakaguchi, Y. Horino, M. Miya, J. Hayakawa and M. Satou, Japanese Journal of Applied Physics – Part 2 Letters, 30, L742, (1991)
8. T. Isobe, S. Y. Park, R. A. Weeks, R. A. Zuhr, Journal of Non-Crystalline Solids, 189, 173, (1995).
9. H. B. Liao, R. F. Xiao, H. Wang, K. S. Wong and G. K. L. Wong, Applied Physics Letters, 72, 1817 (1998).
10. Q. F. Zhang, W. M. Liu, Z. Q. Xue, J. L. Wu, S. F. Wang, D. L. Wang and Q. H. Gong, Applied Physics Letters, 82, 958 (2003).
11. C. C. Chen, M. Hashimoto, J. Shi, Y. Nakamura, O. Nittono and P. B. Barna Journal of Applied Physics, 93, 6273 (2003).
12. C. P. Luo and D. J. Sellmyer, Applied Physics Letters, 75, 3162 (1999).
13. M. F. Casula, A. Corrias, A. Falqui, V. Serin, D. Gatteschi, C. Sangregorio, C. D. Fernandez and G. Battaglin, Chemistry of Materials, 15, 2201 (2003).
14. M. Bowker, P. Stone, R. Bennett, and N. Perkins, Surface Science, 497, 155 (2002).
15. Y. Matsumura, M. Okumura, Y. Usami, K. Kagawa, H. Yamashita, M. Anpo and M. Haruta, Catalysis Letters, 44, 189 (1997).
144
16. R. Takahashi, S. Sato, T. Sodesawa, N. Nakamura, S. Tomiyama, T. Kosugi and S. Yoshida, Journal of Nanoscience and Nanotechnology, 1, 169 (2001).
17. P. Burattin, M. Che and C. Louis, Journal of Physical Chemistry B, 104, 10482, (2000).
18. M. Epifani, C. Giannini, L. Tapfer and L. Vasanelli, Journal of The American Ceramic Society, 83, 2385 (2000).
19. W. G. Zhang, W. M. Liu, B. Li, G. X. Mai, Journal of the American Ceramic Society, 85, 1770 (2002).
20. J. Livage, Current Opinion in Solid State and Materials Science, 2, 132 (1997).
21. L .L. Hench, and J. K. West, Chemical Reviews, 90, 33 (1990).
22. A. S. Holmes, R. R. A. Syms, M. Li and M. Green, Applied Optics, 32, 4916 (1993).
23. E. Zakar, M. Dubey, B. Piekarski, J. Conrad, R. Piekarz and R. Widuta, Journal of Vacuum Science and Technology A, 19, 345 (2001).
24. T. Brylewski and K. Przybylski, Applied Superconductivity, 1, 737(1993).
25. H. Krug, N. Merl and H. Schmidt, Journal of Non-Crystalline Solids, 147, 447 (1992).
26. R. L. Roncone, L. A. Wellerbrophy, L. Weisenbach and B. J. J. Zelinski, Journal of Non-Crystalline Solids, 128, 111 (1991).
27. K. Shinmou, N. Tohge and T. Minami, Japanese Journal of Applied Physics – Part 2 Letters, 33, L1181 (1994).
28. H. Tada, A. Hattori, Y. Tokihisa, K. Imai, N. Tohge and S. Ito, Journal of Physical Chemistry B, 104, 4585 (2000).
29. E. Kim, G. M. Whitesides, L. K. Lee, S. P. Smith and M. Prentiss, Advanced Materials, 8, 139 (1996).
30. N. L. Jeon, P. G. Clem, R. G. Nuzzo and D. A. Payne, Journal of Materials Research, 10, 2996 (1995).
145
31. R. J. Collins, H. Shin, W. R. DeGuire, A. H. Heuer and C. N. Sukenik, Applied Physics Letters, 69, 860 (1996).
32. D. R. Clarke, Journal of the American Ceramic Society, 75, 739 (1992).
33. C. Decker, Macromolecular Rapid Communications, 23, 1067 (2002).
34. K. G. Sharp, Advanced Materials, 10, 1243 (1998).
35. M. C. Breslin, J. Ringnalda, L. Xu, M. Fuller, J. Seeger, G. S. Daehn, T. Otani and T. L. Fraser, Materials Science and Engineering A, 195, 113 (1995).
36. D. Hyman and M. Mehregany, IEEE Transactions on Components and Packaging Technology, 22, 357 (1999).
37. J. R. Coutu, P. E. Kladitis, K. D. Leedy and R. L. Crane, Journal of Micromechanics and Microengineering, 14, 1157 (2004).
38. P. Tengvall, I. Lundstrom, L. Sjoqvist, H. Elwing and L. M. Bjurstein, Biomaterials, 10, 166 (1989).
39. J. P. Bearinger, C. A. Orme and J. L. Gilbert, Surface Science, 491, 370 (2001).
40. J. M. Wu, S. Hayakawa, K. Tsuru and A. Osaka, Scripta Materialia, 46, 101 (2002).
41. A. S. Zuruzi and N. C. MacDonald, Advanced Functional Materials, 15, 396 (2005).
42. Z. Z. Hou, N. L. Abbott and P. Stroeve, Langmuir, 14, 3287 (1998).
43. T. Choudhury, S. O. Saied, J. L. Sullivan and A. M. Abbott, Journal of Physics D – Applied Physics, 11, 1185 (1989).
44. D. Gonbeau, C. Guimon, G. Pfisterguillouzo, A. Levasseur, G. Meunier and R. Dormoy, Surface Science, 254, 81 (1991).
45. J. C. Fuggle, E. Kallne, L. M. Watson and D. J. Fabian, Physical Review B, 16, 750 (1977).
46. N. H. Turner and A. M. Single, Surface and Interface Analysis, 15, 215 (1990).
146
47. T. D. Thomas and P. Weightman Physical Review B, 33, 5406 (1986).
48. H-. I. Chen, C-. K. Hsiung and Y-. I. Chou, Semiconductor Science and Technology, 18, 620 (2003).
49. Massalski T B, Murray J L, Bennett L H and Baker H (Editors), Binary Alloy Phase Diagrams: Volume 1 1986 (Metals Park, Ohio: American Society for Metals)
50. T. Solomun, Surface Science, 331-333, 52 (1995).
51. J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of X-Ray Photoelectron Spectroscopy: A Reference Book of Standard Spectra for Identification and Interpretation of XPS Data, Minnesota, Physical Electronics Inc. (1995).
147
Chapter 6: Attachment of mouse fibroblasts on
nanostructured TiO2
6.1 Introduction
At present there are tremendous efforts towards using microelectronics
devices and microsystem (MEMS) devices for biological applications1-3. The
cellular milieu/electronic material interface may need to be engineered and
optimized to suit these applications. Various approaches have been
investigated for engineering the cellular milieu/electronic material interface.
A few research groups have demonstrated that grafting of organic moieties
such as peptides, proteins and sugars on surfaces allows modulation of cell
attachment on surfaces4-9. Use of plasma polymerized tetraglyme coating
(similar to polyethylene glycol or PEG) on the other hand results in
biocompatible surfaces with non-fouling properties10. Areas on a device
where cells are not desired can be arbitrarily defined by patterning and
removing the tetraglyme layer. In addition to modifying the chemical
moieties on the surface, other techniques include engineering the surface
topography. In this way it was found that transformed as well as primary glial
cells preferentially attached to silicon pillar arrays rather than smooth silicon
surfaces11. Similarly, etching of Si results in a biocompatible porous Si that
148
sustains growth of calcium phosphates and viability of primary rat
hepatocytes12,13.
The excellent in-vivo biocompatibility of titania is well known and use of
bulk Ti implants for dental and bone prosthetic applications is well
documented14,15. Recent work indicated that nanophase titania, with grain
sizes from ~20-40 nm enhanced osteoblast cell functions in vitro16. Hence,
NST seems to be an attractive material for optimizing the cellular
milieu/electronic material interface. The objective of this chapter is to
investigate the attachment of mouse fibroblasts cells on blanket
(unpatterned) NST films as well as patterned NST features. As a
comparison, two other types of substrates were studied – Si substrates
coated with either plasma-deposited SiO2 (PECVD SiO2) or sputter-
deposited titania. At present, PECVD SiO2 is the most common passivation
layer for BioMEMS devices.
6.2 Attachment of cells on surfaces
Attachment of cells on various materials was investigated by performing
cell seeding experiments. The area of a sample covered with cells, after
duration of cell seeding was then determined using image analysis.
Morphology of cells was studied using optical microscopy, laser scanning
confocal microscopy and scanning electron microscopy.
149
Materials preparation
PECVD SiO2 and smooth titanium oxide films were coated on separate
electronic grade N-type silicon wafers (Mitsubishi Electronic Materials).
Plasma enhanced chemical vapor deposition were used to deposit 200 nm
thick PECVD SiO2. A coating of ~150 nm thick titanium oxide films was
deposited by reactive sputtering Ti in an Ar/O2 plasma. These wafers were
then cut into 1 cm square tabs using an automated saw. Tabs were then
soaked, with ultrasonic agitation, sequentially in acetone, isopropyl alcohol
and deionised water (18 MΩcm) for 5 min each. The tabs were then blown
dry with nitrogen and heated at 120 °C on a hotplate for 10 min to remove
moisture. Porous NST used were formed from 500 nm Ti films that was
aged for about 2.5 hr in hydrogen peroxide solution and subsequently
annealed at 800 °C for 8 hr. Prior to annealing the hydrated titania gel layer
was soaked in DI water for about 30 min.
Cell seeding
Mouse fibroblasts cells from a cell line were maintained in Dulbecco’s
modified Eagle’s medium (DMEM) supplemented with 10 % (v/v) fetal
bovine serum (FBS) and 1 % penicillin/streptomycin antibiotics in 5 % CO2
ambient at 37 °C. Cell cultures were passaged every 3 days. Cell seeding
experiments were carried out in 12 well plates with one tab per well. Cells
were dislodged by incubating in 5 ml of 0.05 % trypsin / 0.5 mM EDTA
150
solution. After incubation, trypsin was inhibited by adding 5 ml of serum
containing media. The cell suspension was concentrated by centrifuging at
1500 revolution per minutes (rpm) for 5 min to form a pellet of cells. The
cells were then resuspended in warm DMEM. Concentration of cells in the
suspension was measured using a hemacytometer.
Cells were plated at a density of ~105 cells per well. Each well contains
one tab. The cells were then incubated at 37°C in 5 % CO2 / 95 % air
environment. After the desired incubation time, cells were rinsed twice with
warm 1 X phosphate buffered saline (PBS) and fixed with 4%
paraformaldehyde in PBS. Tabs were then observed under optical
microscopy and micrographs taken. The percentage of area covered by
cells in a micrograph was calculated using Scion Image - a free software
from the National Institutes of Health. Micrographs taken of bare Si tabs as
well as PECVD SiO2 and TiO2 coated tabs have good brightness and
contrast. However, micrographs of NST-coated tabs were dark due to
scattering. For the latter, micrographs taken using scanning electron
microscopy were used to determine percentage of area covered. To
calculate the percentage of area covered by cells, at least five readings
were used.
For laser scanning confocal microscopy studies, cells were stained using
Texas Red-dye obtained from a commercial vendor. Preparation of cells for
scanning electron microscopy studies involves dehydrating cells using a
151
critical point dryer. Ethanol was used as the intermediate solvent. Prior to
drying, cells were soaked in phosphate buffered saline-ethanol solutions at
increasing concentration of ethanol to prevent shrinkage of cells. To reduce
charging during SEM, tabs were coated with ~5 nm Ti layer. Samples were
observed in SEM mode of the FEI dual beam focus ion beam system.
6.3 Results and discussion
6.3.1 Seeding of fibroblast on various materials
Fibroblast cells exhibit enhanced attachment to NST surfaces after short
seeding times. Figure 6.1 shows the average area of various surfaces
covered by fibroblast cells after cell seeding. The area covered by cells for
NST surfaces is consistently greater than that of PECVD SiO2 for culture
times up to 18 hrs. After 24 hr culture time, the area coverage for all
surfaces is ~45 %. There is no significant difference between those for
sputtered and NST surfaces. To determine the level of significance in
differences between average values of area coverage, the t-probability value
in student’s t-test is used. It is generally accepted that for a comparison
between two data sets a t-probability value less than 0.05 indicates a
significant difference in the means. Only t-probability values for comparisons
between PECVD SiO2 and NST are given in Figure 6.1. The t-probability
values for cell seeding times of 3 to 18 hours are less than 0.05. In contrast,
152
for 24 hr cell seeding time, t-probability value is 0.9912. These values
indicated that there is significant difference in means of corresponding data
sets of PECVD SiO2 and NST at seeding times up to 18 hr.
Figure 6.1. Trend of average area covered by cell of various surfaces with seeding time. Error bars indicate one standard deviation. t-
probability values are for comparisons between area coverage on PECVD SiO2 and NST.
In addition to differences in area coverage, cells attached on PECVD and
NST surface exhibit different morphology. Figure 6.2 (a) and (b) show
confocal scanning microscopy images of fibroblasts after 6 hrs seeding time.
Cells seeded on PECVD SiO2 have spherical globular morphology with little
153
contact to the surface. In contrast cells seeded on NST exhibit a flat
morphology with cells spread out on the surface. It is now widely accepted
that cells with little preference to adhesion on a substrate adopt spherical a
configuration17,18. The flat spread out morphology of fibroblasts on titania
surfaces then indicates enhanced adhesion relative to PECVD SiO2
surfaces. After 24 hr seeding time, cells seeded on PECVD SiO2 surfaces
also exhibit the flat spread out morphology.
Figure 6.2. Confocal microscopy images of fibroblast seeded for 6hr on (a) PECVD SiO2 and (b) NST. (Courtesy of Blaine Butler)
Further evidence for adhesion is the formation of numerous processes
shown in Figure 6.3 (a) to (c). These processes are about 75 nm in diameter
and can be up to ~7 µm long. It was expected that these processes would
154
penetrate these pores as the pore diameter range in size from about 50 to
250 nm. However, instead of going through pores, tips of processes were
observed to expand into lamella-like structures (~400 nm diameter) that
span the opening of large pores. Figure 6.3 (c) shows a single lamella-like
structure at the tip of a process. Recent studies have reported that
processes of human osteoblast-like cells are able to penetrate porous
alumina with 200 nm average pore diameter19.
155
Figu
re 6
.3. S
cann
ing
elec
tron
mic
rosc
opy
imag
es o
f a fi
brob
last
afte
r 6hr
on
NST
. Not
e th
e la
mel
la-li
ke
stru
ctur
e in
(c)
156
6.3.2 Morphology of fibroblasts on patterned NST
SEM micrographs of L-cells cultured for 3 days on arrays of patterned
NST pad are shown in Figure. 6.4. Generally, cells have flat and spread out
morphology similar to those on blanket layers. However, a surprising
observation was the influence on cell shape by arrays of patterned NST
pads. It was found that fibroblast cells may attach to and take the shape of
NST pads, Figures 6.4 (b) and (c) are SEM micrographs of a cell that has
taken the shape of a 20 µm pad. It may be observed that the cell is flattened
with lamellapodia wrapping the periphery of the pad – inset is a tilted view.
Figure 6.4 (d) shows a cell that has completely taken the shape of a square
pad. Such a strong interaction between cells and NST pads was not
expected.
Shape of single cells on synthetic surfaces had been shown to be
modulated by organic cues on surfaces5,20,21. Attachment of cultured cells to
synthetic surfaces is mediated by adsorption of proteins from serum in the
culture medium5,22-24. Hence by directing adsorption of proteins to spatially
defined regions using organic cues that had been deposited on surfaces
prior to cell seeding, it had been demonstrated that cell shape could be
controlled5,20,21. In the present study, it is speculated that because of the
large surface area of the NST, protein adsorption on pads is significantly
greater than on smooth silicon dioxide mask. In addition, it is likely that
157
proteins adsorbed on surface of NST do not lose any their activity.
Increased protein adsorption then enhances cell attachment on titania pad
which results in the observed cell patterning. However, because protein
adsorption is not specific to the NST pads, cells also attached and spread
on the silicon dioxide mask. Another factor that may influence the
attachment of cells on surfaces is surface morphology. It is possible that the
topography of pads enhance the attachment of cells.
Figure 6.4. Scanning electron microscopy images of fibroblasts seeded on patterned NST.
158
6.4 Conclusions
In conclusion, we have investigated the attachment of mouse fibroblast
cells on PECVD SiO2, NST and sputtered titania. It was found up to about
18 hrs, significantly more attachment of fibroblast cells occur on NST than
on PECVD SiO2 as indicated by area coverage of cells on these surfaces.
However after 24 hr, the area coverage of cells on all surfaces is similar.
Overall, the area coverage of cell on NST and sputtered titania surfaces is
similar at all times. This enhanced initial attachment of cells is suggested to
be due to enhanced adsorption of protein on titania surfaces. It is speculated
that adsorbed proteins on titania surfaces do not lose their activity.
Morphology of cells on titania surfaces are flat and spread out. When
surfaces consisting of patterned NST pad arrays with PECVD SiO2 mask,
some fibroblast cells took the shape of pads.
159
6.5 References
1. J. T. Santini, M. J. Cima, R. L. Langer, Nature, 397, 335 (1999).
2. D. V. McAllister, P. M. Wang, S. P. Davis, J. H. Park, P. J. Canatella, M. G. Allen, M. R. Prausnitz, Proceedings of the National Academy of Science, 100, 13755 (2003).
3. L. Leoni, T. A. Desai, Advanced Drug Delivery Reviews, 56, 211 (2004).
4. J. H. Lee, J. W. Jung, I. K. Kang, H. B. Lee, Biomaterials, 15, 705 (1994).
5. C. S. Chen, M. Mrksich, S. Huang, G. M. Whitesides, D. E. Ingber, Science, 276, 1425 (1997).
6. M. Zhang, T. A. Desai, M. Ferrari, Biomaterials, 19, 953 (1998).
7. C. S. Giannoulis, T. A. Desai, Journal of Materials Science: Materials in Medicine, 13, 75 (2002)
8. H. Hu, Y. Ni, V. Montana, R. C. Haddon, V. Parpura, Nano Letters, 4, 508 (2004).
9. P. H. Weigel, R. L. Schnaar, M. S. Kuhlenschmidt, E. Schmell, R. T. Lee, Y. C. Lee, S. Roseman, Journal of Biological Chemistry, 354, 10830 (1979).
10. V. H. Pan, Y. Hanein, D. Leach-Scampavia, K. F. Böhringer, B. D. Ratner, D. D. Denton. Proceedings of the IEEE Conference on Micro Electro Mechanical Systems (MEMS), Interlaken, Switzerland, January 21-25, 435 (2001).
11. H. G. Craighead, C. D. James, A. M. P. Turner, Current Opinion in Solid State and Materials Science, 5, 177 (2001).
12. L. T. Canham, C. L. Reeves, D. O. King, P. J. Branfield, J. G. Crabb, M. C. L. Ward, Advanced Materials, 8, 850 (1996).
13. V. Chin, B. E. Collins, M. J. Sailor, S. N. Bhatia, Advanced Materials, 13, 1877 (2001).
160
14. T. Albrektsson, P. I. Brånemark, H. A. Hansson, B. Kasemo, K. Larsson, I. LundstrÖm, D. H. McQueen, R. Skalak, Annals of Biomedical Engineering, 11, 1 (1983).
15. D. M. Brunette, P. Tengvall, M. Textor, P. Thomsen, Titanium in medicine: materials science, surface science, engineering, biological responses and medical applications. Springer-Verlag, New York, (2001).
16. T. J. Webster, R. W. Siegel, R. Bizios R, Biomaterials, 20, 1221 (1999).
17. L. V. Domnina, O. Y. Ivanova, L. B. Margolis, L. V. Olshevskaja, Y. A. Rovensky, J. M. Vasiliev, I. M. Gelfand, Proceedings of The National Academy of Sciences – USA, 69 248 (1972).
18. J. Folkman, A. Moscona, Nature 273, 345 (1978).
19. M. Karlsson, , E. Pålsgård, L. Wilshaw, L. Di Silvio, Biomaterials, 24, 3039 (2003).
20. R. Singhvi, A. Kumar, G. P. Lopez, G. N. Stephanopoulos, D. I. C. Wang, G. M. Whitesides, D. E. Ingber, Science, 264, 696 (1994).
21. A. Kumar, A., N. L. Abbott, E. Kim, H. A. Biebuyck, G. M. Whitesides, Acc. Chem. Res., 28, 219 (1995).
22. M. D. Pierschbacher, E. Ruoslahti, Nature 309, 30 (1984).
23. P. Knox, J. Cell Sci. 71, 51 (1984).
24. Hayman, E. G., Pierschbacher M. D., Suzuki, S. & Ruoslahti E. Exp. Cell Res. 160, 245 (1985).
161
Chapter 7: Conclusions and future work
7.1 Conclusions
A novel process has been developed to integrate crack-free crystalline
nanostructured titania (NST) into microsystems. This process involves
reacting patterned Ti films with aqueous hydrogen peroxide (H2O2) solution
followed by annealing. NST formed is porous with walls of pores having
thicknesses and pore diameters ranging from 25 nm – 50 nm and 50 nm –
200 nm. Crack elimination is achieved by oxidizing Ti films, pre-patterned
below a threshold dimension, in aqueous hydrogen peroxide solution. NST
formed is amorphous but transformed to anatase after annealing at 300 °C
for a few hours.
Oxidation kinetics of Ti films occurs by nucleation and growth
mechanism. It was found that grain size and thickness of films affects the
oxidation kinetics. Grain size was found to affect both nucleation and growth
stages. Films with finer grain size have shorter nucleation period and higher
growth rate. For thin films (25 and 50 nm), growth occurs at a constant rate
until oxidation in complete. For thicker films (100, 150 and 200 nm films),
growth rate decreases after a certain thickness of porous titania has been
formed. This change in rate is attributed to a change in the mechanism
controlling growth of the oxide layer. During the initial period of growth (sub-
162
stage I) growth of oxide is controlled by reaction of Ti species with hydrogen
peroxide molecules. During the later stage (sub-stage II), diffusion of
hydrogen peroxide molecules through the oxide layer is the rate controlling
mechanism.
Functionality of the NST formed and compatibility of the process
developed were demonstrated by exploring three applications. First,
prototype gas sensors were fabricated on both Si and plastic substrates.
These sensors were able to sense hydrogen and oxygen gases at parts per
million levels. Second, integrated micrometer-scale interpenetrating Au-NST
network nanocomposites were fabricated. Although Au has been used in
this work, other metals amenable to electroless deposition would be
expected to work as well. Patterned mictometer sale Au-NST
nanocomposite features have been integrated into MEMS micro-switches as
contact materials. Third NST features have been used as porous cell
adhesion layer in devices. Initial attachment of fibroblast cells on NST in
greater than commonly used PECVD-deposited SIO2.
7.2 Future work
Although the NST formed has been characterized using a variety of
techniques in this dissertation study, further characterization work is
163
required. One aspect that has not been studied thoroughly is porosity. In
future work it is suggested that gas adsorption studies be done to
characterize porosity. This could be done in conjunction with small angle x-
ray scattering for smaller pore sizes. Furthermore, other methods of
controlling pore size, besides Ti film thickness, need to be investigated. In
addition, the mechanical properties of NST have not been studied. The
mechanical properties could be investigated by growing patterned NST on Ti
thin foils and performing stress strain test.
In the area of applications, methods need to be developed to dope the
NST. Doping or incorporating metallic species for example could be
beneficial for gas sensing applications. Functionalizing of NST surfaces with
chemical moieties has already been done in conjunction with the research
group of Professor Moskovits and has shown positive initial results.
However, more work is required in this area to show the functionality of
these grafted molecules. In the area of wear resistant contacts, work is
required to generate data on wear properties of nanocomposite at various
device operating regimes. In addition, the contact resistance of these
nanocomposites needs to be investigated. Further work is also required
towards implementing NST in BioMEMS application. One area that needs
further study is adsorption of protein on NST. In particular, work is required
to characterize kinetics of proteins adsorption.
164