Interfacial Engineering of MicroStructured Materials
by
Aimee Poda
A dissertation submitted to the Graduate Faculty of
Auburn University
in partial fulfillment of the
requirements for the Degree of
Doctor of Philosophy
Auburn, Alabama
May 14, 2010
Keywords: Interfacial Engineering, MicroStructured Devices, Self-Assembled Monolayers,
Catalytic Microreactor
Copyright 2010 by Aimee Poda
Approved by
William R. Ashurst, Chair, Assistant Professor of Chemical Engineering
Christopher B.Roberts, Uthlaut Professor of Chemical Engineering
Ram B.Gupta, Alumni Professor of Chemical Engineering
Barton Charles Prorok, Associate Professor of Materials Engineering
Mario Richard Eden, Associate Professor of Chemical Engineering
Abstract
The tribological behavior of octadecyltrichlorosilane self assembled monolayers (OTS-
SAMs) has been successfully exploited to reduce energy losses and to produce adequate adhe-
sion barrier properties on many MEMS surfaces. Unfortunately, performance discrepancies
are reported in the literature between films produced on smooth surfaces as compared to typ-
ical MEMS surfaces maintaining topographical roughness. Rational explanations in terms
of reproducibility issues, production considerations, and the scale of measurement technique
have been introduced to account for some of the variation. The tribological phenomena
at the micro-scale are complicated by the fact that rather than inertial effects, the forces
associated with the surface become dominant factors influencing the mechanical behavior
of contacting components. In MEMS, real mechanical contacts typically consist of a few
nanometer scale asperities. Furthermore, various surface topographies exist for MEMS de-
vice fabrication and their corresponding asperity profiles can vary drastically based on the
production process.
This dissertation presents research focusing on the influence of topographical asperities
on OTS film properties of relevance for efficient tribological improvement. A fundamental
approach has been taken to carefully examine the factors that contribute to high quality
film formation, specifically formation temperature and the role of interfacial water layer
associated with the sample surface. As evidenced on smooth surfaces, the characteristics
for successful tribological performance of OTS films are strongly dependent on the lateral
packing density and molecular orientation of the monolayer. Limited information is available
on how monolayers associate on topographical asperities and whether these topographical
asperities influence the interfacial reactivity of MEMS surfaces. A silica film produced from
a low temperature, vapor-phase hydrolysis of tetrachlorosilane with a tunable topography is
ii
introduced and leveraged as a novel investigative platform for advanced analytical investi-
gations often restricted to use on smooth surfaces. This tunable surface allows intellectual
insight into the nature of surface properties associated with silica surfaces, the uptake of
interfacial water and the subsequent influence of surface morphology on OTS film forma-
tion. FTIR analysis was utilized for an examination of interfacial properties on both smooth
Si(100) surfaces and on the tunable MVD topography in combination with an investigation
of OTS film formation mechanism. A dilute etchant technique is developed to provide topo-
graphic contrast for AFM imaging to allow direct examination of film packing characteristics
in relation to surface asperities. A relationship between monolayer adsorption characteris-
tics and topographical asperities with observed variations in monolayer order resultant from
surface roughness has been elucidated. Results show that the packing structure of OTS
monolayers is dependent on the local asperity curvature which is qualitatively different from
that observed on flat surfaces.
In addition, a difference in surface reactivity is observed as a result of different surface
topographies with thicker silica layers maintaining a thicker interfacial water layer result-
ing in a higher coverage of OTS monolayers at similar reaction times and conditions. This
work shows changes in surface reactivity as a consequence of different morphological sur-
face characteristics and preparation procedures. Additional research is presented on a new
class of SAM, namely octadecylphoshonic acid and its monolayer formation mechanism and
properties are compared to conventional OTS monolayers. This monolayer is translated to
investigative probes based on Aluminum oxide specifically tailored for a tribological com-
parison across multi-scale friction regimes.
iii
Acknowledgments
First and foremost, I must thank my family for the patience, understanding, love, and
unending support during my pursuit of this degree. Without this support and constant
encouragement, this degree, and the work detailed in this dissertation, would not have been
possible.
I especially wish to thank my research advisor and friend, Dr. Bob Ashurst for his
guidance, understanding and support during my pursuit of this high honor. The insightful
discussions we have had along with the advice he has given and expectations he has set have
led me to discover my inner strength and ability to achieve the highest of standards, for that
I am extremely grateful.
iv
Table of Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi
I Interfacial Engineering of Microelectromechanical Systems . . . . . . . . . . . . . 1
1 Introduction to MEMS Research . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1 Background Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 Interfacial Phenomenon in MEMS Devices . . . . . . . . . . . . . . . 3
1.1.2 Overcoming Interfacial Interactions with Self-Assembled Coatings . . 6
1.1.3 Self-Assembled Octadecyltrichlorosilane Monolayers asaCandidate for
Overcoming Interfacial Interactions . . . . . . . . . . . . . . . . . . . 6
1.1.4 Disparity in Tribological Improvements . . . . . . . . . . . . . . . . . 13
1.1.5 Self-Assembled Octadecylphosphonic Acid Monolayers as a Candidate
for Overcoming Interfacial Interactions . . . . . . . . . . . . . . . . . 15
2 Investigative Summary for Part I . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1 Analytical Tools and Techniques . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.1 Contact Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.2 Ellipsometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.3 Atomic Force Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.4 Fourier transform infrared (FTIR) spectroscopy . . . . . . . . . . . . 23
3.1.5 TGA Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Experimental Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
v
3.2.1 Sample Preparation and Cleaning . . . . . . . . . . . . . . . . . . . . 28
4 Formation of Octadecyltrichlorosilane Monolayers on a Molecular Vapor De-
posited (MVD) Silica Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.1 Introduction to the MVD silica layer . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Utilization of MVD silica layer as a Novel Platform to Study OTS Monolayers 35
4.2.1 Investigation of Water Layers on the MVD Silica Surface. . . . . . . . 36
4.2.2 Influence of Formation Time and Temperature on OTS Monolayer For-
mation on a Highly Hydrated MVD Silica Surface . . . . . . . . . . . 43
4.2.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3 High Surface area MVD Silica Support for TGA Analysis of OTS Monolayers 52
4.3.1 Production of Fibrous MVD Silica Support . . . . . . . . . . . . . . . 53
4.3.2 Analysis of OTS Monolayers on Fumed Silica and Fibrous MVD Silica
Supports without Post Treatment . . . . . . . . . . . . . . . . . . . . 55
4.3.3 Analysis of OTS films on Fumed Silica and Fibrous MVD Silica after
Thermally Annealing with Different Gases . . . . . . . . . . . . . . . 60
4.3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5 OTS Island Formation on Topographical Asperities: Implications Toward Tribo-
logical Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.2 Introduction of an Etchant Procedure for Obtaining Topographical Contrast
with AFM Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.3 Reduced Temperature Deposition of OTS Films as Influenced by Surface To-
pography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6 Influence of MVD Surface Topography on Interfacial Water Layers- Implications
Toward OTS Monolayer Formation . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
vi
6.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7 Influence of Cleaning Procedures on Interfacial Water Layers on Silicon Oxide
Surfaces- Implications Toward OTS Monolayer Formation . . . . . . . . . . . . 95
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
8 Overall Conclusions / Contribution to OTS Monolayer Formation . . . . . . . . 107
9 Self-Assembled Octadecylphosphonate Monolayers on Metal Oxide Substrates . 112
9.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
9.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
9.2.1 Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
9.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
9.2.3 Conclusions to Phosphonate Work . . . . . . . . . . . . . . . . . . . . 125
II Interfacial Engineering of a Catalytic Fischer Tropsch MicroReactor . . . . . . . 128
10 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
10.1 Background Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
10.1.1 Introduction to MicroReactors . . . . . . . . . . . . . . . . . . . . . . 130
10.1.2 Catalyst Considerations . . . . . . . . . . . . . . . . . . . . . . . . . 132
10.1.3 Introduction to Fischer Tropsch Synthesis . . . . . . . . . . . . . . . 133
10.1.4 Catalysts for Fischer Tropsch synthesis . . . . . . . . . . . . . . . . . 136
10.2 Tunable Fischer Tropsch Microreactor Realization . . . . . . . . . . . . . . . 139
10.3 Development of catalyst material for incorporation into the micro reactor system141
10.3.1 Micro fibrous Catalyst Support System (MCSS) . . . . . . . . . . . . 141
10.3.2 Qualification of Micro fibrous Catalyst Support System (MCSS) . . . 142
10.3.3 Catalyst Support Manufacturing Procedure . . . . . . . . . . . . . . 147
10.3.4 Utilization of Catalyst Support in Standard FT Reactor . . . . . . . 148
vii
10.4 Systematic Probing of the Operating Parameters . . . . . . . . . . . . . . . . 151
10.4.1 Phase 1. Investigation of the Effect of Pressure Drop . . . . . . . . . 151
10.4.2 Residence Time of Distribution . . . . . . . . . . . . . . . . . . . . . 153
10.5 Microreactor Utilization for Fischer Tropsch Synthesis . . . . . . . . . . . . . 155
10.6 Results and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
III Interfacial Engineering of Microfluidic Devices . . . . . . . . . . . . . . . . . . . 171
11 Surface Engineering of Poly(dimethylsiloxane) Microfluidic Devices . . . . . . . 172
11.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
11.2 Background Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
11.3 Investigation of Water Vapor Diffusivity in PDMS Devices as Influenced by
Curing Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
11.4 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.4.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.4.2 Wafer preparation and cleaning . . . . . . . . . . . . . . . . . . . . . 176
11.4.3 Teflon Mold Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . 177
11.4.4 PDMS Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
11.4.5 Deuterated Water Loading and Plasma Treatment for Device Sealing 178
11.5 Analytical Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
11.5.1 Contact Angle Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 179
11.5.2 Fourier transform infrared (FTIR) spectroscopy . . . . . . . . . . . . 180
11.5.3 Thermogravimetric Analysis and Mass Spectroscopy . . . . . . . . . . 180
11.6 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
11.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
11.8 Investigation of MVD silica Layers for Water Vapor Barrier Improvement . . 190
11.8.1 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
viii
List of Figures
1.1 SEM images of surface morphology of MEMS fabricated by (a) Si micro machin-
ing, (b) SOI, and (c) electro forming processes adapted from reference [1]. . . . 5
1.2 Representation of the three parts of self-assembled monolayer illustrating the
head groups, tail groups, and carbon chain backbones. . . . . . . . . . . . . . . 7
1.3 Proposed reaction mechanism of OTS monolayer adsorption onto a hydrated
surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Submonolayer growth regimes for OTS film formation as influenced by tempera-
ture adapted from reference [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1 Illustration of the polarized light electrical vector as influence by p-polarization
at an angle of incidence greater than 80 ?. . . . . . . . . . . . . . . . . . . . . . 24
3.2 Influence of molecular tilt on the orientation of methylene stretches for phospho-
nate molecules on metal surfaces. Methylene stretches from an all trans (per-
pendicular orientation) are not identified with specular reflectance spectroscopy
(p-polarized IR) due to an electric field oriented normal to the substrate at an
angle of incident of > 80?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Schematic (left) and photograph (right) of the custom vacuum deposition system
and FTIR employed in this work. . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.1 Custom built teflon ATR assembly sealed from ambient conditions with inlet and
outlet purge assemblies for controlling relative humidity. . . . . . . . . . . . . . 37
4.2 Two dimensional AFM scans of (A) a clean oxidized Si(100) surface, RMS rough-
ness is 0.14 nm, and (B) Silica layer deposited on clean silicon surface, RMS
roughness is 1.3 nm, thickness is 18.0 nm. . . . . . . . . . . . . . . . . . . . . . 38
4.3 ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on an
oxidized silicon ATR crystal at increasing relative humidities. The peak positions
of ice-like 3119 cm?1 and the position of the O-H bending mode is marked at
1635 cm?1. The direction of increasing relative humidity is from bottom to
top. Spectra correspond to relative humidities of 20%, 40%, 60% and 80% with
an accuracy of ?2.0%. [Inset View]-Background spectra for water adsorption
study illustrating O-H stretching region associated with hydrogen bonded surface
silanols. Note peak absence at 1635 cm?1 indicative of physisorbed water. . . . 39
ix
4.4 ATR-FTIR spectra of adsorbed water on the surface of an oxidized silicon ATR
crystal at increasing relative humidities. Ice-like and liquid water are indicated
with arrows at 3250 cm?1 and 3400 cm?1, respectively, and the position of the
O-H bending mode is marked at 1635 cm?1. The direction of increasing relative
humidity is from bottom to top. Spectra correspond to relative humidities of
20%, 40%, 60% and 80% with an accuracy of ?2.0% . . . . . . . . . . . . . . . 40
4.5 AFM micrographs of OTS monolayer formation (A) A 20 second OTS submono-
layer deposition on native silicon oxide at 10 ?C. (B) A 20 second OTS submono-
layer deposition on native silicon oxide at room temperature. . . . . . . . . . . . 44
4.6 Ellipsometric film thickness vs immersion time for octadecyltrichlorosilane films
on oxidized Si(100) and MVD silica layer at 10 ?C and room temperature. See
text for film formation conditions and details of extraction of thickness from
ellipsometric parameters. Experimental uncertainty of the thickness is estimated
at ?2.0 ?A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.7 Water contact angle vs immersion time for octadecyltrichlorosilane films on oxi-
dized Si(100) and MVD silica layer at 10 ?C and room temperature. . . . . . . . 46
4.8 Representative FTIR spectra of OTS monolayer formation as a function of im-
mersion time: (Top) OTS monolayer formation on oxidized Si(100) at 10 ?C and
(Bottom) OTS monolayer formation on a oxidized Si(100) at room temperature. 50
4.9 Representative FTIR spectra of OTS monolayer formation as a function of im-
mersion time: (Top) OTS monolayer formation on MVD silica layer at 10 ?C and
(Bottom) OTS monolayer formation on a MVD silica layer at room temperature. 51
4.10 SEM image of reconditioned fibrous MVD silica support with average fiber width
of 10.0 ?m and surface area of 60 m2/g. . . . . . . . . . . . . . . . . . . . . . . 55
4.11 TGA plots of fumed silica (top) before and (bottom) after adsorption of OTS. . 57
4.12 Illustration of fumed silica (Cabosil M5) from manufacturer?s website with a BET
surface area of 200m2/g. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.13 TGA analysis of blank fibrous MVD silica and OTS monolayer films on fibrous
MVD silica fibers (with and without reconditioning) without post treatment. . . 60
4.14 Weight change as a function of thermal treatment under nitrogen from 150 ?C
to 600 ?C at a ramp rate 10 ?C/min for three OTS monolayers on fibrous MVD
silica previously annealed for 1 hour at 120 ?C three different environments (He,
N2 and Air) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.15 Derivative weight loss in (%/min) as a function of thermal treatment under nitro-
gen from 150 ?C to 600 ?C with a ramp rate 10 ?C/min for three OTS monolayers
annealed for 1 hour at 120 ?C under three different environments (He, N2 and Air). 63
x
5.1 Example polysilicon surface and line profile of surface roughness . . . . . . . . . 67
5.2 (Right) Simplified schematic of MVD system (Left) Photograph of MVD Silica
system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.3 Two dimensional AFM scans of (A) a clean silicon surface, RMS roughness is
0.14 nm, and (B) 20 second OTS island film deposited on clean silicon surface ?A?. 80
5.4 (A) AFM image of a MVD silica layer surface with thickness of 50.0 nm and a
RMS roughness of 1.00 nm. (B) AFM image of a 20 second OTS deposition at
room temperature on surface ?A? after 60 seconds of etching. (C) AFM image
of a 20 second OTS deposition at 10 ?C on surface ?A? after 60 seconds of etching. 80
5.5 FTIR analysis of 20 second room temperature OTS island film deposition on an
oxidized silicon ATR and on a MVD silica layer before and after etching. . . . . 81
5.6 FTIR analysis of 20 second 10 ?C OTS island film deposition on an oxidized
silicon ATR and on a MVD silica layer before and after etching. . . . . . . . . . 81
5.7 (Left) Two-dimensional representation of AFM data A-E of MVD deposited silica
layers. (Right) Three dimensional representation of AFM data A-E on MVD
deposited silica layers (2 x 2 ?m) represented. Scaling arbitrarily set to highlight
data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.8 (A-E) 20 second OTS island film deposition on MVD silica surfaces of varying
roughness and topography after exposure to etchant procedure. . . . . . . . . . 82
5.9 Illustration of proposed influence of topographical asperities on packing ability
of OTS molecules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.1 ATR-FTIR Spectra of adsorbed water at 45% relative humidity on various silica
layers ranging in thickness from 18 to 65 nm. Bending mode at 1635 cm?1 is
constant but varying degrees of hydrogen bond strength presented. . . . . . . . 91
6.2 ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on
an oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions
of ice-like 3119 cm?1 and the position of the O-H bending mode is marked at
1632 cm?1. [Inset View] Corresponding AFM Image 5?m x 5?m. Dotted line is
a scan of pure liquid water on the MVD silica layer scaled down in size for peak
position comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.3 ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on an
oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of
ice-like 3225 cm?1 and the position of the O-H bending mode is marked at 1632
cm?1. [Inset View] Corresponding AFM Image 5?m x 5?m. . . . . . . . . . . . 92
xi
6.4 ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on an
oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of
ice-like 3183 cm?1 and the position of the O-H bending mode is marked at 1632
cm?1. [Inset View] Corresponding AFM Image 5?m x 5?m. . . . . . . . . . . . 92
6.5 ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on an
oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of
ice-like 3107 cm?1 and the position of the O-H bending mode is marked at 1632
cm?1. [Inset View] Corresponding AFM Image 5?m x 5?m. . . . . . . . . . . . 93
7.1 ATR-FTIR spectra of adsorbed water on the surface of an oxidized silicon ATR
crystal at increasing relative humidities. (Top) Native oxide exposed to a 30
minute UV ozone treatment with a measured oxide thickness of 1.8 nm. Ice-
like and liquid water are indicated with arrows at 3258 cm?1 and 3400 cm?1,
respectively, and the position of the O-H bending mode is marked at 1635 cm?1.
(Bottom) HF Removal of native oxide followed by Piranha oxidation with an
oxide thickness of 0.9 nm. Ice-like and liquid water are indicated with arrows
at 3255 cm?1 and 3400 cm?1, respectively, and the position of the O-H bending
mode is marked at 1635 cm?1. The direction of increasing relative humidity is
from bottom to top. Spectra correspond to relative humidities of 20%, 40%, 60%
and 80% with an accuracy of ?2.0% . . . . . . . . . . . . . . . . . . . . . . . . 104
7.2 ATR-FTIR spectra of adsorbed water on the surface of an oxidized silicon ATR
crystal at increasing relative humidities. (Top) HF removal of native oxide fol-
lowed by RF oxygen plasma treatment with a measured oxide thickness of 2.0
nm. Ice-like and liquid water are indicated with arrows at 3257 cm?1 and 3400
cm?1, respectively, and the position of the O-H bending mode is marked at 1635
cm?1. (Bottom) HF Removal of native oxide followed by UV Ozone oxidation
with an oxide thickness of 2.0 nm. Ice-like and liquid water are indicated with
arrows at 3259 cm?1 and 3400 cm?1, respectively, and the position of the O-H
bending mode is marked at 1635 cm?1. The direction of increasing relative hu-
midity is from bottom to top. Spectra correspond to relative humidities of 20%,
40%, 60% and 80% with an accuracy of ?2.0% . . . . . . . . . . . . . . . . . . 105
7.3 Comparison interfacial water layer thickness as influenced by cleaning method-
ologies on Si(100) with increasing levels of relative humidity. . . . . . . . . . . . 106
9.1 Water contact angle (squares) and ellipsometric film thickness (diamonds) vs.
immersion time for ODP films on oxidized aluminum oxide surfaces at room
temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
9.2 AFM image of a 5 ?m x 5 ?m plasma cleaned MVD aluminum wafer with RMS
roughness of 4.3 nm and oxide thickness of 2.5 nm. . . . . . . . . . . . . . . . . 117
9.3 AFM image of a 5 ?m x 5 ?m 240 minute deposition of ODP on plasma cleaned
MVD aluminum wafer with RMS roughness of 4.67 nm, contact angle 109.0? and
ODP thickness of thickness of 2.5 nm. . . . . . . . . . . . . . . . . . . . . . . . 117
xii
9.4 Representative FTIR spectra of ODP monolayer formationon ?plasma activated?
aluminum surface as a function of immersion time at room temperature. . . . . 118
9.5 Calculated inclination angles and ICH2/ICH3 ratios as a function of immersion time121
9.6 Electron micrograph of spherical alumina powder with a distribution in size rang-
ing from 75?m to 100?m utilized for ODP deposition and subsequent FTIR anal-
ysis and TGA analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
9.7 Representative FTIR spectra of ODP monolayer formationon ?plasma activated?
aluminum oxide spheres as a function of immersion time at room temperature. . 125
9.8 TGA weight loss as a function of temperature under nitrogen at a ramp rate of
10 ?C/min comparison between two different immersion times. Scan of uncoated
alumina spheres are removed from baseline. . . . . . . . . . . . . . . . . . . . . 126
9.9 Photograph of holder designed for holding AFM Cantilevers, SPM Probes, and
Quartz Crystal Micro balance Assemblies for simultaneous phosphonate mono-
layer deposition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
10.1 Typical Fischer Tropsch product distribution based on chain growth probability. 158
10.2 Representation of gas flow through a micro fibrous catalytic support system. . . 159
10.3 Representation of gas flow through a standard packed bed reactor. . . . . . . . 159
10.4 Calcination of unsupported cobalt nitrate powder in 100 ml/min air and a ramp
rate of 10 ?C/min. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
10.5 Reduction of unsupported Co3O4 with a 100 ml/min 3% H2/ Helium flow and
ramp rate of 10 ?C/min. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
10.6 Calcination of silica-supported cobalt nitrate powder in 100 ml/min Air and a
ramp rate of 10 ?C/min. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
10.7 Water evolution during the reduction of Co3O4 with a 100ml/min 3% H2/Helium
flow and ramp rate of 10 ?C/min as a function of pore size. . . . . . . . . . . . 161
10.8 [Left] MCSS Sheet prior to solution impregnation. [Right] Individual particle
from MCSS Sheet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
10.9 [Left] MCSS Sheet that has been impregnated and dried by blotting method and
subsequently calcinated [Right] Individual particle after calcination. . . . . . . . 162
10.10Illustration of the microreactor system fabricated for this work. . . . . . . . . . 163
10.11Measured pressure drop across the microreactor as a function of compression as
fitted by Darcy?s law. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
xiii
10.12Reynolds number calculations in a MCSS as a function of variable compression,
void volume, and volumetric flow rate. . . . . . . . . . . . . . . . . . . . . . . . 164
10.13F(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 1016 ?m com-
pression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
10.14F(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 508 ?m com-
pression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
10.15E(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 1016 ?m com-
pression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
10.16E(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 1016 ?m com-
pression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
10.17F(t) Break-through curve He ? O2 at variable flow rates, T= 298 K at 1016 ?m
compression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
10.18E(t) Break-through curve He ? O2 at variable flow rates, T= 298K at 1016 ?m
compression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
10.19F(t) Break-through curve He ? O2 at variable flow rates, T= 298K at 508
?mcompression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
10.20E(t) Break-through curve He ? O2 at variable flow rates, T= 298K at 508
?mcompression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
10.21Residence time distribution as a function of temperature at a set flow rate of 5.0
ml/min. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
10.22Photograph of microreactor fabricated for this work. . . . . . . . . . . . . . . . 170
11.1 Reaction diagram of platinum catalyzed PDMS curing. The R group is either
CH3 or H. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
11.2 (A) Illustration of mold platform used for fabrication of the PDMS test devices
and (B) PDMS diffusion cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
11.3 Representation of the gravimetric system utilized for this work. . . . . . . . . . 187
11.4 Representative FTIR spectra of PDMS (1) 24-hr room temperature cure after
30-sec oxygen plasma (2)1hr at 100 ?C followed by 24 hr room temperature cure
after 30 sec oxygen plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
11.5 Example of data collected from TGA-MS experiments. Both weight loss and ion
current are measured as a function of time. Ramp was 10 ?C/min to 95 ?C and
held until until no observable weight loss. . . . . . . . . . . . . . . . . . . . . . 188
xiv
11.6 Diffusion coefficients as a function of temperature. . . . . . . . . . . . . . . . . 189
11.7 Data used to calculate activation energy. . . . . . . . . . . . . . . . . . . . . . . 189
11.8 PDMS film (top), silica coated PDMS film (bottom). . . . . . . . . . . . . . . . 192
xv
List of Tables
5.1 Water contact angle (CA), thickness, and RMS roughness (RMS) for deposited
films. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
10.1 Macro scale conversion data for cobalt impregnated silica supported nickel catalyst.151
xvi
Part I
Interfacial Engineering of Microelectromechanical
Systems
1
Chapter 1
Introduction to MEMS Research
Knowledge of the fundamental mechanisms of friction has remained elusive despite over
300years of careful thought and experimentation [3]. Tribological properties in nano- and mi-
cro scales between two sliding solid surfaces have drawn much attention, as they significantly
affect the performance of microelectromechanical systems and provide crucial information
for understanding the fundamental mechanisms and origin of friction and wear [4]. As me-
chanical devices are made ever smaller in size they are limited by these very phenomena. The
expansion of interest in friction, wear, and lubrication at small length scales has been fueled
by both scientific curiosity and technological need. New high resolution force probes have
provided unprecedented access to fundamental processes at interfaces, which nevertheless
remain poorly understood due to their inherent complexity.
Tribological performance is highly dependent on surface interactions. When two surfaces
are pressed against each other initial solid-solid contact occurs at the surfaces? high points.
This area where the asperities touch is usually an extremely small fraction of the total
area covered by the surfaces, but the forces generated between the contacting atoms are
responsible for most tribological phenomenon. In order to alleviate these adhesion-related
problems, both the topography and the chemical composition of the contacting surfaces must
be studied and controlled. One commonly used method of tailoring the surface properties is
based on deposition of self-assembled monolayers (SAMs).
Polycrystalline silicon, silicon oxide and aluminum oxide are major components of many
MEMS devices found to consist of a multitude of topographical asperities. As SAMs have
been applied to surfaces of varying topographical complexity, many discrepancies associated
with tribological performance have been identified. SAM growth kinetics and interfacial
2
properties as deposited on MEMS devices are often inferred from information learned on
smooth surfaces, due to the inherent difficulties with analyzing surfaces of topographical
complexity. Research indicates that monolayers need to maintain a specific packing density
and molecular orientation as deposited on surfaces of low roughness in order to provide a high
tribological response. In translating these monolayers to surfaces with inherent roughness
and varying topographical asperities, it becomes essential to determine how monolayers
associate on these ?points of contact? as this will be a determining factor for the effectiveness
of tribological improvement provided by the monolayer. This research investigates SAM
adsorption on various surface topographies and various surface types that are similar to
those that exist in MEMS devices. Valuable comparisons are made between surfaces of
higher complexity and surfaces traditionally utilized for SAM observation in an attempt to
determine the influence of topographical asperities on OTS monolayer performance.
1.1 Background Research
1.1.1 Interfacial Phenomenon in MEMS Devices
Micro electro-mechanical systems (MEMS) are miniature size devices that contain sens-
ing and actuating components in very small packages. Several applications, such as ac-
celerometers, digital mirror devices (DMD(TM)), pressure sensors, gyroscopes, resonators,
and many others have been realized using MEMS technologies, and many new ones are under
development [5]. The use of small micromechanical devices has proven difficult as indicated
by the inability to overcome the failure modes and reliability issues of the current test struc-
tures. Commercially successful MEMS devices, to date, have either non moving parts or
contacts whose lateral motion is very restricted. Many more exciting applications can be
attained with MEMS devices consisting of moving, touching, and rubbing structures. These
include gears and motors that can enable much more complicated mechanical functions at
the micro- and nanoscale. However, the effects of adhesion, friction, and wear of MEMS
3
devices are challenging the development and commercialization of more sophisticated micro
machines.
The tribological behavior of contacts in MEMS technologies differs from those in macro-
scopic engineering structures. At the macroscopic scale, millions of asperities give rise to the
parametric relationships described by Amontons law, which states that the friction coeffi-
cient is independent of apparent contact area and applied load. In MEMS, real mechanical
contacts typically consist of a few nanometer-scale asperities that touch. At these small
scales, Amontons law breaks down and individual asperity contact behavior must be con-
sidered [1]. Additionally, forces that are negligible at the macro scale become significant at
the microscopic scale. At the macro scale, gravity dominates over adhesion. However, at
the micro- and nanoscale, the gravitational body forces are negligible and adhesion becomes
significant. Adhesion is generally caused by capillary, electrostatic and van der Waals forces,
and in some cases by chemical forces such as hydrogen bonding in solid bridging [6]. At
separations on the order of a micrometer, there has been reported occurrence of unwanted
sticking phenomenon. This phenomenon aptly named ?stiction? is present during the release
process when the surface tension of the draining rinse liquid draws the micro structure into
contact with the underlying substrate (release stiction). In-use-stiction occurs, when inten-
tionally or accidentally, parts come into contact. Strong interfacial adhesion between micro-
and nanoscale structures (asperities) induces high friction and wear, leading to failure of
MEMS devices.
The topography of MEMS devices and their corresponding asperity profiles are influ-
enced by the fabrication process. Three main MEMS fabrication processes are surface micro
machining (SMM), silicon-on-insulator (SOI) technology, and electro forming. A more in-
depth review of the different micro fabrication methods can be found elsewhere [7]. Each
method employed for material deposition, patterning, and etching to create MEMS struc-
tures, leaves behind unique surface morphologies on both sidewall and in-plane surfaces. An
example of sidewall surfaces from each of the micro fabrication methods indicated above are
4
shown in Figure 1.1 [1]. Because of this surface roughness ranging from a few to a hundred
nanometers, only a few asperities make actual physical contact when opposing surfaces with
similar surface morphologies are brought together under a given load. As these surfaces slide
past one another, some contacts are broken, while new ones are formed. However, the total
number of contacting asperities is relatively constant during sliding, assuming the total load
and adhesive forces do not vary.
Figure 1.1: SEM images of surface morphology of MEMS fabricated by (a) Si micro machin-
ing, (b) SOI, and (c) electro forming processes adapted from reference [1].
Various surface engineering techniques have been introduced to reduce adhesion in
MEMS. One general approach is to avoid the formation of water layers on the surfaces
of MEMS/NEMS by surface treatments that provide a stable hydrophobic surface, thereby
eliminating capillary forces [8]. Another approach is to change the surface interaction forces
by using surface roughening or texturing techniques to reduce the real area of contact [9,10].
Many attempts have been made to create MEMS devices with a specific surface topography
in order to control the tribological processes, unfortunately many of these processes have
proven unreproducible and expensive for mass production. As a consequence, much research
has been dedicated to the chemical modification of MEMS devices with coatings designed to
tailor the interfacial properties. The chemical modification of MEMS devices must render
the surfaces hydrophobic in order to eliminate capillary forces, be stable both under ambient
5
and elevated temperatures, be hard and wear resistant, and electrically conductive to min-
imize charge trapping [6]. In addition, these coatings must be conformal to insure uniform
deposition to MEMS devices.
1.1.2 Overcoming Interfacial Interactions with Self-Assembled Coatings
One commonly used method of tailoring the surface properties is based on deposition
of self-assembled monolayers (SAMs). SAM coatings have been demonstrated to provide
anti-stiction and low friction characteristics [11]. SAMs are composed of molecules that
spontaneously chemisorb and organize into a close-packed assembly on a substrate [12]. They
are ideal for studying the fundamental chemical and physical phenomena, since they can be
attached to a variety of different substrates, form well-defined and uniform interfaces, and
can be tailored structurally to study the resulting effects on friction and other properties [13].
Figure 1.2 illustrates a self-assembling molecule, which includes an anchoring group with
a strong preferential adsorption to the substrate, an alkyl chain and a terminal functionality
(head group). The organization of the self-assembling molecules in SAMs is governed by
the interplay between the packing of the molecules on the substrate, the interaction of
the molecular head groups with the substrate and any interactions between the molecular
terminal groups both with each other and with any liquids of vapors in contact with the
SAM-modified surface.
1.1.3 Self-Assembled Octadecyltrichlorosilane Monolayers as a Candidate for
Overcoming Interfacial Interactions
Octadecyltrichlorosilane SAMs have been demonstrated as one of the most promising
candidates for the tribological improvements applicable to MEMS devices. The formation
and structure of octadecyltrichlorosilane (OTS) monolayers, as deposited by the solution
methods have been studied extensively on various substrates and under different deposition
6
Figure 1.2: Representation of the three parts of self-assembled monolayer illustrating the
head groups, tail groups, and carbon chain backbones.
conditions [2,14?22]. When physisorbed on an appropriate surface, the OTS molecules
organize themselves into crystalline or near crystalline order depending on temperature,
with a terminal end that is hydrophobic. This tendency has proven helpful in for many
applications including biosensors [23,24] microelectronics [25], thin film optics [26], water-
resistant coatings [27], and anti-corrosive coatings [28]. These films have been successfully
deposited on a variety of substrates including silica, alumina, mica, glass [29?32] and are
frequently employed on surfaces that are rough, such as polycrystalline silicon and silicon
nitride [33].
Atomically smooth surfaces have served as model systems allowing the characterization
of the growth kinetics and interfacial chemistry for both full monolayer films and partial
OTS SAMs. The formation of OTS monolayers is postulated to occur through a condensa-
tion reaction between a hydroxylated substrate surfaces and the hydrolyzed chlorine atoms
of OTS. The hydrolyzed OTS molecules physisorb onto the substrate surface via hydro-
gen bonding and finally form Si(Silane)-O-Si(Substrate) bonds and Si(Silane)-O-Si(Silane) cross
linking covalent bonds [34]. This is illustrated in Figure 1.3.
7
Figure 1.3: Proposed reaction mechanism of OTS monolayer adsorption onto a hydrated
surface.
The self assembly process is highly sensitive to experimental conditions such as reaction
time, reaction temperature, solvent type, concentration, and the post deposition immobi-
lization procedure [19,35?41]. Two mechanisms of growth are described in the literature.
In the mechanism known as the ?uniform? model, silane molecules initially are distributed
homogeneously over the substrate in a disordered, liquid like manner [15?17,42]. The aver-
age tilt angle of the adsorbed chains is greater than 15? because of lower packing densities
and gauche conformations for the chains. With increasing coverage, the disordered film is
transformed into a more ordered, quasi-crystalline structure. In the mechanism known as
the ?island? model, domains of densely packed well-ordered, fully extended OTS molecules
are heterogeneously grown on the substrate, and these domains are separated by uncovered
regions of the substrate [2,19]. These islands typically have dimensions on the order of
hundreds of nanometers. With increasing deposition time, more domains are adsorbed until
complete coverage of the substrate is achieved. In the ?island model? OTS film is considered
to be close-corresponding to fully extended chains, with the chemisorbed chains in an all
trans conformation with an average tilt angle <15? away from the surface normal [12]. The
8
two mechanisms of growth are distinguished in the literature based on formation temperature
and will be described in the sections below.
The importance of the water content on the substrate surface in determining the re-
activity, surface coverage, and chain packing of OTS monolayers has been emphasized, and
conflicting results have been presented as to the role of substrate surface properties such as
the availability of free OH groups and the level of hydration required for an ordered OTS
film [17,38,42]. Although the presence of an interfacial water layer has been concluded as
essential, debate still remains as to the degree of hydration necessary for uniform coverage
and to the role this water layer plays in surface attachment of OTS molecules. Angst et
al. found that the degree of hydration on the surface of SiO2 was essential for film order,
showing that a films produced on non hydrated SiO2 were less ordered than films of OTS
deposited on well-hydrated SiO2, providing evidence that the contact angles of water and
thicknesses of the monolayers were lower on the non hydrated surfaces [17]. Wang et al.
used FTIR to studied the adsorption of OTS monolayers on a fully dehydrated and dehy-
droxylated silica surface of fumed silica in comparison to an as received silica surface and
a superhydrated silica. The dehydrated and dehydroxylated surface, which maintains only
isolated or geminal OH groups adopted a disordered structure. In the case of OTS attached
to a hydrated surface known to exist at 3.1 OH/nm2 they found a slight increase in the
trans conformation, but noted that when excess surface water is present on a superhydrated
fumed silica, the trans bands of OTS become even more pronounced suggesting a higher
ordered monolayer [43]. On the other hand, LeGrange et al. deposited OTS on hydrated
and dehydrated substrates of fused silica [38] observing ordered, complete SAMs of OTS on
hydrated substrates and decreased coverages on dehydrated substrates. Tripp and Hair used
infrared spectroscopy to determine that there is no direct reaction of OTS with the surface
hydroxyl groups and the OTS binds only to a small extent with the first water layer on the
surface of fumed silica. It was proposed that a nominal amount of water was needed for
monolayer formation with that too little water resulting an incomplete monolayer, whereas
9
with thick water layers the OTS polymerized [44]. Several groups have shown that the den-
sity of siloxane bonds between the adsorbed silane film and substrate is small [38,41,42]. It
has been speculated that OH groups on the surface affects the mobility of the adsorbed OTS
molecules [18] acting as nucleation and bonding sites for the formation and stabilization of
the monolayers, but without an intervening water layer successful OTS monolayer formation
is restricted. Parikh and Brzoska suggested that a mobile water layer on the surface was
necessary to decouple the OTS monolayer from the underlying substrate allowing improved
lateral packing of the hydrocarbon tails. However, they concluded that on flat surfaces,
it was not possible to directly measure the amount of adsorbed water or to determine the
hydration state of the substrate [19,41].
Partial OTS monolayers deposited on smooth surfaces have revealed that the self-
assembly process results in the formation of films with distinct characteristics as a function of
deposition temperature. These regimes are illustrated in Figure 1.4. Each regime is charac-
terized by changes in molecular arrangement including packing density, chain conformation,
and orientation with respect to the substrate [45]. The occurrence of three distinct temper-
ature regimes have been equated to the notion of a transient Langmuir film during SAM
formation [2,41]. The (hydrolyzed) precursor molecules retain their mobility for sufficiently
long times allowing for the observation of transformations between different condensed two
dimensional phases of the film [40]. The existence of two characteristic temperatures have
been determined, T0 and Tc. For T < T0, self-assembly proceeds initially by island growth
only, the substrate between islands remaining essentially unfilled; such a growth pattern is
expected if a Langmuir film nucleates liquid-condensed (LC) domains coexisting with the
gas phase below the triple point. Thus, T0 can be identified with Tt. For T > T0, island
growth is observed to occur more slowly, while the substrate between islands is gradually
filled in with OTS molecules. This growth regime, which has been observed by various
authors, is attributed to nucleation of LC domains coexisting with a liquid-expanded (LE)
phase [46,47]. This mechanism operates only up to a characteristic temperature Tc, above
10
which there is no island nucleation but homogeneous growth instead. More specifically, for
low temperatures of T < 16 ?C, the OTS-based monolayer forms by island growth, the is-
lands consist of fully extended chains, and essentially no OTS molecules are found on the
substrate between islands. For T > 40 ?C, growth is homogeneous and disordered with no
island growth observed. For 16 ?C< T < 40 ?C, islands of extended chains are observed
(islands) surrounded by a uniform background of chains that are disordered (e.g., contain
gauche defects). Figure 1.4 is adapted from the works of Carraro et al. and illustrates the
three growth regimes.
Figure 1.4: Submonolayer growth regimes for OTS film formation as influenced by temper-
ature adapted from reference [2].
Thermal stability
Investigations of the thermal degradation mechanisms of OTS monolayers, evidence
that, the thermal stability of OTS monolayers is dependent on the environmental conditions
under study. Kluth et al. studied the thermal decomposition mechanism on Si(100) surfaces
by annealing under vacuum, finding that decomposition occurs through cleavage of C-C and
Si-C bonds followed by the removal of the headgroup Si-O. Results indicate that the silicon
11
head groups remain on the surface even after the monolayer has decomposed until about 827
?C [48]. In contrast, Kim et al. investigated OTS thermal decomposition in air finding that
the monolayers are stable up to about 220 ?C and above 220 ?C the monolayers primarily
decompose through C-C bond cleavage, with a gradual reduction in chain length. They
also show the siloxane headgroups remaining on the surface following the decomposition
of hydrocarbon fragments until about 480 ?C, offering, that an air environment accelerates
the decomposition process [49]. Other studies have shown that water contact angles on
octadecyltrichlorosilane (OTS)-coated surfaces remain the same upon annealing for 5 min to
temperatures as high as 400 ?C, in a nitrogen environment, and that above this temperature
the contact angle was observed to decrease, indicating degradation of the films [50]. Whereas,
Calistri-Yeh et al.. have studied the thermal stability of OTS SAMs on Si(100) by annealing
their specimens for 2 h at temperatures over 125 ?C in air with observed permanent structural
changes of octadecyltrichlorosilane (OTS) monolayers in air above 127 ?C [51]. Mirji et
al. examined OTS monolayers on planer Si(100)/SiO2 and mesoporous SBA-15 substrates
finding that the OTS layers are thermally stable up to a temperature of 260 ?C in air [29].
No direct comparison between different annealing conditions and the thermal decomposition
temperatures has been examined or determined in the literature.
Chemical and Immersion Stability
Immersion stability studies of organosilane monolayers deposited on silicon oxide sur-
faces, indicate that, while these films are stable in dry environments for long periods of
time [52], the stability is severely compromised in aqueous conditions. Maccarini investi-
gated the immersion stability of OTS films on silicon, showing that, the films did not exhibit
any appreciable changes after 24 h of exposure to water [53]. Extending the water expo-
sure time to the OTS film has yielded different results. For example, Parker monitored the
contact angle of various monolayers on silicon oxide surfaces in a five week immersion ex-
periment, concluding that, OTS monolayers immersed in water were relatively stable [54].
12
While the work of Anderson et al. obtained similar contact angle data over a 5 week period,
but included additional ellipsometric and AFM data that suggested the OTS monolayers
stability was compromised [55].
The chemical stability of OTS SAMs has not been reported extensively in terms of
solvent interactions. Iimura et al. studied the chemical stability in organic solvents (e.g.
benzene, ethanol, chloroform), in acidic (pH 1) conditions and in alkaline solutions (pH
13) observing no dramatic changes in the intensity of the observed CH2 stretching bands
in all three conditions [56]. In contrast, Geerken et al. investigated the chemical stability
of OTS in different aqueous solutions with varying pH-values for several days, finding that
the coatings are stable from pH 2 to pH 10. But, found that when treated with sodium
hydroxide (pH-13), the contact angle decreased dramatically suggesting that the Si-O bond
is susceptible toward hydrolysis at alkaline conditions [57].
1.1.4 Disparity in Tribological Improvements
For any given pair of interfaces, there are many possible frictional mechanisms at work
including phononic, electronic, and viscoelastic dissipation [58]. The recent development of
experimental tools, like thesurface force apparatus(SFA)[59]andscanning probemicroscopy
(SPM) has allowed a molecular level study on the friction and wear [60]. Researchers have
used model systems such as mica, graphite, atomically smooth flat metals, and self-assembled
monolayers to reveal the details of these underlying phenomena. It is well understood that
frictional behavior and appropriate models to describe this behavior are intimately linked to
length scales, with measurements taken on the nanoscale often not translating into reliable
information on the micro scale and even still different than macro scale measurements. Re-
search shows that values measured for microscopic friction are much lower than the values
obtained at the macro scale. As explained by Bhushan, contact stresses at low loadings do
not exceed the sample hardness, which minimizes plastic deformation. And when measured
for small contact areas and very low loads used in microscopic studies, indentation hardness
13
and modulus of elasticity, are higher than the macro scale. Lack of plastic degradation and
improved mechanical properties reduce the degree of wear. In addition, the small apparent
areas of contact reduce the number of particles trapped at the interface, thus minimizing the
?ploughing? contribution to frictional force [61].
In addition, there are several challenges in associating the frictional properties observed
with AFM to tribological behaviors of MEMS. The sliding velocity of the contact in AFM
is typically 24 orders of magnitude slower than that of real contacts in MEMS. Because of
the sharpness of the AFM contact, the contact pressure is typically on the order of a few
gigapascals, which is several orders of magnitude higher than the contact pressures expected
in real systems. The frictional response at the nanoscale contains an adhesive component.
The adhesive friction components decrease as the size of contact increases from the nanoscale
to the micro scale, and eventually become negligible in the macro scale. Finally, relating
the behavior of one single asperity to larger contact phenomena requires understanding how
to model multiple contacts that are continually being broken, reformed, and sheared at
different contact pressures at which elastic and plastic deformation may constantly occur.
The integration of single asperity behaviors to capture the behavior of a real rough interface
has not yet been achieved. Most MEMS applications lie in micro structural level resulting in
the development of in-situ direct measurement platforms using micro machined structures
which measure stiction [62] and friction and wear [63,64] at the micro scale.
Investigations of SAM coated smooth surfaces, despite the measurement scale, have
shown that the frictional properties are influenced by molecular tilting, inter atomic interac-
tions, interface stability, adhesion to substrate, and gauche defect formation [65]. Dissipation
of frictional energy in SAMs through the formation of gauche defects has been thoroughly
investigated by Salmeron and coworkers, concluding that increased disorder correlates with
higher gauche defects and an easier dissipation of frictional energy [66]. The tribological
behaviors of several self-assembled monolayers have been reported [25,33,67?69], and one
14
characteristic of these films is that friction has been shown to be closely associated with film
structure [70?73].
SAMs considerably reduce friction and adhesion proving useful in various MEMS devices
[64]. Unfortunately, as these films have been applied to the surfaces of MEMS devices, the
tribological improvement afforded by these films is consistently less than reported on smooth
surfaces regardless of the measurement technique and length scale examined. In addition,
a variation in the improvement afforded exists across the same length scale. Because most
MEMS devices consist of a multitude of topographical features dependent on the fabrication
process, it is suggested by this author that the variation in friction measurements appears
to not only be linked to the measurement range, but dependent on a combination of the
topography of the surface and interfacial coating quality. This is a fundamental issue and
requires a more in depth analysis of OTS film formation on MEMS devices.
1.1.5 Self-Assembled Octadecylphosphonic Acid Monolayers as a Candidate for
Overcoming Interfacial Interactions
Phosphonate monolayers are a new class of SAMs that show significant potential to
define new directions in MEMS technology. Phosphonate SAMs provide a model system for
exploring the relationship between friction and molecular structure, as well as for determin-
ing how attaching SAMs to a variety of substrates influences friction. Moreover, phosphonate
SAMs are technologically important because they can be deposited on mica and many dif-
ferent metal oxide surfaces, unlike the more commonly studied alkane thiol SAMs [74?76].
These materials have also been deposited on conventional silicon surfaces although some
controversy exists as to whether or not these material maintain the same stability as there
self assembled counterparts on metal oxide surfaces [77]. These material have the potential
to overcome the limitations of silane-based SAMs with respect to immersion and thermal
stability [78].
15
The necessity for an ultra thin, lubricant film to minimize adhesion, friction, and wear
between surfaces in contact for MEMS/NEMS is clear. The tribological mechanisms for sur-
faces coated with phosphonate SAMs have not yet been studied extensively, although these
monolayers have recently begun to attract the attention of tribology research groups [25,79].
Frictionforce measurements with thephosphonate monolayer coated smoothalumina demon-
strated a remarkable improvement in performance when compared to untreated alumina
surfaces the coated surfaces exhibited remarkable robustness and resistance to wear over
each experiment. In addition to offering a physical protective barrier, the monolayers, by
virtue of their inert hydrophobic nature, also provide a degree of chemical protection. The
hydrophobic coatings have been shown to repel humidity and protect alumina substrates
from embrittlement caused by water, and the nature of the phosphonate alumina bond also
appears to prevent chemical attack of the alumina by any water that does penetrate the
monolayer [80].
A study of alkylphosphonic acids on flat Al sheets showed that SAM formation of
alkyl phosphonates on Al surfaces relies on the hydroxylation of the oxide (alumina) layer.
Generally, chemisorption of alkylphosphonic acid occurs by proton dissociation to form an
alkylphosphonate species. The phosphonic acids undergo a condensation reaction with sur-
face bound aluminohydroxyl (-Al-OH) species to form aluminophosphonate compounds with
H2O as a byproduct [81]. An inelastic electron tunneling spectroscopy study of the adsorp-
tion and structure of alkyl phosphonic and alkyl phosphoric acids suggested that tridentate
species were formed with the aluminum oxide surface [82]. It has been shown that ODP ad-
sorbs preferentially to Aluminum surfaces that are hydroxylated in air and that an increase
in the surface hydroxyl density can accelerate adsorption kinetics [83]. Such an acceleration
can be explained by the adsorption of the phosphonic acid via surface hydrogen bonds prior
to the condensation reaction, leading to the finally adsorbed phosphonate [84]. Due to the
nature of the surface reaction between alkylphosphonic acids and OH surface species this self
16
assembled monolayer offers a unique opportunity for tribological improvement and should
be compared with the existing knowledge of OTS monolayers.
17
Chapter 2
Investigative Summary for Part I
Part I (Chapters 4-8), investigates the interfacial structure of self-assembled coatings as
influenced by surface topography and surface reactivity to unravel some of the discrepancies
associated with tribological performance. Initial work focuses on OTS monolayers, identified
as one of the most promising candidates for the tribological improvements, restricted in use
due to inconsistent results when translated to MEMS surfaces. Literature reveals many
factors relevant for the production of consistent monolayers. This work attempts to expand
on that knowledge by including a ?missing link?- Is it possible for monolayers to maintain
their molecular order and conformational strength on topographical asperities?
Tocreate surfaces ofcomparable topography, as encountered onMEMS devices, a Molec-
ular Vapor Deposited (MVD) silica layer is introduced. This surface is a novel platform,
with a tunable topography amenable to advanced surface analysis, often restricted to use
on smooth surfaces. The reactivity of the MVD silica surface toward OTS film formation is
compared to traditional silicon oxide surfaces with an examination of film formation kinetics
and interfacial water structure. The MVD silica layer allows for the visual examination of
the self assembly process on individual topographical asperities, in conjunction with, a spec-
troscopic examination of the reactive surface and the final conformational order of the films.
Interfacial water structure is identified as a key element in preparing ideal OTS monolayers
and a correlation between water structure and surface topography is identified and explored.
In addition, this silica layer is not a line of sight deposition and has been translated
onto surfaces of varying geometrical complexity, enabling the production of a high surface
area material. This material is utilized for studies of monolayer loading and the influence of
curing mechanisms on the stability of OTS films.
18
This work is followed in Chapter 9, by an examination of the self assembly of octade-
cylphosphonic acids on aluminum oxide surfaces. This work is aimed at producing high
quality films on surfaces with varying degrees of curvature (spherical balls) and on sur-
faces that exhibit interfacial surface roughness. The expanded goal is comparing tribological
characteristics across nanoscale and micro scale regimes. For consistency these surfaces are
produced under identical conditions and at the same time. An investigation of the deposition
mechanism of ODP films is performed on an aluminum oxide surface and compared with a
spherical aluminum oxide powder. After determining the appropriate deposition character-
istics, these films are translated onto QCM crystals, and SPM probes / AFM cantilevers,
with spherical aluminum oxide balls of 100 ?m diameter attached for tribological analysis
by an outside collaborating team.
19
Chapter 3
Methodology
3.1 Analytical Tools and Techniques
In the study of monolayers and thin films on surfaces, both their surface and bulk
properties are important. In this chapter, several analytical methods are introduced for
characterization of the monolayers and thin films produced throughout this document. Con-
tact angles with water as liquid are measured to evaluate wetting properties and to get
information on surface order. Ellipsometry is used to measure the thickness and uniformity
of freshly prepared films. FTIR spectroscopy is used to learn about molecular orientation,
packing behavior and coverage. TGA analysis is used to obtain material loading information
in conjunction with surface imaging techniques (SEM and AFM) to learn about the surface
topography and particle size distributions.
3.1.1 Contact Angle
To study the hydrophobicity and the uniformity of the modified substrates, the static
contact angle of a water drop lying on the substrate (sessile drop) was measured. The
quality of a stable monolayer and thin film can be estimated from wetting measurements.
The contact angle is not a property of the liquid or the substrate but the interaction between
them. The angle of a drop on a solid surface is the result of the balance between the cohesive
forces in the liquid and the adhesive forces between the solid and the liquid. If there is no
interaction between or wetting the solid and the liquid, the contact angle will be 180?.
As the interaction between the solid and liquid increases, the liquid spreads until ? = 0?.
Water contact angle measurements are obtained by the sessile drop method on a Ram?e-Hart
20
model 200 automated goniometer (Ram?e-Hart, Inc. Mountain Lakes, NJ) using DROPimage
Standard software. Measurement error for this technique is ?2.0?.
3.1.2 Ellipsometry
An ellipsometer measures the changes in the polarization state of light when it is re-
flected from a sample. If a thin film on a surface changes in thickness, then its reflective
properties will also change. Measuring these changes in the reflective properties can allow
for a determination of the actual change in film thickness. The most important application
of ellipsometry is to study thin films. In the context of ellipsometry a thin film is one that
ranges from essentially zero thickness to several thousand angstroms. The sensitivity of an
ellipsometer is such that a change in film thickness of a few angstroms is usually easy to
detect. For a clean surface, ? and ? are the output of an ellipsometry measurement and can
be interpreted into layer thickness provided the right model for the surface is utilized. These
two parameters are directly related to the complex index of refraction (n) of the surface. In
order to interpret the ellipsometry data into layer thickness it is necessary to know or assume
values for the refractive index for each separate layer on the surface.
?n = n(1?ik) (3.1)
The complex refraction index is a representation of the optical constants of a material, it is
represented by Equation 3.1 where n is the ordinary refraction index and k is the extinction
coefficient. The real part or index of refraction, n, defines the phase velocity of light in
material:
? = c/n (3.2)
where ? is the speed of light in the material and c is the speed of light in vacuum. The
imaginary part or extinction coefficient, k, determines how fast the amplitude of the wave
decreases. The extinction coefficient is directly related to the absorption of a material and
21
is related to the absorption coefficient by
? = 4pik/? (3.3)
where ? is the absorption coefficient and ? is the wavelength of light. Once a film having
a different index of refraction nf is coated on the surface, ? and ? are related to the
complex indices of both the film and the substrate, and to the film thickness. Typically, the
compensator is set at an angle of 70? and the experimental data are expressed as
?? = ?0 ?? (3.4)
?? = ?0 ?? (3.5)
Where ?0 and ?0 are the ellipsometric angles characteristic of the clean substrate surface,
and ? and ? are those measured for the substrate with the film. From n and k, we can
calculate the thickness of the film.
In the experiments that follow, ellipsometric measurements are conducted on a nulling
type ellipsometer (Rudolph AutoEL III, Rudolph Research, Fairfield NJ) equipped with a
He-Ne laser (? = 632.8 nm) at a 70? angle of incidence relative to the surface normal. Herein,
reported values for film thicknesses are the averages from four different positions measured
on each sample.
For films near the cycle thickness for the particular substrate-film combination under
investigation (extremely thin films fall into this category), simultaneous determination of
both the film thickness and refractive index is difficult, given that the measured parameters,
the angles ? and ?, are very insensitive to changes in refractive index in these regimes. For
this reason, an accepted value for the refractive index of silicon oxide, n = 1.46, is used in this
investigation for measurements of silicon dioxide produced from oxygen plasma and MVD
silica layer treatment. The refractive index of the silicon surface is set at n = 3.858+0.018i.
22
For ellipsometric measurement of the organic films on silicon oxide, a single film model
has been proposed [85] and is used in this work. Briefly, the thickness of the initial oxide
layer is measured using a refractive index of n = 1.46. After the organic film is deposited, the
sample is remeasured using the same refractive index and the total thickness of the organic
layer is the difference between these two measurements.
For ellipsometric measurement of the organic films on aluminum oxide, a double film
model is used. A single model is used to determine the initial thickness of the aluminum
oxide layer with set values of ks = 6.384 and ns = 1.428 and nl = 1.615. After determination
of the initial oxide layer thickness tl the values are used to determine the thickness of the
organic film on top of the aluminum oxide layer. The sample is re-measured using a refractive
index of 1.45 and the total thickness of the organic layer is reported.
3.1.3 Atomic Force Microscopy
Atomic force microscope (AFM) images are collected in air (21 ?C,?45% relative humid-
ity) on a commercial AFM (Pacific Nanotechnology Nano-R AFM, Pacific Nanotechnology,
Santa Clara, CA) in non-contact mode using silicon AFM tips with resonant frequencies
in the range of 150-210 kHz and force constants in the range of 4.5-14 N/m (MikroMasch,
Wilsonville, OR). Images are collected at a scan rate of 0.8 to 1 Hz. Optimal tip geometry
is represented by a pyramid shape with an ideal tip radius of 10 nm.
3.1.4 Fourier transform infrared (FTIR) spectroscopy
Specular reflectance FTIR spectroscopy
Infraredreflection absorptionspectroscopy (IR-RAS),alsoknown by thesynonym RAIRS
and specular reflectance FTIR has been used for nearly 40 years to study thin films deposited
or adsorbed on metal surfaces. The high reflectivity of the metal surface, combined with a
grazing angle of incidence, give a relatively long path length in the thin film with the result
that nanometer-thick films can readily be studied. For s-polarized light, the reflected beam is
23
out-of-phase with the incident beam, giving a node in the electric field at the surface. For p-
polarized light the two fields are in phase, resulting in a strong preference for vibrations with
transition dipoles normal to the surface: the so-called surface selection rule. Since s-polarized
light carries no interfacial information, s-polarized light is commonly used as a background
scan. As a result, a ration of the s-polarized light scan with the p-polarization scan, is often
used used to eliminate artifacts arising from bulk-phase absorptions. This technique is very
useful giving information about the direction of the transition dipoles of an adsorbed organic
thin film. Theoretical consideration of the IR spectroscopy of the monolayers adsorbed on a
metal surface showed that the reflection-adsorption spectrum is measured most efficiently at
high angles of incidence, typically at 80-85?, and that only the component of incident light,
which is parallel to the plane of incidence, gives measurable absorption [86]. Figure 3.1
presents a schematic description of a mono molecular film on a mirror, with an incident light
and direction of the polarization. In order to study orientation, polarized light is directed
on the film. The polarized light electrical vector coinciding with the dipole of the infrared
active moiety increases in absorption intensity, thereby revealing the band assignment and
the orientation of the molecular group.
Figure 3.1: Illustration of the polarized light electrical vector as influence by p-polarization
at an angle of incidence greater than 80 ?.
In the results that follow, phosphonate monolayers, were characterized using FTIR
spectroscopy (PerkinElmer Spectrum 2000 spectrometer) equipped with a liquid nitrogen
cooled in MCT detector. The samples were analyzed using a specular reflectance accessory
with an angle of incidence of 83?. The intensity of the methylene vibrations in the spectra
24
is a direct function of the tilt angle of the alkyl chain. This unique property of the grazing-
incident experiment allows the calculation of the molecular orientation from FTIR spectra.
For example, as illustrated in Figure 3.2, the molecule on the right illustrates an optimal
rectangular adsorbed molecule of an alkyl phosphonic acid molecule on a metal surface with
both the symmetric and asymmetric methylene (CH2) vibrations (?s and ?a), parallel to
the metal plane. Therefore, the methylene groups in a perpendicular, all-trans optimal
configurations would not have any transition dipole moments in the direction normal to the
investigated surface which is necessary for signal. As a consequence the alkyl chain will not
be picked up by the p-polarized light. Once the alkyl chain tilts from the normal to the plane
as shown Figure 3.2 on the left, the symmetric and asymmetric vibrations of the methylene
groups are no longer are parallel to the surface and thus will appear in the grazing incident
spectra. A measure of the degree of order for films investigated with specular reflectance
spectroscopy can be obtained by the ratio of the intensity of the symmetric and asymmetric
methylene groups.
Figure 3.2: Influence of molecular tilt on the orientation of methylene stretches for phos-
phonate molecules on metal surfaces. Methylene stretches from an all trans (perpendicular
orientation) are not identified with specular reflectance spectroscopy (p-polarized IR) due to
an electric field oriented normal to the substrate at an angle of incident of > 80?.
25
Attenuated Total Reflection Spectroscopy
Attenuated total reflectance (ATR), is a infrared spectroscopy technique, which enables
samples to be examined directly in the solid or liquid state without further preparation. ATR
uses a property of total internal reflection called the evanescent wave. A beam of infrared
light is passed through the ATR crystal in such a way that it reflects at least once off the
internal surface in contact with the sample. This reflection forms the evanescent wave which
extends into the sample, typically by a few micrometers. The beam is then collected by a
detector as it exits the crystal. In order to utilize the evanescent effect crystals are utilized
that are made from an optical material with a higher refractive index than the sample being
studied. In the case of a liquid sample, pouring a shallow amount over the surface of the
crystal is sufficient. In the case of a solid sample, it is pressed into direct contact with the
crystal to ensure a more intimate contact avoiding any trapped air as that would distort the
results.
To obtain internal reflectance, the angle of incidence must exceed the so-called critical
angle. This angle is a function of the real parts of the refractive indices of both the sample
and the ATR crystal:
?c = sin?1(n2/n1) (3.6)
Where n2 is the refractive index of the sample and n1 is the refractive index of the
crystal. The evanescent wave decays into the sample exponentially with distance from the
surface of the crystal over a distance on the order of microns. The depth of penetration of
the evanescent wave dp is defined as the distance form the crystal-sample interface where
the intensity of the evanescent decays to 1/e (37%) of its original value. It can be given by:
dp = ?/2pin1[sin2??(n2/n1)2]12 (3.7)
26
Where ? is the wavelength of the IR radiation. For instance, if the ZnSe crystal (n1=
2.4) is used, the penetration depth for a sample with the refractive index of 1.5 at 1000 cm?1
is estimated to be 2.0 ?m when the angle of incidence is 45?. If the Ge crystal (n1= 4.0)
is used under the same condition, the penetration depth is about 0.664 ?m. The depth of
penetration and the total number of reflections along the crystal can be controlled either by
varying the angle of incidence or by selection of crystals. Different crystals have different
refractive index of the crystal material.
In the results that follow, OTS monolayers and interfacial water layers, were character-
ized using FTIR spectroscopy (PerkinElmer Spectrum 2000 spectrometer) equipped with a
liquid nitrogen cooled in MCT detector. The samples were analyzed using horizontal atten-
uated total reflection (HATR) accessory, which consists of a steel plate with an attenuated
total reflection (ATR) element sealed to it that rests on top of an optical alignment box. The
optical alignment box contains two planar mirrors and focal mirrors which can be adjusted
to focus the infrared beam on the incident face of the ATR crystal. All spectra are recorded
at room temperature at 4.0 cm?1 resolution with 500 co-added scans.
Throughout this document both trapezoidal silica ATR elements and ZnSe ATR ele-
ments are used (50 x 20 x 2 mm3, 45? face angle) enabling acquisition of spectral data from
4000 to 1500 cm?1 for the silicon and 4000 to 550 cm?1 for the ZnSe. The refractive index
of silicon is nearly constant over the spectral range under study, and a value of 3.42 is used
herein. The refractive index is about 2.67 at 550 nm, and about 2.40 at 10.6 ?m. For water,
the refractive index at 1635 cm?1 is taken as 1.32 [87]. Consequently, the critical angles for
total internal reflection for the systems under study are as follows: for silicon/water, 22.7?
Since the HATR accessory operates at a fixed angle of incidence of 45?, and the calculated
critical angles are far less than 45?, dispersion effects can be neglected [86].
27
3.1.5 TGA Analysis
This system consists of an isolated thermo balance, an IR furnace, quartz sample cham-
ber, purge inlets and outlets, and a capillary connection to a mass spectrometer. Sample
weight loss is measured as a function of temperature under various environments. The
thermo balance is maintained at 40 ?C in a well insulated, gas purged chamber isolated from
the furnace by a water cooled plate. Evolved gas products are analyzed by a quadrapole
mass spectrometer (Pfieffer Thermostar) via a heated transfer line. The technical specifica-
tions for this unit can be found in more detail on the manufacturers? website, but include
a weighing accuracy of ?0.1% and a weighing precision of ?0.01%. The furnace consists
of four symmetrically placed IR lamps with a quartz lined sample compartment that is a
chemically inert and resistant to adsorption of off-gas products.
3.2 Experimental Techniques
3.2.1 Sample Preparation and Cleaning
Plasma Cleaning
Silicon samples are prepared by dicing the silicon wafers into 8mm ? 8mm squares.
Both silicon samples and silicon ATR crystals are cleaned using the same procedure as
described below. The silicon samples are sonicated in acetone for ten minutes and then
again in isopropanol for ten minutes and then dried under a stream of nitrogen. Samples
are etched for ten minutes in concentrated HF to remove the native oxide layer, rinsed in
copious amounts of deionized water, and then dried under a stream of nitrogen. Samples
are then loaded into a custom-built vacuum deposition system, the design of which is based
on a previously described system [88], for oxygen plasma treatment. Briefly, the vacuum
deposition system consists of a rotary vane pump (base pressure 0.4 Pa) which is coupled to
a glass reaction chamber that contains perforated electrodes which can be biased to create an
in situ, capacitively-coupled radio frequency (RF) plasma (13.56 MHz). The glass reaction
28
chamber is also coupled to a vapor delivery system which allows for the introduction of
the vapors of various volatile precursors into the glass reaction chamber. MKS Baratron
capacitance manometers are used to monitor the system pressure up to 1.3?104 Pa. A
simplified schematic and photograph of the vacuum deposition system can be seen in Figure
3.3.
Figure 3.3: Schematic (left) and photograph (right) of the custom vacuum deposition system
and FTIR employed in this work.
After introducing the samples into the vacuum system, the system is evacuated to a
pressure of less than 2.6 Pa. Oxygen gas is then allowed to flow through the system, and
an oxygen background is established by multiple pump-purge cycles with oxygen gas. Then,
the oxygen pressure is allowed to stabilize around 33.3 Pa, at which point the chamber is
isolated from the pump. An RF plasma is struck at 25 W forward power, and the samples
are exposed to the plasma for five minutes. This treatment grows an oxide layer about
2.0 nm thick on the silicon surface. The samples are removed from the vacuum deposition
system and etched again for ten minutes in concentrated HF, rinsed in copious amounts
of deionized water, and dried under a stream of nitrogen. An additional oxygen plasma
treatment is employed to re-grow an oxide layer. This iterative cleaning method is repeated
until inspection by contact angle and AFM indicate that a clean (contact angle < 5?), flat
(RMS roughness ?0.2 nm) surface has been obtained, but usually two iterations suffice.
29
UV Ozone Cleaning
Etched Silicon Samples
Silicon samples and silicon ATR crystals are cleaned using the same procedure as described
below. The silicon samples are sonicated in acetone for ten minutes and then again in
isopropanol for ten minutes and then dried under a stream of nitrogen. Samples are etched
for ten minutes in concentrated HF to remove the native oxide layer, rinsed in copious
amounts of deionized water, and then dried under a stream of nitrogen. The samples are
then placed into a UV Ozone cleaning chamber and treated with UV ozone in air. Samples
are placed approximately 1.0 cm below the UV ozone bulbs. The treatment duration is
10 minutes with a measured oxide thickness of about 2.0 nm thick on the silicon surface.
Inspection by contact angle indicate that a clean reactive oxide surface with a contact angle
< 5?.
Native Silicon Samples
Treated with the same procedure as described above except etching step is neglected and
the treatment duration is extended to 30 minutes. The measured oxide thickness about 2.0
nm thick on the silicon surface. Inspection by contact angle indicate clean reactive oxide
surface with a contact angle of < 5?.
Piranha Cleaning
Silicon samples and silicon ATR crystals are cleaned using the same procedure as de-
scribed below. The silicon samples are sonicated in acetone for ten minutes and then again
in isopropanol for ten minutes and then dried under a stream of nitrogen. Samples are
etched for ten minutes in concentrated HF to remove the native oxide layer, rinsed in co-
pious amounts of deionized water, and then dried under a stream of nitrogen. Samples are
exposed to a Piranha solution consisting of (3:1 v/v) H2SO4 and H2O2 for a 30 minute du-
ration. The measured oxide thickness was ? 1.0 nm thick on the silicon surface. Inspection
by contact angle indicate that a clean reactive oxide surface with a contact angle < 5?.
30
MVD Silica Layer Deposition
Deionized water and tetrachlorosilane are loaded into clean glass vials, and then con-
nected to the vapor delivery system. The precursors are degassed by employing multiple
freeze-pump-thaw cycles with liquid nitrogen. For vapor coating, freshly cleaned silicon
samples are loaded into the vacuum deposition system. The system is evacuated until the
pressure is less than 2.6 Pa, and then the samples are treated with an oxygen plasma by
the method described in Section 3.2.1 for substrate surface conditioning. After the plasma
pre-treatment, the chamber is again evacuated to a pressure less than 2.6 Pa, and then water
vapor and tetrachlorosilane vapor is admitted into the chamber and allowed to react while
being isolated from the pump. Detailed precursor dosing requirements associated with the
various silica layer thicknesses have been previously reported and will not be discussed fur-
ther. Using the taxonomy of Anderson, [55] the silica layers produced in this work range
from type I to type IV surfaces when referring to the reference article previously indicated.
OTS Deposition
Prior to coating, freshly cleaned silicon samples and silicon ATR crystals are treated
with an oxygen plasma by the method described in Section 3.2.1. Samples coated with
MVD silica layer are used immediately after their preparation. The samples are coated with
octadecytrichlorosilane by the method described below. Clean silicon samples and MVD
silica layer samples are coated simultaneously from a single coating solution. It is also
important to note that all glassware used were cleaned via the same method as the silicon
wafers and is pre-coated with the appropriate precursor to eliminate competitive adsorption
with the walls of the glassware. OTS growth is carried out under atmospheric conditions
with about ?45% relative humidity. Freshly cleaned silicon samples and freshly prepared
silica layer samples are placed into a 50 ml beaker of Optima grade hexanes either chilled
to 10 ?C or at room temperature and allowed to equilibrate for 2 minutes prior to OTS
deposition. Solutions of OTS (0.5 mM) are prepared in 50 ml of Optima grade hexanes in a
31
secondary beaker. To grow liquid condensed islands, the OTS adsorption reaction is allowed
to proceed for the specified times at 10 ?C, significantly lower than the critical temperature
for the LC phase in this system. Then the samples are removed, rinsed in two stages of
neat hexanes that are also chilled to 10 ?C. The samples are then dried under a stream of
nitrogen. The samples are allowed to set for 1 hour at room temperature prior to analysis. A
similar procedure is utilized for deposition of the room temperature (mixed regime) samples
at a deposition temperature of 22 ?C.
Cleaning of Aluminum and Aluminum Oxide Samples
Cleaning of Alumimum Samples
E-Beam Sputtered aluminum samples with 4500 ?A of aluminum metal on 1000 ?A of ther-
mally grown silicon oxide are diced into into 8mm ? 8mm squares and 2.54cm ? 5.08cm
rectangles. Prior to use the aluminum samples are sonicated in acetone for ten minutes and
then again in isopropanol for ten minutes and then dried under a stream of nitrogen. Prior to
coating, freshly cleaned sputtered aluminum samples are treated with an oxygen plasma by
the method described in Section 3.2.1 for 2.0 minutes to activate the surface. Inspection by
contact angle and AFM indicate that a hydrophillic clean surface with contact angle < 5?)
and surface roughness of ?4.3 nm RMS. The measured aluminum oxide thickness was ?2.5
nm.
Cleaning of Aluminum Spheres
Aluminum oxide spheres were obtained from Denka Corp. (China) and sieved to collect
particles ranging in size from 75 to 100 ?m. The BET surface area for the spheres gives a
surface area of 0.2 m2/g. The sieved aluminum oxide spheres are sonicated in acetone for
10 minutes and filter dried and then sonicated again in isopropanol for 10 minutes and filter
dried. The spheres are then placed into an oven at 120 ?C for 1 hour for final drying. Prior to
coating, the spheres are treated with an oxygen plasma by the method described in Section
32
3.2.1 for 2.0 minutes to activate the surface and to remove hydrocarbon contamination. The
uncoated samples are placed into an oven set at 60 ?C for 10 minutes prior to analysis.
Phosphonate Deposition
A 5.0 mM phosphonate solution is prepared by sonicating 200 mg of Octadecylphos-
phonic Acid (97% Strem Chemical [4724-47] in 100 ml of 2-Propanol. After activation of
the aluminum surface with plasma treatment, as decribed above, the samples are submerged
into the 5.0 mM solution for different time durations. These samples are removed from
solution and rinsed 3 times in neat 2-Propanol and dried under a stream of nitrogen. For
a 4 hour deposition representative of full monolayer coverage on the surface of sputtered
aluminum the thickness was measured at 25.1 ?A, with a measured contact angle 110? and a
RMS roughness of ?3.7 nm.
Aluminum oxide spheres are coated following a similar procedure, but under continual
stirring, and subsequent rinse with 2-Propanol performed using vacuum filtration and vac-
uum collection. These samples are subsequently dried in an oven set at 60 ?C for 10 minutes
for solvent removal prior to analysis.
33
Chapter 4
Formation of Octadecyltrichlorosilane Monolayers on a Molecular Vapor Deposited (MVD)
Silica Layer
4.1 Introduction to the MVD silica layer
Previous studies have shown that the vapor phase hydrolysis of tetrachlorosilane results
in the deposition of a thin, uniform silica film with chemical properties similar to those of
conventionally prepared silica materials. This MVD silica layer is a highly hydroxylated
surface, but unlike conventional silica surfaces, it possesses no observable free surface silanol
groups, a fact believed to be due to the unusually high concentration of hydroxyl groups
present. Despite the lack of free silanol groups, the silica layer has been shown to support
the attachment of organosilicon compounds [55]. A novel property of the MVD silica surface
is the ability to create a reactive surface with a variable degree of surface roughness as well
as different topographical asperities (i.e. peak sizes, peak shapes). This tunable surface is
used to investigate the influence of surface topography on the interfacial characteristics of
subsequently deposited layers as illustrated in the following sections.
This silica layer can be deposited at low temperatures to a wide variety of substrates,
including glasses, metals, fibers, polymers, and plastics and can overcome the limitations
associated with traditional methods for depositing silicon oxide layers that requires tem-
peratures in excess of 300 ?C [89]. Prior work by Anderson et al. has demonstrated that
the vapor deposited thin silica film can be utilized as a novel substrate for in situ surface
studies on silica using infrared spectroscopy. Infrared spectra of silica films prepared by this
method compare favorably with spectra collected by the conventional thin film technique.
Furthermore, the uniformity of the thin vapor deposited silica film on the supporting infrared
window reduces the likelihood of portions of the film falling off during an experiment, an
34
improvement over the traditional thin film method. This method is the only experimental
technique put forth so far that can be used to study the vapor deposited silica films as they
exist on surfaces, and it can also be used to investigate reactions on the newly formed, pris-
tine silica layers. In fact, this material has been deposited on the surface of FTIR crystals
allowing for an investigation of the interfacial properties associated with the self assembly
of OTS monolayers, as evidenced, in the following sections. In addition, this silica layer is
not a line of sight deposition and has been deposited onto surfaces of varying geometrical
complexity, enabling the production of an advanced high surface area material utilized for
advanced analysis with TGA commonly limited to high surface area particles.
4.2 Utilization of MVD silica layer as a Novel Platform to Study OTS Mono-
layers
The role of free surface hydroxyl groups, hydration requirements, and the nature of
the surface attachment of OTS molecules on silicon oxide surfaces remains controversial
despite extensive investigation. Given the MVD silica layers reactivity toward organosilane
deposition, in spite of, the lack of free surface silanol groups, in conjunction with the ability
to examine the self assembly process and interfacial reactivity with FTIR, we were motivated
to study the growth characteristics of OTS monolayers as compared to those observed on
traditional silica surfaces.
In this chapter, we report the results of a comparative study of the adsorption of OTS
films on oxidized Si(100) surfaces representative of a traditional silica surface and a MVD sil-
ica layer. These surfaces are analyzed to determine the amount of physisorbed water present
under adsorption conditions in order to ascertain the influence of the water structural layer
on the OTS monolayer adsorption mechanism and overall film quality. This investigation
includes the effects of reaction temperature and immersion times on the organization of an
OTS monolayer, and includes investigation of both the submonolayer and full monolayer
35
growth regimes. The film properties and formation kinetics are examined using infrared
properties, ellipsometry, AFM, and contact angle measurements.
Experimental Details
Materials
Silicon wafers, p-type, with (100) orientation were obtained from University Wafer
(Boston, MA). Octadecyltrichlorosilane and tetrachlorosilane were obtained from Gelest,
Inc. (Morrisville, PA) and used as received. Concentrated hydrofluoric acid (HF, 49%),
acetone, isopropanol, and Optima grade hexanes (0.01% water) were obtained from Fisher
Scientific (Fair Lawn, NJ). Deionized water (18 M?-cm) was obtained from a Millipore
filtration system.
Methods
Cleaning methods, MVD silica layer formation, and OTS monolayer formation tech-
niques for the various samples have been previously described in Chapter 3.
4.2.1 Investigation of Water Layers on the MVD Silica Surface.
Initial efforts focused on investigating the interfacial region of the surface of silicon sub-
strates under adsorption conditions. The amount and structure of adsorbed water layers
associated with the MVD silica layer is compared to those found on traditional silicon oxide
surfaces. Asay and Kim have reported on the data analysis methods for studying the evolu-
tion of the adsorbed water layer thickness on silicon through ATR-FTIR spectroscopy [90],
and the same technique is employed in the present study. Briefly, the thickness of the ad-
sorbed water layer is determined by comparing the intensity of the water O-H bending mode
absorption located at 1635 cm?1 to the intensity of the same peak for a bulk liquid water
sample on the same surface under study, which is limited by the depth of penetration of
the evanescent wave. The thickness determined in this manner can be converted to the
36
number of monolayers by dividing by the mean van der Waals diameter of water, which is
2.82?A [90?92]. The position of the O-H bending mode located at 1635 cm?1 has been shown
as constant and is not influenced by the degree of hydrogen bonding or structure of water
present on the ATR surface [93].
The adsorption of water onto the MVD silica surface and the oxidized silica surface
was conducted by varying the relative flow rates of two nitrogen steams- one is dry and the
other is saturated with water vapor. This set-up was adapted by works done by Barnette et
al. [94] and allowed for investigation of water adsorption under varying relative humidities
ranging from 0% to 80%. All experiments were run at 27 ?0.5 ?C. A clean oxidized silicon
ATR crystal was placed into a teflon ATR holder, designed for this work as shown in Figure
4.1, and purged with nitrogen until no difference was noted in the spectra. This spectra
was utilized as the background spectra for examining the water adsorption on the oxidized
Si(100) surface. An atomic force micrograph for the oxidized Si(100) surface is illustrated in
Figure 4.2 with a corresponding ellipsometric oxide thickness of 2.0 nm and RMS roughness
value of 0.18 nm.
Figure 4.1: Custom built teflon ATR assembly sealed from ambient conditions with inlet
and outlet purge assemblies for controlling relative humidity.
An atomic force micrograph for the MVD silica layer is illustrated in Figure 4.2 B with a
corresponding average ellipsometric thickness of 18.0nm and RMS roughness value of 1.3nm.
FTIR analysis of the MVD silica layer after extensive nitrogen purging is represented in the
37
Figure 4.2: Two dimensional AFM scans of (A) a clean oxidized Si(100) surface, RMS
roughness is 0.14 nm, and (B) Silica layer deposited on clean silicon surface, RMS roughness
is 1.3 nm, thickness is 18.0 nm.
inset view on Figure 4.3. Major peaks present in the spectra include a broad feature centered
at 3400 cm?1 indicative of the OH stretching frequencies of associated silanols present in a
distribution of geometrical arrangements and chemical environments, while isolated silanols
known vibrate at 3747-3750 cm?1 are absent in the MVD silica films. This MVD silica
surface lacks the peak associated with physisorbed water normally located at 1635 cm?1
indicating a MVD silica layer void of physisorbed water. The ATR spectra for the MVD
silica film is utilized as the background spectra for experiments involving water absorption
on the MVD silica layer. Contact angle measurements indicated a CA = <5.0? indicative
of a highly hydroxylated surface on both the oxidized Si(100) surface and the MVD silica
surface.
After the background spectra were collected for the oxidized silicon ATR and the MVD
silica layer coated ATR, the films were exposed to increasing levels of relative humidity,
which resulted in the adsorption of physisorbed water as indicated by the broad feature at
3100 and the peak at 1635 cm?1. The pure water spectra collected for calculating maximum
adsorption intensity was scaled in magnitude to the 80 % RH spectra to visually compare
peak positioning for bulk liquid water applied to each crystal. Water structure is determined
by examination of the water O-H stretching region of the spectrum from 3700 to 2800 cm?1.
The peak position of the OH vibrational peak is very sensitive to the degree of hydrogen
38
Figure 4.3: ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on an
oxidized silicon ATR crystal at increasing relative humidities. The peak positions of ice-like
3119 cm?1 and the position of the O-H bending mode is marked at 1635 cm?1. The direc-
tion of increasing relative humidity is from bottom to top. Spectra correspond to relative
humidities of 20%, 40%, 60% and 80% with an accuracy of ?2.0%. [Inset View]-Background
spectra for water adsorption study illustrating O-H stretching region associated with hy-
drogen bonded surface silanols. Note peak absence at 1635 cm?1 indicative of physisorbed
water.
bonding and is often used to study the structure of water [95?97]. While liquid water
exhibits a broad vibrational peak at 3400 cm?1 condensed water at ambient conditions has
been shown to exhibit both liquid-like peak characteristics and ice-like peak characterisics
as indicated by two peaks centered near 3200 cm?1 and 3400 cm?1 corresponding to ice-like
and liquid water, respectively [98].
In the case of silicon oxide covered with silanol groups, the structure of adsorbed water
at room temperature as a function of relative humidity (RH) was discussed previously [93].
For comparison purposes this data is presented in Figure 4.4. Each line represents a different
partial pressure relative to saturation pressure of water (P/Psat) or relative humidity at 27?.
This study, indicated that the ice-like and liquid water isotherms grow together, and at
nearly an equal rate, across all relative humidities studied.
The results of water adsorption onto the surface of a MVD silica-modified silicon ATR
crystal are presented in Figure 4.3. In order to determine the thickness of bulk water that
39
Figure 4.4: ATR-FTIR spectra of adsorbed water on the surface of an oxidized silicon ATR
crystal at increasing relative humidities. Ice-like and liquid water are indicated with arrows
at 3250 cm?1 and 3400 cm?1, respectively, and the position of the O-H bending mode is
marked at 1635 cm?1. The direction of increasing relative humidity is from bottom to top.
Spectra correspond to relative humidities of 20%, 40%, 60% and 80% with an accuracy of
?2.0%
the evanescent wave interacts with, the thickness of the silica layer is subtracted from the
depth of penetration of the evanescent wave. For example, for the silicon/silica system,
the depth of penetration is approximately 505 nm at a 45? angle of incidence. In this
experiment, the silica layer is 18.0 nm thick. This means that the adsorption measured
for a bulk water sample on a silica film corresponds to 487.0 nm of water (505-18.0 nm).
For the MVD silica surface the depth of penetration is determined by considering the silica
layer as impervious to liquid water. Prior research performed by Anderson et al. has shown
this to be a valid means to determine the depth of penetration and more details can be
found in that work [93]. At 80% RH, the amount of adsorbed water is equivalent to nearly
20 monolayers, and is approximately 4.5 times greater than on the clean silicon surface, at
similar relative humidities. The structure of the adsorbed water on the MVD silica layer is
dominated almost entirely by ice-like water. Furthermore, the ice-like water structure has
an adsorption peak centered near 3119 cm?1, which is 127 cm?1 lower than on clean silicon
oxide. It was concluded that more absorbed water was on the surface of the MVD silica,
40
which was more tightly bound compared to that existing on traditional silica surfaces. This
was evidenced by a water absorbance spectrum dominated by an ice-like structure shifted
to lower energy. This shift to lower energy indicates that water has a stronger degree of
hydrogen bonding with the MVD silica-coated surface relative to the clean silicon oxide [99].
It is postulated that the MVD silica layer maintains a more ordered water structure due
to a higher degree of hydrogen bonding caused by the high concentration of hydroxyl groups
present. It is also suspected that the high degree of hydrogen bonding induces preferential
water orientation in subsequent layers contributing to the stability of the increased water
monolayers. Du et al. found that interfacial orientation and bonding structure of water
are strongly influenced by the electrostatic interaction and hydrogen bonding of molecules
with the surface at fused quartz/water interfaces [97]. Work by Yang et al. on water/silica
interfaces suggests that a more ordered water structure (ice-like) is associated with vicinal
silanol groups, due to a higher local surface electrical field on the silica [100]. As previously
described, the MVD silica layer is highly hydroxylated and maintains no free OH groups and
the surface consists of mainly vicinal groups. GunKo et al. provides further evidence toward
electrostatic influences showing that the hydrophobidization of silica surfaces (limiting the
availability of OH groups) modifies the state of water by lowering the electrostatic field [101].
The extent of the electrostatic influence on subsequent layers is still a subject of debate.
Ewing [102] argued that only the first interfacial layer of water is affected by the electric
field while the work by Shen [103] indicates that electric fields can induce polar ordering
of interfacial water molecules up to a few monolayers. Argyris et al. performed modeling
studies on the water structure on crystalline silica substrates indicating a correlation between
the local density of hydroxy groups and preferential orientation of the interfacial structure of
water, illustrating that water molecules in the first layer of a fully hydroxylated surface have
hydrogen down orientations, where as nonhydroxylated surfaces have water molecules with
OH bonds parallel to the surface [104]. Work done by Ostroverkhov utilizing Phase-Sensitive
Surface Spectroscopy on a quartz/water interface assigns ice like stretching regions existing
41
between 3050 and 3200 cm?1 with the peaks indicated at the low-frequency component,
usually labeled as the ice like peak, first appearing as a negative peak centered at 3050
cm?1 at low pH, but reverting to a dominant positive peak centered at 3200 cm?1 at high
pH. Suggesting that as pH increases, the quartz surface becomes more deprotonated (and
negatively charged with a resulting surface field), and thus indicating that water molecules
exist with their H bound to the surface [105]. All of these works are supportive of our
postulated interfacial orientation induced by an electrostatic interaction from the highly
hydroxylated silica surface.
A second explanation for the observed increase in hydrogen bonding and ice like ori-
entation of water molecules on the MVD silica surface could be attributed to the inherent
topography associated with the MVD silica layer. Recent studies suggest that water con-
fined in nanoscopic regions exhibit different thermodynamic characteristics in comparison
to bulk measurements [106]. Both Goertz and Li found that the viscosity of water for
nanometer scaled separations were approximately 106 times greater than that of bulk water
for highly hydrophyllic surfaces indicating a more structured water present in nanoscopic
regions [107,108]. Unfortunately, for the present analysis, it is difficult to elucidate answers
in terms of nanoscale confinement as it has been shown that ATR infrared spectroscopy have
large penetration lengths (>100 nm) that hinder the study of the buried interface [98]. As
discussed by Castrillon et al. when examining water in nanoscale confinement, a distinction
must be made between: (i) interfacial water, and (ii) confined water, as interfacial water
is commonly understood as being found between fluid phases (bulk vapor or liquids) and a
solid substrate, while confined water exists between two solid surfaces [109]. The confine-
ment length scales for the water existing between the solid surfaces as experienced in this
system is greatly over examined and the influence of nanoscopic regions is indistinguishable
from the interfacial water.
Results show that the MVD silica layer supports a higher amount of water on the surface
at increasing levels of relative humidity (approximately 4.5 times greater) than on traditional
42
silicon oxide surfaces. This increased amount of water is found to be more highly hydrogen
bound to the surface in comparison to the monolayers investigated on traditional silicon oxide
surface, suggesting that the MVD coated surface maintains a more ordered water structure
due to a stronger degree of hydrogen bonding compared to water layers formed on traditional
silicon oxide surfaces.
4.2.2 Influence of Formation Time and Temperature on OTS Monolayer For-
mation on a Highly Hydrated MVD Silica Surface
As evidenced in the previous section, the MVD silica layer supports a higher amount
of water on the surface at increasing levels of RH as compared to traditional silica surfaces.
For the results presented in this section, the relative humidity was held constant at ? 45
%, consistent with prior research notations found to produce high quality OTS films as
applicable to traditional silicon oxide surfaces.
In order to examine the OTS monolayer formation mechanism on the MVD silica surface
as compared to traditional oxidized silica surfaces, two regimes of OTS film growth are
examined. The first is a room temperature case representative of the mixed regime (LC and
LE phase) followed by a case at T = 10 ?C representative of the lower temperature regime T
< T0 (only LC phase) with film formation times ranging from 5.0 seconds to 1.0 hour. A 20
second OTS submonolayer deposition on the smooth Si(100) surface at room temperature is
shown in Figure 4.5B and the 10 ?C case is shown in Figure 4.5A. These figures clearly show
a distinction between the different deposition phases as influenced by the temperature. For
the reduced temperature case presented in Figure 4.5A only island growth is present with
no deposition in the surrounding areas consistent with previously reported results [2]. The
islands formed at reduced temperatures maintain average height values of roughly 2.0 nm as
shown in the height profile in Figure 4.5A. The room temperature case, as shown in Figure
4.5B, shows island growth surrounded by a secondary disordered phase. The height profile
of the room temperature case is illustrated in Figure 4.5B and is significantly lower than the
43
reduced temperature deposition with island average height values at roughly 1.4 nm. This
decreased height value is attributed to the presence of two phases on the surface made up
of a LC phase (ordered islands-light features) surrounded by a LE phase (disordered-dark
features) randomly distributed. Hence, the measured height values represent the difference in
thickness between the disordered and ordered island phases. Due to the roughness associated
with the MVD silica layer surface, AFM could not be utilized to examine the submonolayer
growth regime on the MVD silica surface. Generally, not enough contrast was visible between
the OTS film and the underlying substrate and the features remained hidden and undiscern
able from the underlying surface topography.
Figure 4.5: AFM micrographs of OTS monolayer formation (A) A 20 second OTS sub-
monolayer deposition on native silicon oxide at 10 ?C. (B) A 20 second OTS submonolayer
deposition on native silicon oxide at room temperature.
The values measured for the thickness of the OTS film as a function of immersion time
ranged from 6.0 to 28.3 ?A for both the room temperature case and the 10 ?C case on the
oxidized Si(100) surface. The values measured on the MVD silica layer ranged from 3.7 to
44
31.5 ?A. All data sets showed an increase in thickness as a function of immersion time. The
maximum thickness obtained on the oxidized Si(100) surface is 28.3 ?A, consistent with values
found in the literature [42] and close to the theoretical value of 26.2 ?A [65]. A refractive
index of 1.45 is used in calculating the OTS thickness, yet this value may not be reliable for
low coverage samples as partial OTS monolayer can have differences in the refractive index
that are dependent on the phase present [15]. Also, the standard deviation of a data set
of multiple ellipsometric measurements of the MVD silica layer thicknesses is on the order
of ?1.0 nm. Such a large standard deviation for the MVD silica layer thickness makes it
impossible to determine the true thicknesses for the films deposited onto the MVD silica
layer surface, since the precursor used in this study form films with thicknesses of roughly
three nanometers or less. Average values are presented in Figure 4.6 and should not be
considered conclusive values.
The contact angle measurements for all experimental data sets showed an increase with
respect to immersion time. At increased immersion times the values obtained for the 10 ?C
case on the oxidized Si(100) surface and the MVD silica layer were similar to those obtained
at room temperature regardless of the deposition conditions. At decreased immersion times
the contact angle for the room temperature cases was significantly lower on both with values
of 39.0? and 41.0?, in comparison to 63.8? and 69.8? at reduced temperatures. This data was
consistent with the differences in overall coverages as measured by ellipsometry and indicated
that the OTS films assembled more rapidly at the reduced temperature compared to the
room temperature conditions. All of the contact angle data regardless of the temperature
regime plateaus at higher deposition times at values greater than 100? and are indicative of
a complete monolayer.
Although AFM provides information on the microscopic structure of OTS film growth,
it does not reveal molecular structure and conformation changes within the film and can not
provide visual contrast on the MVD silica surface. To obtain structural information FTIR
analysis was performed on the films as a function of immersion time and temperature on the
45
Figure 4.6: Ellipsometric film thickness vs immersion time for octadecyltrichlorosilane films
on oxidized Si(100) and MVD silica layer at 10 ?C and room temperature. See text for film
formation conditions and details of extraction of thickness from ellipsometric parameters.
Experimental uncertainty of the thickness is estimated at ?2.0 ?A.
Figure 4.7: Water contact angle vs immersion time for octadecyltrichlorosilane films on
oxidized Si(100) and MVD silica layer at 10 ?C and room temperature.
46
oxidized Si(100) and the MVD silica surfaces. For FTIR analysis, the background spectra
for films deposited on the oxidized Si(100) surface, was a clean uncoated silicon ATR crystal
plasma treated to grow an oxide layer thickness 2.0 nm. For the films deposited on MVD
silica layer the background spectra was the as deposited MVDsilica layer. To characterize the
conformational order of the octadecylsiloxane monolayers, we use the methylene stretching
modes as a benchmark. On a native silicon oxide surface the frequencies of the methylene
antisymmetric and symmetric stretches are in the range of 2915-2920 and 2846-2850 cm?1,
respectively, for alkyl chains in an all-trans conformation with extended chains, and shift
toward 2928 and 2856 cm?1 when the chains are in a liquid like disordered phase [110].
Our data as demonstrated on an oxidized Si(100) surface and illustrated in Figure
4.8 is in agreement with work previously done by Lee et al. showing that at the lower
reaction temperature 10 ?C the intensity of the FTIR peaks in the vs(CH2) and va(CH2)
bands increase in proportion to immersion time [36]. At the reaction temperature of 10
?C the peak positions for the vs(CH2) and va(CH2) bands remain constant at 2850 and
2918 cm?1 regardless of immersion time. The OTS SAMs prepared at the higher reaction
temperature are notably different. In Figure 4.8 the intensity of the peaks also increases in
proportion to increasing the immersion time. But, there is a notable decrease in the wave
number of the methylene antisymmetric and symmetric stretches ranging from 2919 and
2852 cm?1 at the lower immersion times to values at higher immersion times of 2918 and
2850 cm?1. The resulting increase in intensity in both cases indicate that the density of
the OTS monolayer increases in proportion to immersion time. Additional information can
be obtained by examining the peak widths as a function of immersion time. For the room
temperature case the peak widths at lower immersion times are very broad becoming more
narrow at increased film formation times. Research provided by Parikh et al. explained
that this peak broadening at room temperature was indicative that the monolayers formed
consist of a mixture of conformationally ordered and disordered chains representative of both
a LE phase and a LC phase. Thus, indicating that for the room temperature case, as the
47
deposition times increase, the film becomes more dense and the chains become more ordered
as evidenced by the decreasing peak widths and shift in the antisymmetric and symmetric
bands to lower energy. For the reduced temperature case, the peak widths remain constant
regardless of immersion time indicating only an ordered LC phase present that increases in
density in proportion to immersion time.
The results presented in Figure 4.9 are the FTIR spectra for films formed on the MVD
silica layer as a function of immersion times and temperatures under study. These spectra
follow the same trends as identified on the native silicon oxide. The peak positioning and
increasing intensities are consistent with what was observed on the oxidized Si(100) surface
indicating that the growth mechanism as influenced by time and temperature are equivalent
mechanistically as measured with FTIR on the MVD silica layer.
48
4.2.3 Conclusions
A MVD silica layer has been demonstrated, amenable to FTIR analysis allowing for
an examination of interfacial properties in combination with an investigation of OTS film
formation mechanism. This MVD silica layer consists of a highly hydroxylated surface
that lacks free OH groups possessing a high affinity to physisorbed water resulting in an
approximately 4Xincrease inamount absorbedas afunction ofrelative humidity ascompared
to traditional silicon oxide surfaces. Results indicate that the MVD silica layer supports an
ordered water structure that is more tightly bound due to a higher degree of hydrogen
bonding associated with the hydroxylated surface. An examination of the growth process
of OTS films from submonolayer coverage to full monolayer deposition as influenced by
temperature is in agreement with films produced on native silicon oxide surfaces. In contrast
to previously reported results, the quality of the films obtained are comparable to those
formed on traditional silicon oxide surfaces, despite a higher amount of water present and
a lack of free OH groups. These results suggest that the role of free OH groups on the
formation mechanisms for OTS film formation are inconsequential, further highlighting the
importance of interfacial water on the OTS film formation process. We show that high
quality OTS films can be formed on a highly hydrated surface suggesting that the OTS
deposition mechanism is not only influenced by the amount of water present on the surface,
but may dependent how tightly bound the water is to the surface, indicating a link between
the interfacial waters? availability and reactivity. This work supports the use of an MVD
silica layer as a novel platform to study OTS monolayer formation, further enabling the
ability to probe the influence of surface topography and the interfacial water layer on the
self-assembly process.
49
Figure 4.8: Representative FTIR spectra of OTS monolayer formation as a function of
immersion time: (Top) OTS monolayer formation on oxidized Si(100) at 10 ?C and (Bottom)
OTS monolayer formation on a oxidized Si(100) at room temperature.
50
Figure 4.9: Representative FTIR spectra of OTS monolayer formation as a function of
immersion time: (Top) OTS monolayer formation on MVD silica layer at 10 ?C and (Bottom)
OTS monolayer formation on a MVD silica layer at room temperature.
51
4.3 High Surface area MVD Silica Support for TGA Analysis of OTS Mono-
layers
Introduction
While the OTS formation processes and bonding structure have been the subject of
extensive investigation, comparatively less is known about monolayer stability as influenced
by post processing steps. Monolayer stability has great technological importance because
the reliability of a device often depends critically on the integrity of its surface functional-
ity imparted by the coating. In addition, an understanding of the films thermal stability
is necessary for successful integration with MEMS devices in terms of processing and use
conditions. Prior understanding of the degradation mechanisms of OTS films has relied heav-
ily on techniques such as FTIR, contact angle analysis, X-ray reflectivity, and ellipsometry
which give an indication of the state of order of the alkane chains and structural changes as
influenced by temperature.
TGA analysis is another technique that can provide information about the monolayer
formation and degradation mechanisms, namely, the lateral surface uniformity as indicated
by overall surface coverage. TGA analysis has found limited use due to the inability to resolve
small changes in mass relative to the low film loading associated with planer substrates (i.e.
Si(100)). As aconsequence, highsurface areasupport materialsareoftenused toprovide high
loadings necessary for experimental resolution when examining thin films. High surface area
materials such as silica particles and mesoporous materials such as SBA-15 have proven ef-
fective at enabling a limited investigation of monolayer loading and degradation mechanisms.
Unfortunately, in the case of fumed silica, the available free volume for monolayer attachment
is affected by two factors, namely, the state of aggregation/agglomeration and the curvature
of the primary particles. Since fumed silica consists of aggregated/agglomerated particles,
only a limited portion of the surface area is actually accessible to OTS molecules [111]. The
curved surfaces of the silica particle offer additional constraints to monolayer formation with
52
the degree of curvature influencing the efficiency of space occupied by all-trans octadecyl
chains at full surface coverage [112]. In the case of mesoporous SBA-15, a high surface area
support has been demonstrated, that consists of both a curved surface in combination with
a large amount of pores. The size of the pores in the SBA-15 has been shown to influence
OTS molecule adsorption.
In this section, a high surface area support is demonstrated that has a lower degree of
curvature in comparison to the previously reported silica support materials that is nonporous
in nature. This high surface area material created through the deposition of MVD silica
layers onto cellulose templates is capable of sustaining OTS loading values measurable with
TGA analysis. In addition the properties of this support material enable the deposition of
monolayers with film characteristics more similar to those found on a planar silicon oxide
surfaces. It is suggested that this material provides a more valid substrate for comparing
thermal events as related to monolayer films with TGA analysis.
Materials
Silicon wafers, p-type, with (100) orientation were obtained from University Wafer
(Boston, MA). Cabosil M-5 (200 m2/g) fumed silica was obtained from Cabot Corporation.
Octadecyltrichlorosilane and tetrachlorosilane were obtained from Gelest, Inc. (Morrisville,
PA) and used as received. Acetone, isopropanol, and Optima grade hexanes (0.01% water)
were obtained from Fisher Scientific (Fair Lawn, NJ). Deionized water (18 M?-cm) was
obtained from a Millipore filtration system. Cellulose cotton supports were obtained from
Fisher Scientific.
4.3.1 Production of Fibrous MVD Silica Support
A templating technique was utilized to produce the fibrous silica support. The MVD
silica layer was deposited onto cellulose cotton balls. Briefly, the cotton balls were cleaned
with a 30 sec O2 RF plasma treatment explained in section 3.2.1 and followed with 10 cycles
53
of MVD silica layer deposition as described in section 3.2.1 (20 torr H2O and 20 torr of
SiCl4 per cycle). The thickness of the silica layer was approximated at 1 ?m as measured
by ellipsometry on a Si(100) reference wafer cleaned and oxidized as indicated in Chapter 3.
After removal from the vacuum system, the templated MVD silica was placed into a furnace
and heat treated to 700 ?C for 4 hours to remove the cellulose material, revealing the MVD
silica support. Prior analytical evaluations were performed with TGA purged under air
that determined this heat treatment procedure resulted in complete removal of the cellulose
leaving a thin white fibrous material. Thermal treatment in this manner as illustrated by
Anderson et al. on MVD silica surfaces, has shown that when the MVD silica surface
is treated at temperature > 400 ?C that that surface silanols in closest proximity to each
other, bridge bond into new Si-O-Si bonds lowering the reactivity of the silica surface toward
organosilane deposition. To counteract this effect the MVD silica support was reinserted
into the MVD deposition chamber and coated with an additional cycle of the MVD silica
in essence reactivating or reconditioning the surface with available surface silanols. The
available surface area of the silica support was measured at 60 m2/g with BET. A SEM image
is of the MVD silica surface after reconditioning is shown in Figure 4.10 revealing a fibrous
material with a low degree of curvature with an average width of 10.0 ?m approximately 4
times greater that the length of an extended OTS molecule. This material is considered to
be nonporous based on BET surface area results, in conjunction to, the calculations based
on uniform MVD silica coverage to the exposed surface area of the cotton fiber. These
calculations resulted in approximately 12 the measured value, as expected, the surface area
of the reactive material is dual sided in nature reflecting reactivity of the outer MVD silica
layer and the MVD silica layer that was originally attached to the cellulose fiber.
54
Figure 4.10: SEM image of reconditioned fibrous MVD silica support with average fiber
width of 10.0 ?m and surface area of 60 m2/g.
OTS deposition
A 1-hourroomtemperature deposition of OTS monolayers was performed on bothfumed
silica powder and MVD silica fibers (pre and post reactivation) following the procedure
illustrated in Section 3.2.1. Samples were analyzed immediately without the 1 hour set up
time previously described in section 3.2.1.
4.3.2 Analysis of OTS Monolayers on Fumed Silica and Fibrous MVD Silica
Supports without Post Treatment
Fumed Silica
Thermogravimetric analysis of fumed silica powder before and after OTS adsorption is
shown in Figure 4.11. The thermal analysis was carried out in the temperature range from
30 to 650 ?C, in nitrogen, at a heating rate of 10 ?C/min. The TGA pattern of fumed silica
powder shows a total weight loss of ? 1.0% over the temperature range from ambient to 650
?C. The initial weight loss up to a temperature of 120 ?C can be associated with the breakup
55
of the hydrogen bonded network, excess solvent removal and removal of adsorbed water from
silica. The second stage of weight loss from 120 to 650 ?C can be associated with desorption
of the adsorbed water, as well as, water generated by the condensation of free silanols.
A multi-step weight loss is observed in case of OTS monolayers on fumed silica. The
weight loss occurred in three distinct regions, from ambient to 120 ?C and 120 to 350 ?C and
350 to 650 ?C. The total weight loss observed is 2.5% in the entire temperature range from
ambient to 650 ?C. The weight loss in the first step is approximately the same as the weight
loss in case of bare fumed silica and is attributed to physisorbed water. In the temperature
range, 120 - 350 ?C, the weight loss of the OTS coated support was higher than that of
fumed silica, and was attributed to Si-O-Si cross linking events and the release of associated
water, although, it is difficult to determine if this loss might also represent the start of
decomposition. We attribute the decomposition of the OTS layer to the weight loss in the
temperature range 350 to 650 ?C as has been reported for a monolayer thermally treated
in nitrogen. To obtain an estimate of the extent of coverage of OTS on the surface, 100 %
coverage was calculated assuming a maximum possible value of 5 molecules/nm2, and using
200 m2/g as the surface area of fumed silica. Therefore, the percent coverage for fumed silica
was 8.4%, and much lower than the amount calculated, suggesting an incomplete monolayer
and non uniform coverage. For the temperature range under investigation the silica molecules
are assumed to still remain on the surface. Low loadings are typical of silica materials, as
illustrated in Figure 4.12, due to the aggregated/agglomerated particles, which allow only
a limited portion of the surface area to be accessible to OTS molecules. An additional
consideration for the observed loadings stem from the assumption of an ideal monolayer
consistent with an all trans packing density (5.0 molecules/nm2). These films were produced
at room temperature and previous investigations have demonstrated disorder will exist in
the films. In assuming no diffusion limitations due to aggregated particles and completely
accessible surface area, the density of an ideal all trans OTS monolayer is estimated at 1.1
molecules/nm2 and suggests more tilt with the OTS molecules indicative of disorder.
56
Figure 4.11: TGA plots of fumed silica (top) before and (bottom) after adsorption of OTS.
Figure 4.12: Illustration of fumed silica (Cabosil M5) from manufacturer?s website with a
BET surface area of 200m2/g.
57
Fibrous MVD Silica Support
The weight loss events observed on the fibrous MVD silica and the OTS coated fibrous
MVD silica (conditioned) were similar to those noted on the fumed silica. As a reference,
the unconditioned MVD silica was included in the analysis showing that as expected no
deposition of OTS occurred given the decreased functionality associated with the formation
process. In general, the MVD silica blank followed the same trends as observed for the fumed
silica and the observed weight loss temperatures were comparable. One exception found on
the OTS coated MVD silica fiber is a larger weight loss event occurring between ambient
temperatures and 120 ?C. We attribute this to the fact that no post treatment was performed
on the OTS modified MVD silica surface and residual solvent from the coating process likely
remained. Generally, these supports were gelatinous after rinsing in solution and it seemed
to take longer for the solvent to completely evaporate compared to the fumed silica. Weight
loss in the second step ranging from 120 to 350 ?C was larger than that of the blank MVD
silica and was attributed to Si-O-Si cross linking events and the release of water associated
and/or the beginning of thermal decomposition. The weight loss in the range of 350 to
650 ?C was attributed to the decomposition of OTS. The same procedure, as previously
described, was used to estimate the surface coverage associated with the OTS film. For the
temperature range under investigation the silica molecules are assumed to still remain on
the surface. The film coverage on the MVD silica fiber was calculated at 51.0% from the
loss associated in the range of 350 to 650 ?C. This value was significantly higher than what
was obtained on the fumed silica surface under identical deposition conditions suggesting
a 6X increase in material present. This increase in material is significant considering that
the available surface area for the fibrous silica was only 60 m2/g as compared to 200 m2/g
for the fumed silica. Performing a similar calculation to that performed on the fumed silica
the density of an ideal all trans OTS monolayer on the surface of the MVD fibrous silica is
estimated at 2.1 molecules/nm2.
58
A higher calculated weight loss in the second step ranging from 120 to 350 ?C, in
combination with the higher overall loadings, were suggestive that the weight loss events
in this range can be attributed to the release of water from Si-O-Si cross linking events
and not the beginning of thermal decomposition. From this observation a need for post
processing of the OTS layers is indicated, such that, thermal decomposition can be observed
without interference from possible cross linking events. The section that follows supports this
conclusion and includes an analysis of the thermal decomposition mechanism as influenced
by different post processing environments.
It is more feasible that the accessible surface area for the fibrous silica is greater as
compared to the fumed silica verses more molecules per unit area as the depositions were
performed under similar deposition conditions. The MVD silica when deposited on a planer
substrate, as shown in the previous section, exhibits a similar reactivity toward OTS film
formation as compared to traditional silica materials. In conclusion, results suggest that due
to the lack of porosity, a low degree of surface curvature, and the increased loadings obtained,
that the fibrous silica, provides a more representative support material for the determination
of loading information for comparison to the films produced on planer substrates.
59
Figure 4.13: TGA analysis of blank fibrous MVD silica and OTS monolayer films on fibrous
MVD silica fibers (with and without reconditioning) without post treatment.
4.3.3 Analysis of OTS films on Fumed Silica and Fibrous MVD Silica after
Thermally Annealing with Different Gases
It has been suggested that heating OTS films to temperatures between 120 and 150 ?C,
results in the stabilization of the monolayer, through the formation of cross-linking between
adsorbed OTS molecules. [44,113]. Sung et al. has indicated a timescale for cross linking
or grafting to the substrate showing that in solution the (hydrolyzed) precursor molecules
deposited at 10 ?C retain their mobility for sufficiently long times indicating that the lifetime
of the mobile state is 10 minutes [40]. The results presented by Wang et al. suggest that
thermal annealing creates an adsorbed monolayer that is no longer mobile, either because
of attachment to the underlying substrate, to other OTS molecules, or due to removal of
adsorbed water [112]. Conditions necessary for appropriate post processing and the influence
of curing conditions on the stability of an OTS monolayer remain as undefined parameters. It
60
is the intent of this work to utilize TGA-MS to examine the influence of different annealing
gases on the stability and degradation scheme of OTS Monolayers produced on an MVD
silica surface.
Experimental
OTS Monolayers are deposited onto the MVD silica support following the same method-
ology as described in the previous section. After deposition the monolayer/support system is
placed into the TGA under different environments (Air, Nitrogen, and Helium) and ramped
to 120 ?C and thermally annealed at this temperature for a 1 hour duration. Typical loading
amounts are 10 mg of the OTS / MVD fiber support. After annealing, the gas is changed to
nitrogen and the system is purged of the initial curing gas. The temperature is subsequently
ramped at 10 ?C to 600 ?C and the thermal weight loss events are monitored.
Results
A distinction between the different annealing gases and decomposition features is appar-
ent from the TGA analysis. Two distinct weight loss events are evident from the thermally
annealed OTS film, most evident, in the differential plots shown in Figure 4.15. The first
event is in the range of 150 ?C to 400 ?C and second from 400 ?C to 600 ?C. The weight
loss data for the helium and nitrogen treatment are comparable, in the range from 150 ?C
to 400 ?C, but the profile obtained for air annealing is different. The decomposition of OTS
monolayers, on the MVD silica thermally annealed under air starts at a lower temperature
(250 ?C) compared to those treated under helium and nitrogen (370 ?C). It has been shown
that thermal decomposition occurs at lower temperatures in an air environment, as such air
may be trapped in the film from the annealing process contributing to the increased ther-
mal degradation at lower temperatures. The data indicate that the rate of decomposition is
slightly faster for the helium annealed samples in the 400-600 ?C regime. Based on the overall
weight loss of the OTS monolayer an estimated OTS coverage ranging from to 51 to 38% for
61
the three cases is determined. It is not suspected that the annealing procedure influences the
overall coverage attained on the MVD silica (i.e. lower on He and N2). It is suspected that
thermally annealing the OTS monolayer under a nitrogen and helium environment produces
an OTS film with a higher thermal stability in the range of 150 to 600 ?C, but that treatment
to a higher temperature would eventually result in the same overall weight loss and loading.
It appears that the OTS films cured under air undergo thermal decomposition at a faster
rate at lower temperatures and result in the most significant weight loss. But, in the case of
the nitrogen cured sample more fragments will remain up to 600 ?C.
Figure 4.14: Weight change as a function of thermal treatment under nitrogen from 150
?C to 600 ?C at a ramp rate 10 ?C/min for three OTS monolayers on fibrous MVD silica
previously annealed for 1 hour at 120 ?C three different environments (He, N2 and Air)
62
Figure 4.15: Derivative weight loss in (%/min) as a function of thermal treatment under
nitrogen from 150 ?C to 600 ?C with a ramp rate 10 ?C/min for three OTS monolayers
annealed for 1 hour at 120 ?C under three different environments (He, N2 and Air).
63
4.3.4 Conclusions
A novel high surface area silica support capable of sustaining loadings measurable with
TGA analysis is introduced. This support exhibits a similar reactivity toward OTS film
decomposition, as compared to, other high surface area silica materials and sustains higher
loadings. In addition, results suggest that due to the lack of porosity, low degree of surface
curvature, and the increased loadings obtained, that the fibrous silica, provides a more
representative support material for the determination of loading information for comparison
to the films produced on planer substrates. This support has been utilized to determine the
influence of annealing environments on the thermal stability of OTS films. It was shown
that films annealed under an air environment decomposed at lower starting temperatures
and to a greater extent than those annealed under inert atmospheres. This support will
enable future examinations of the influence of different annealing procedures and annealing
environments.
64
Chapter 5
OTS Island Formation on Topographical Asperities: Implications Toward Tribological
Response
5.1 Motivation
To date, the interfacial phenomenon responsible for causing stiction, friction, and wear
hinder the commericialization of MicroElectroMechanical Systems (MEMS) [50,114?117].
Consequently, substantial research has focused on ways to overcome the interfacial phe-
nomenon by controlling the topography and the chemical composition of the contacting sur-
faces [6,118]. The most common method for overcoming the interfacial interactions focuses
on tailoring the surface properties with self-assembled monolayers (SAMs) [11]. Octadecyl-
trichlorosilane (OTS) monolayers have recieved considerable research and attention [2,19?22]
with demonstrated success for tribological improvement [6,33,68]. Unfortunately, variation
in the amount of tribological improvement afforded by these films has been identified as
the films have been translated from smooth surfaces to MEMS surfaces [68,119,120]. Some
of the observed variation has been explained in terms of differences in topographical con-
tact points, associated tribological mechanisms and the length scale of the measurement
technique [121?123]. But, reported variation exists between films produced on similar sur-
faces regardless of topographical characteristics as analyzed with the same measurement
technique [124?127].
OTS monolayer growth kinetics and interfacial properties as deposited on MEMS sur-
faces are generally accepted from investigations of film formation on smooth surfaces. These
investigations have revealed distinct regimes of growth dependant on formation temperature
that are differentiated by changes in packing density, chain conformation, and orientation
with respect to the substrate [45]. Despite the revelation of three distinct growth regimes,
65
most published literature investigating OTS monolayers for friction and adhesion improve-
ment have relied heavily on OTS SAMs produced at room temperature (mixed regime). Yet,
recent works indicate differences in friction improvements attainable based on the phase of
OTS present. Work done by Flater et al. illustrated a disparity in friction measurements on
monolayers deposited from the mixed regime on smooth silicon surfaces. The monolayers,
which were deposited at room temperature, consisted of a mixture of the liquid condensed
(LC) phase represented by densely packed islands and a surrounding disordered liquid ex-
panded phase (LE) [72]. These results showed that OTS islands resulting from a LC phase
produced a lower friction value when compared to the surrounding disordered LE phase.
It was postulated that gauche defects in the chains or vacancies in the LE phase result in
an increase in friction. A congruent study by Masuko et al. showed that the friction re-
ducing performance for an OTS monolayer on a smooth silicon wafer corresponded to the
SAMs molecular density and the degree of molecular orientation [128]. The importance of
the deposition temperature on OTS film formation is evident with different friction charac-
teristics obtainable based on the phase present. It is suspected that some of the reported
variability in friction measurements as observed on similar surfaces can be explained with
this observation.
When OTSmonolayers aredeposited ontoMEMS surfaces, itbecomes equally important
to determine the interfacial properties, namely, molecular arrangement and packing density
on the topographical points of contact (asperities). Research indicates that the packing
density of monolayers not only depend on deposition temperatures, but that the adhesion
between SAMs and silicon oxide surfaces can vary significantly when assembly takes place
on surfaces with curvature as compared with flat surfaces [129,130]. Molecular dynamics
simulations of SAMs under compression have shown, during the initial stages of compression
by an asperity, that without strong lateral chain-chain interactions that gauche defects within
a monolayer appear and begin to propagate, reducing overall order within the film [131]. The
loosely packed nature of a disordered film increases the frictional forces. Thus, understanding
66
the relationship between monolayer order on the contacting asperities may help predict their
ability to act as lubricant films for MEMS.
An example AFM image for polysilicon is shown in Figure 5.1 and illustrates the to-
pographical complexity associated with MEMS devices. A line scan shows the inherent
roughness of these surfaces and gives a better representation for the surface features in
contact during friction measurements.
Figure 5.1: Example polysilicon surface and line profile of surface roughness
67
Unfortunately, limitations exist in the ability to identify what phases are present on
the topographical features, due to an inability to inspect the exact location of the mono-
layer coating. Atomic Force Microscopy (AFM) is a valued technique for topographical
investigation and has been successful in monitoring the growth mechanisms and coating
quality on smooth surfaces [2,21,30,132,133]. But with added topographical roughness,
AFM, can not be readily utilized to examine the submonolayer to full monolayer growth
process on technologically relevant substrates, such as polycrystalline silicon. Generally,
the monolayer characteristics remain hidden and undiscernable from the underlying surface
roughness. Consequently, alternative techniques are employed to obtain information about
the OTS film quality on surfaces with inherent roughness and/or curved surfaces. Techniques
such as FTIR [44,134], micro-gravimetric analysis [135] and TGA [29,112] have provided
structural information such as lateral packing density (loading) and orientation. However,
the data obtained is only indicative of the average or bulk values, and rely on certain assump-
tions about packing density for calculations. They can not reveal details about how the films
exist on individual asperities. Recently, Bush et al. showed that Kelvin Probe Microscopy
could be utilized to study OTS film formation on polysilicon surfaces. Their work focused on
obtaining contrast between the LC and LE phase on polysilicon surfaces. Results suggested
that partial OTS films formed at low temperatures preferentially deposited along the grain
boundaries of polysilicon, leaving the topographical features uncoated [45]. The reason for
this prefential absorbtion was not completely investigated and warrants further research.
In this investigation, OTS films deposited from the liquid condensed phase, are applied
onto surfaces of varying topographical complexity to examine molecular assembly as influ-
enced by surface asperities. A MVD silica surface, amenable to FTIR analysis, known to
support the same growth mechanisms of OTS monolayers [136] as compared to traditional
silicon surfaces is utilized. This surface enables a tunable degree of surface roughness with
additional control over different topographical asperities (i.e. peak sizes, peak shapes). In
order to provide topographic contrast for AFM imaging, an etchant technique is developed,
68
which as described, removes only the MVD deposited silica that is not ?protected? by densely
packed OTS islands. The film characteristics on the MVD silica surfaces are compared to
films produced on an oxidized Si(100) surfaces as examined via ellipsometry, contact angle
analysis, atomic force microscopy (AFM), and FTIR.
Experimental Details
Materials
Silicon wafers, p-type, with (100) orientation were obtained from University Wafer
(Boston, MA). Octadecyltrichlorosilane and tetrachlorosilane were obtained from Gelest,
Inc. (Morrisville, PA) and used undistilled. Concentrated hydrofluoric acid (HF, 49%),
acetone, isopropanol, and Optima grade hexanes (0.01% water) were obtained from Fisher
Scientific (Fair Lawn, NJ). Deionized water (18 M?-cm) was obtained from a Millipore
filtration system.
Methods
Cleaning methods, MVD silica layer formation, and OTS monolayer formation tech-
niques for the various samples have been previously described in Chapter 3.
Etching Process
A dilute solution of HF is utilized for the removal of exposed silica. This solution is
made from a 1:700 v/v HF:Deionized water. Each sample was allowed to etch for 1 minute
and removed from the solution and then rinsed with copious amounts of Deionized water.
The samples are then dried under a stream of nitrogen. More details are provided in the
sections that follow.
69
5.2 Introduction of an Etchant Procedure for Obtaining Topographical Con-
trast with AFM Imaging
In order to provide topographic contrast for AFM imaging, a unique etchant technique
was developed and utilized, which removes only the MVD deposited silica that is not ?pro-
tected? by densely packed OTS islands. This work was prompted by studies done by Komeda
et al. [137] using a similar methodology to reveal properties of the SiO2 near the SiO2/Si
interface.
Our study utilized, ATR-FTIR and AFM analysis to examine OTS film characteristics
as influenced by exposure to a dilute HF etchant. This investigation included examination
of both submonolayer films and full monolayers deposited at room temperature deposition
(mixed regime LC and LE phases) conditions and those deposited at reduced temperatures
T=10 ?C (LC phase). A comparison was made to films deposited on MVD silica surfaces and
traditional oxidized (Si(100) surfaces). Etch rate was determined for the various MVD silica
morphologies, indicating that an etch time of 1 minute was sufficient to remove exposed
MVD silica providing contrast for OTS film investigation with AFM. It was determined
that the dilute HF etching technique had no influence on a full monolayer film regardless of
formation temperature, but distinct differences were observable in the submonolayer growth
regime.
Two cases are presented to illustrate the use of the etching procedure. The first case
consists of two experiments examining a 20 second submonolayer OTS deposition at room
temperature (mixed regime) on a bare oxidized silicon ATR and on a MVD silica layer coated
oxidized silicon ATR. The bare oxidized silicon surface had a 0.2 nm RMS roughness and
oxide thickness of 1.9 nm and the an AFM images of this surface is illustrated in Figure
5.3A. A MVD silica layer was deposited onto a second oxidized silicon ATR and is referred
to as MVD silica 0 with an oxide thickness of 30.0 nm and a 1.0 nm RMS roughness. An
AFM image of the MVD silica layer is shown in Figure 5.4A. The second case consists of two
experiments examining a 20 second submonolayer OTS film deposition at 10 ?C(LC regime)
70
on a similar oxidized silicon ATR surface and a second ATR that was coated with same
MVD silica layer previously detailed.
For the FTIR analysis, the background spectra for the OTS Island coated MVD silica
layer was the ?as deposited? MVD silica layer and the background spectra for the silicon
oxide layer was a clean plasma oxidized silicon ATR crystal with 2.0 nm of oxide. To char-
acterize the conformational order of the octadecylsiloxane monolayers, we use the methylene
stretching modes as a benchmark. On a native silicon oxide surface the frequencies of the
methylene antisymmetric and symmetric stretches are in the range of 2915-2920 and 2846-
2850 cm?1, respectively, for alkyl chans in all-trans conformation with extended chains and
shift toward 2928 and 2856 cm?1, repectively when the chains are in a liquid like disordered
phase [110].
We have previously demonstrated that the growth process of octadecyltriclorosilane
(OTS) films on MVD silica layers from submonolayer coverage to full monolayer deposi-
tion are consistant with the growth mechanisms found on oxidized silicon surfaces [136].
Briefly, this work showed that when OTS films are deposited at reduced temperatures, the
molecules densely assemble at the early stage of formation and grow two dimensionally with
the same packing density representative of an all trans conformation and as the immersion
time increased there was a monotonous increase in coverage with a continued all trans con-
formation [36]. The OTS films prepared at room temperature are notably different. The
intensity of the peaks also increases in proportion to increasing the immersion time indicating
a density increase. But, there is a decrease in the wave number of the methylene antisym-
metric and symmetric stretches ranging from 2919 and 2852 cm?1 at the lower immersion
times to values at higher immersion times of 2918 and 2850 cm?1 indicating the formation
of a more ordered film as film growth progresses. Additional information was obtained by
examining the peak widths as a function of immersion time. For the reduced temperature
case, the peak widths remain relatively constant regardless of immersion time. But, for the
71
room temperature case the peak widths at lower immersion times are very broad and be-
come more defined as the deposition time increases. This increase in peak width for the room
temperature case compared to that of the 10 ?C case was indicative that the monolayers pro-
duced at room temperature exhibit both a LE phase and LC phase simultaniously evidenced
by conformationally ordered and disordered chains. The same trends were observed for the
cases presented here.
The FTIR spectra shown in Figure 5.5 illustrates the influence of the etchant on OTS
submonolayer films by examining the methylene antisymmetric and symmetric peaks before
and after the contrast enhancing etchant procedure. For the 20 second room temperature
case on the oxidized silicon ATR crystal the methylene antisymmetric and symmetric peaks
are very broad and shifted to a higher wavenumber indicative of a more disordered film
exhibiting characteristics of both a LE phase and LC phase. The same trend is evident for
films grown on the MVD silica layer, but the broadening effect is greater suggesting that
a higher degree of disorder exists on this film. It is suspected that this higher degree in
disorder is associated with the increased roughness associated with the MVD silica layer in
this study. After the etchant treatment, the methylene antisymmetric and symmetric peaks
shift to 2918 cm?1 and 2850 cm?1 and the broadening effect is diminished, which is more
in agreement with alkyl chans in all-trans conformation. These results further suggest that
the LE phase was removed with only the LC phase remaining as indicated by peak location.
An AFM image of the OTS film after the etchant treatment is presented in Figure 5.4B,
consisting of nothing but the remaining LC phase. Prior to the etchant treatment the AFM
image was uninformative and did not allow for visualization of the OTS deposition.
Figure 5.6 shows the results of a 20 second 10 ?C OTS film deposition on both the
oxidized silicon and the MVD silica layer before and after the etchant treatment. In this
instance the methylene antisymmetric and symmetric peaks remain constant regardless of
the etchant treatment on both surfaces. This result indicates that the etchant procedure
does not remove densely packed OTS Islands. An AFM image of the 10 ?C deposition case
72
on the MVD silica layer is presented in Figure 5.4C. After the etchant treatment, the OTS
coated MVD silica layer consists of densely packed islands similar to those found on a smooth
silicon surface under the same deposition conditions.
Our work has therefore confirmed that after dilute HF etching, the OTS films deposited
at reduced temperatures on both the native oxide surfaces and MVD silica layers remain
unchanged. But, the films produced in the mixed regime were found to be susceptable to
selective removal of the disordered phase. It was concluded that the packing density for
OTS films contributed to the differences. It is also noted that the MVD silica layer has
a higher degree of surface roughness in comparison to Si(100) surfaces and the degree of
disorder is increased on this surface under room temperature conditions, but for the reduced
temperature case no change in monolayer order is observed after etching. Utilizing the
dilute etchant procedure allows for an increase in obtainable contrast for AFM imaging by
removing MVD silica layers not protected by a densly packed OTS phase. This technique was
utilized for the remainder of this work focusing on submonolayer films produced at reduced
temperatures providing topographic contrast for AFM imaging.
5.3 Reduced Temperature Deposition of OTS Films as Influenced by Surface
Topography
In this work, densly packed octadecyltrichlorosilane islands resulting from low temper-
ature deposition conditions, are deposited onto native silicon oxide surfaces and onto silicon
oxide surfaces created via molecular vapor deposition (MVD). A two-dimensional AFM scan
of a clean silicon sample after the pre-deposition oxygen plasma treatment can be seen in
Figure 5.3A. The cleaning procedure detailed in Section 3.2.1 produces a clean, flat silicon
oxide surfaces, as evidenced in the figure. Various MVD silica layers exhibiting different
topographies were produced for this study. The procedure described in Section 3.2.1 pro-
duces a thin, rough, silica-like surface which is very rich in reactive sites for further surface
modification by organosilane chemistries [138]. After silica layer deposition, samples are
73
characterized with ellipsometry, contact angle analysis, and AFM. For simplicity, the vapor
grown silica layers examined in this section of work are labeled as MVD Silica 1 - MVD
Silica 5 and the native oxide surface is identified as Si(100) oxide.
Relevant dataforfilmsdeposited ontheclean silicon oxide surface andforfilms deposited
on the MVD silica layer surface are given in Table 5.1. In this work the MVD silica layer
surfaces range from 10-35 nm thick with RMS roughness values ranging from 1.0-6.0 nm.
Due to the non-uniform nature of the silica layer surface, the standard deviation of a data set
of multiple ellipsometric measurements of silica layer thicknesses is on the order of ?1.0 nm.
Such a large standard deviation for the MVD silica layer thickness makes it impossible to
determine the true thicknesses for the films deposited onto the MVD silica layer surface, since
the precursor used in this study form films with thicknesses of roughly three nanometers or
less. As is evidenced in the Table 5.1, all silica layers exhibit a contact angle of < 5? indicative
of a hydrophillic surface capable of sustaining OTS monolayer growth.
It is interesting to note that AFM measurements performed on different morphological
silica layers often yeilded the same RMS roughness values. This is demonstrated in some
of the examples shown in this work. In order to illustrate the topographical differences
in a more pronounced manner the AFM images were transformed into three dimensional
representations on a slightly magnified scale. This is illustrated in Figures 5.7A-E. The
scaling of these images were arbitrarily set in order to give an easier comparison of the
data. Rendering the images in this fashion allows for an easier recognition of the distinct
differences in topographical asperities, that would remain unnoticed, if only RMS roughness
values were compared.
After initial characterization the silica layers were modified with a submonolayer OTS
films deposited from the liquid-condensed phase (10 ?C for 20 seconds) as presented in
Section 3.2.1. The atomic force micrograph shown in Figure 5.3B, shows the OTS islands
obtained on a native oxide Si(100) surface. Island growth was allowed to proceed until islands
formed at an intermediate size of 1 to 2 ?m. It has been shown that additional ordering of
74
SAM molecules occurs in islands in this size range due to a more complete cross-linking or
covalent bonding to the substrate, which eliminates rotational defects or other confirmation
or changes [45].
As shown in Figures 5.8A-E, partial OTS monolayers grown from the liquid condensed
phase on the MVD silica layers, are readily observed after the etching process with AFM
imaging. It is evidenced from the AFM images that the OTS deposition is different on all the
surfaces illustrated. At low roughness values, illustrated in Figure 5.8A, the islands appear
to form unrestricted and similar to those observed on the native oxide surface as shown
in Figure 5.3B. As the MVD silica roughness values increase, in Figure 5.8B-E, islands
preferentially deposit in the lower portions of the topography (valleys) with the highest
points (asperity peaks) being selectively removed during the etching process. This removal,
as previously described, is attributed to the presence of a low density disordered film present
on the highest points.
In addition, the MVD silica layers illustrated in Figures 5.8A-E maintain observable
differences in the shape of the topographical asperties. For example, the MVD silica layers
identified in Figure 5.7C and 5.7D have similar RMS roughness values as measured with
AFM. The topographical asperities associated with the silica layer as represented in Figure
5.7C have a sharper topographical peaks in comparison to those in Figure 5.7D that appear
more rounded. Increased OTS film coverage is observed on the rounded topographical peaks
as shown in Figure 5.8D than on the sharper peaks shown in Figure 5.8C. The same trend
is observed when comparing the sharp peaks associated with Figure 5.7E in comparison to
those on Figure 5.7D with island adsorption location shown in Figure 5.8D-E. This indicates
that OTS film deposition is influenced by the asperity curvature. As a result more of the
densly packed film is able to form on the topographical asperities that maintained a more
gradual rise in the slope of valley to peak. Thus, the higher the peak curvature the less
likely densly packed islands form suggesting a correlation between the shape and size of the
asperity.
75
As illustrated inFigure5.9, itis hypothesized thattheasperity curvatureinhibit effective
chain-chain hydrophobic interactions, which ultimately stabilize the islands. As a result, a
decrease in the packing density allows the etchant to penetrate between the OTS chains and
attack the exposed silica layer, suggesting that little to no deposition of the high density
OTS phase occurred on the highest points of the topographical asperities of high curvature.
A 2-dimensional representation of this hypothesized effect is illustrated in Figure 5.9. It is
concluded, that the OTS chains will have a higher crosslinking probability with asperities of
lower curvature.
A similar conclusion was observed by the work of Wang et al. in analyzing OTS mono-
layer adsorption on the surface of spherical particles. This work derived a means to estimate
the efficiency of space occupied by ?ideal? all-trans octadecyl chains at full surface coverage
as influenced by surface curvature [112] for spherical powder analysis.
The packing efficiency, with which the molecules occupy the available volume is defined
as V/VS. If it is assumed that the octadecyl chain is in an all-trans rigid rod (comparable
to OTS molecules) then the length can be assumed at 2.0 nm and cross section of 0.2nm2,
then the number of OTS molecules, N, which can be accommodated on a d-nm spherical
particle, is pid2/0.20 and the total volume V, occupied by the N molecules, is 2pid2. The
volume available for occupation by the OTS molecules, VS, is pi[(d+ 4)3 ?d3]/6 so V/VS is
calculated according to the following equation (taken from ref.( [112]):
V
Vs =
12d2
[(d+4)3 ?d3] (5.1)
This model was used to explain how, the curvature of the silica particles creates free
volume between the chains, and how the resultant differences in free volume can thus affect
the packing of the OTS alkyl chains, finding that in the case of the 7nm particles, only 60%
of the available volume is in fact occupied by the OTS molecules, whereas 90% and 96% is
occupied for the 40nm, and 106nm beads, respectively.
76
If the topography of the surface is considered to be made up of spherical particles
of varing radius of curvature (hemispherical in this case) then a limiting value of 400 nm is
necessary for 99% coverage with much higher values necessary for 100% coverage. Obviously,
this is asimplistic explanation and cannot bequantitatively determined frommeasured AFM
values and would prove even more difficult to approximate on a polysilicon surface with a
multitude of asperity dimensions. The MVD surfaces represented here show differences in
the saggital curvature and transverse curvature of the surface asperities and limit the validity
of surface curvature models for determining optimal packing density.
In addition to the preferential adsorption described, the films evidenced increasing ex-
tents of OTS island coverage under the same adsorption conditions corresponding with in-
creasing MVD silica layer thickness. This observation is suggestive of an increase in surface
reactivity associated with thicker MVD silica coverages. This effect is presently under in-
vestigation and will be the subject of future reports focusing on the interfacial water layer
found on the MVD silica layer under adsorption conditions and its possible contribution to
this observation.
5.4 Conclusions
In this work Atomic Force Microscopy was utilized to invesitgate the formation of OTS
films on surfaces exhibiting various topographical asperities. A dilute etchant technique is
introduced, providing topographic contrast, for AFM imaging by removing only the MVD
deposited silica that is not ?protected? by densely packed OTS islands. Preferential film
formation is observed in the lower regions of the MVD silica surfaces with the highest points
(asperity peaks) being selectively removed during the etchant process. This removal is at-
tributed to the presence of a low density disordered film present on the highest points. The
extent of coverage to the surface asperities is suspected to depend on the curvature associ-
ated with the asperities. With increasing asperity curvature (sharpness) effective chain-chain
hydrophobic interactions are reduced resulting in a decrease in packing density allowing the
77
etchant to penetrate between the OTS chains and removing the exposed MVD silica layer.
From these results, it is determined that the LC phase do not deposit uniformly on the asper-
ities. Based on this conclusion and due to published literature investigating OTS monolayers
for tribological improvement on MEMS devices, it is suspected that the LC phase will de-
posit differently on every surface investigated and be dependant on the overall profile of the
asperities. It is postulated that this is a contributing factor for why friction measurements
vary from group to group and sample to sample on MEMS devices. The results warrant fu-
ture consideration when trying to utilize OTS monolayers, regardless of temperature regime,
on rough surfaces.
78
MVD Silicon Oxide OTS Island Layer
CA Thickness (?A) RMS (nm) CA OTS Thickness (?A)
Si(100)oxide <5? 18.6 0.17 85.4? 26.2
MVD SilicaLayer1 <5? 119 1.55 92.3? 26.4
MVD SilicaLayer2 <5? 175 1.89 92.2? 28.5
MVD SilicaLayer3 <5? 211 3.96 94.0? 25.4
MVD SilicaLayer4 <5? 255 3.97 88.3? 24.4
MVD SilicaLayer5 <5? 316 6.30 91.4? 27.3
Table 5.1: Water contact angle (CA), thickness, and RMS roughness (RMS) for deposited
films.
79
P
Vent
Pump Chamber
Vapor Delivery System
Figure 5.2: (Right) Simplified schematic of MVD system (Left) Photograph of MVD Silica
system.
Figure 5.3: Two dimensional AFM scans of (A) a clean silicon surface, RMS roughness is
0.14 nm, and (B) 20 second OTS island film deposited on clean silicon surface ?A?.
Figure 5.4: (A) AFM image of a MVD silica layer surface with thickness of 50.0 nm and
a RMS roughness of 1.00 nm. (B) AFM image of a 20 second OTS deposition at room
temperature on surface ?A? after 60 seconds of etching. (C) AFM image of a 20 second OTS
deposition at 10 ?C on surface ?A? after 60 seconds of etching.
80
Figure 5.5: FTIR analysis of 20 second room temperature OTS island film deposition on an
oxidized silicon ATR and on a MVD silica layer before and after etching.
Figure 5.6: FTIR analysis of 20 second 10 ?C OTS island film deposition on an oxidized
silicon ATR and on a MVD silica layer before and after etching.
81
Figure 5.7: (Left) Two-dimensional representation of AFM data A-E of MVD deposited silica
layers. (Right) Three dimensional representation of AFM data A-E on MVD deposited silica
layers (2 x 2 ?m) represented. Scaling arbitrarily set to highlight data.
Figure 5.8: (A-E) 20 second OTS island film deposition on MVD silica surfaces of varying
roughness and topography after exposure to etchant procedure.
82
Figure 5.9: Illustration of proposed influence of topographical asperities on packing ability
of OTS molecules.
83
Chapter 6
Influence of MVD Surface Topography on Interfacial Water Layers- Implications Toward
OTS Monolayer Formation
6.1 Motivation
The adsorption of water under ambient conditions has an important role in interfacial
reactivity of surfaces. These surfaces adsorb water either, molecularly or dissociatively, with
strongly bound surface hydroxyl groups acting as chemically reactive groups influencing
surface properties like surface energy, chemical reactivity, catalytic activity, surface charging
and adsorption and desorption of foreign molecules [139]. The thickness of the water film at
sub-saturation conditions depends critically on the strength of the chemical bonds of water
to the substrate, as well as hydrogen bonds between water molecules within the adsorbed
water layer. The properties of these thin water films, in particular their thickness, structure
and hydrogen-bonding to the substrate, as well as, within the water film are up to now not
understood very well [140]. Investigations on smooth silicon oxide surfaces by Asay and Kim
revealed changes in the molecular configuration and thickness of this adsorbed water layer
as a consequence of increasing humidity [90], indicating the presence of an interfacial water
layer with ice-like and liquid like bonding characteristics.
The behavior of water next to any surface is a result of the delicate balance between
long-range interactions, the short-range water-surface atoms interactions, and the driving
force for water molecules to keep their hydrogen bonds network intact. Although the short
range interactions are not expected to depend significantly on the morphology of the surface,
the hydrogen bonding network is quite sensitive to this aspect, as has been demonstrated, by
recent molecular dynamics simulations where the issue of hydrogen bonding at interfaces is
directly examined [141,142]. Studies by Trakhtenberg et al. of ice layers deposited on smooth
84
surfaces have revealed that the average interaction between the adsorbed water molecule and
the surface is weaker than on a rough one. This is due to the fact that if the roughness is
on the dimension of a few molecules, the adsorbate interact with several sites on the surface
simultaneously. On the other hand, on the smooth surface the interaction is only with a
single site indicating that on a smooth surface, the water molecules distribute more evenly.
As a result, on a smooth surface, the adsorbed water molecules are more free to move and
can arrange themselves in the most stable configuration. It has also been indicated that
due to the relatively weak adsorbate surface interaction, in the case of smooth surfaces, the
adsorbate desorbs from the surface at lower temperature [143]. In the case of a rough surface,
water penetrates through structural irregularities on the surface allowing the water molecules
to interact more with the surface and the surrounding chains, and therefore remain on the
surface [144].
Researchers focusing on OTS monolayer formation have indicated, that the amount of
water present on a silica surface, influences the formation mechanism and monolayer film
properties. Most of these investigations have relied on high surface area materials to quantify
water levels, but little analysis is performed to quantify the bonding structure of water asso-
ciated with the interfacial surface on planer substrates. Through the use of a tunable MVD
silica layer, we have produced thin silica surfaces of varying morphological characteristics
that can be probed with FTIR to examine the interfacial regions. As described is section
4.2.1, the MVD silica layer has been shown to support a more extensive hydrogen bound
water layer than traditional silicon substrates with a comparable reactivity toward OTS
monolayer formation mechanisms, suggesting a link between water availability and surface
reactivity. This investigation, however, examines only one surface type. The investigation
in section 5.2 shows that, island adsorption amounts increased, as a result of an increase in
MVD silica layer thickness regardless of RMS roughness values. At low roughness values, as
illustrated by sample A in Figure 5.8, the islands appear to deposit unrestricted and similar
to the native oxide depositions as shown in Figure 5.3B. At higher roughnesses, the islands
85
appeared to deposit more restricted to the lower portions of the topography (valleys). The
influence of substrate roughness on the reactivity of the interfacial water layer remains an
undefined parameter. The work described in this Chapter attempts to determine if substrate
morphology influences the interfacial water layer that exists on the surface of MVD silica
layers under adsorption conditions and thus its reactivity.
Experimental
Materials
Silicon wafers, p-type, with (100) orientation were obtained from University Wafer
(Boston, MA). Tetrachlorosilane were obtained from Gelest, Inc. (Morrisville, PA) and
used as received. Concentrated hydrofluoric acid (HF, 49%), acetone, and isopropanol were
obtained from Fisher Scientific (Fair Lawn, NJ). Deionized water (18 M?-cm) was obtained
from a Millipore filtration system.
Methods
Cleaning methods and MVD silica layer formation schemes for the various samples have
been previously described in Chapter 3.
6.2 Results
Here, we present new evidence on the effect of surface topography on the structure and
thickness of vapor deposited water layers. Various MVD silica layer surfaces are deposited
onto plasma oxidized silicon FTIR crystals and exposed to ?45% relative humidity. FTIR
analysis of the interfacial water provides information with regards to the bonding structure
and thickness of layers. It is known that water can form a varying degrees of a hydrogen
bond networks depending on the physical and chemical conditions of the environment. When
water is completely self-associated, each water molecule has four hydrogen bonds with its
nearest neighbors forming a tetrahedral arrangement as evidenced on a smooth silicon oxide
86
surface. This is usually seen in crystalline ice structures exhibiting a broad vibrational peak
at 3250 cm?1 at or below the freezing point. Above the freezing temperature, the most
thermodynamically stable structure of condensed water is the liquid form. In the liquid
phase, the water molecule has on average two to three hydrogen bonds and exhibits a broad
vibrational peak at 3400 cm?1.
In this work four MVD silica surfaces of varying topography and thickness are presented.
The vapor deposition of MVD silica onto silicon oxides results in the formation of a robust
silica layer ranging in thickness from 18.0 nm to 65.0 nm thick, as determined by ellipsometry
on coated silicon substrates. AFM micrographs of the various silica films are shown in the
inset views in Figure 6.2 through Figure 6.5. We obtained several images of area 5.0 ?m x 5.0
?m for each substrate. The roughness values given represent the average over three images.
The RMS roughness values range from 1.5 nm to 5.6 nm with a measurement deviation of
?0.1 nm.
The results of water adsorption onto the surface of the MVD silica-modified silicon ATR
crystals are presented in Figure 6.2 to Figure 6.5. The MVD silica layers are utilized as the
background spectra. The presence of the MVD silica layer on the surface determines the
depth ofpenetrationofthe evanescent wave, but thesilica layer itself isconsidered impervious
to liquid water. In order to determine the thickness of bulk water that the evanescent wave
interacts with, the thickness of the silica layer is subtracted from the depth of penetration
of the evanescent wave. For example, for the silicon/silica system representing in Figure
6.2, the depth of penetration is approximately 505 nm at a 45? angle of incidence. In this
experiment, the silica layer is 18.0 nm thick. This means that the adsorption measured
for a bulk water sample on a silica film corresponds to 487 nm of water (505.0-18.0 nm)
at 1632 cm?1 as measured with FTIR. All thickness calculations were based on bulk water
adsorptions calculated in this manner.
The data set presented represent water adsorption at 45% relative humidity consistent
with humidity conditions indicated for uniform OTS monolayer deposition on smooth silicon
87
oxide surfaces. The data in Figure 6.2 through Figure 6.5 show that the adsorbed water
layer that exists on the MVD silica coated surfaces are dramatically different from sample to
sample. In fact, water adsorption thicknesses range from 3.9 nm to 12.69 nm. The structure
is dominated almost entirely by ice-like water in all four cases. But, the extent of hydrogen
bonding, as represented by the shifting of the ice like adsorption peak at 3200 cm?1, indicates
significantly different bonding strengths between the water and the MVD silica layer, as a
consequence of, different sample morphologies.
Data from Figure 6.2 was previously presented in Section 4.2 and illustrates a MVD
silica layer that is 18.0 nm thick with a RMS roughness of 1.5 nm that was compared with
water adsorption on a traditional silicon oxide surface. The absorption spectra for bulk water
on the MVD silica layer is presented, scaled down in size, to allow for a comparison of peak
positioning. The arrows are representative of the 3250 and 3400 cm?1 peaks associated with
an ice like and liquid like structure, respectively and the water adsorption band located at
1632 cm?1 represent water bending modes. The structure of the adsorbed water on the MVD
silica layer was dominated by ice-like water. Furthermore, the ice-like water structure had
an absorption peak centered near 3119 cm?1, which is 127 cm?1 lower than the ice like peak
associated with silicon oxide. It was concluded that more absorbed water was on the surface
of the MVD silica, which was more tightly bound compared to that existing on traditional
silica surfaces. This was evidenced by a water absorbance spectrum dominated by an ice-like
structure shifted to lower energy, greater signal intensity as compared to bulk water on the
smooth silicon oxide, and a similar reactivity toward OTS monolayer formation.
The MVD silica layer in Figure 6.5 is 36 nm thick with an RMS roughness of 5.6 nm and
showed the highest shift in the absorbance spectrum with peak positioning of 3107 cm?1,
which is 143 wavenumber less than the peak associated with ice like water on smooth silicon
oxide surfaces. The peak positioning for the interfacial water observed on MVD silica layer
in Figure 6.4 and in Figure 6.3 were 3183 cm?1 and 3225 cm?1. These MVD layers had
88
measured oxide thicknesses of 65 nm and 30 nm with a RMS roughnesses of 4.8 and 2.6 nm,
respectively.
Trakhtenberg et al. performed FTIR analysis of pure ice and found two forms of ice
distinguished by their FTIR spectra: an amorphous form, which is characterized by a peak
at about 3250 cm?1 with a shoulder at about 3400 cm?1, and a partially crystalline form,
which exhibits an absorption peak at about 3230 cm?1 with two shoulders at about 3150 and
3350 cm?1 [143]. Additional structural investigations have found, ice like water molecules
absorbing with a preferential structure have been shown to exhibit IR characteristics mea-
sured at 3050 cm?1 as a consequence of surface field [105]. The results presented for this
work are more in agreement with the absorbance values characterizing a partially crystalline
form of ice vs. the amorphous form found to exist on smooth silicon oxide surfaces. In fact,
all of the water spectra associated with the MVD silica layer suggest a partially crystalline
form of ice present with varying degrees of hydrogen bonding on the surface at 45% humid-
ity. This shift to lower energy indicates the interfacial water layer has a stronger degree
of hydrogen bonding known to induce preferential orientation with the MVD silica-coated
surface relative to the Si(100) surfaces.
In terms of interfacial water thickness, all four surfaces maintained more water on the
surface, as compared to what has been measured on smooth silicon substrates. The MVD
silica layer shown in Figure 6.2 was the thinnest MVD silica surface with a measured oxide
thickness of 18.0 nm with a calculated interfacial water layer of 3.90 nm at 45% relative
humidity. The thickness of water measured on the thickest MVD silica (65nm) layer was
12.7 nm thick at 45 % relative humidity, a value that is approximately 25% greater that
the amount measured on a smooth silicon oxide surface at 98% relative humidity. The
interfacial water layer thickness trended with increasing thickness of the MVD silica layers.
The values obtained for the MVD silica layers shown in Figures 6.3 and in Figures 6.5 with
silica thicknesses of 30 nm and 36 nm, respectively yields water amounts of 11.0 nm and 10.0
89
nm. The thickness values of the MVD silica layers are within the experimental variance for
the ellipsometric technique and are considered to be very similar.
The roughness values associated with the MVD silica layer are between 7 and 25 times
higher than those measured on a clean silicon oxide surface. It has been suggested by other
researchers that in the case of a rough surface, structural irregularities on the surface allow
water molecules to interact more with the surface forming a more tightly bound network
[144]. Based on the data, it is postulated that the different silica morphologies exhibit
varying degrees of hydrogen bonding which are dependent on topographical morphology.
As explained in a Section 4.2 water found in nanoscopic regions are known to exhibit a
more structured water character with thermodynamic properties differing from the bulk.
Again, the confinement length scales for the water existing between the solid surfaces as
experienced in this system is greatly over examined and the influence of nanoscopic regions
is indistinguishable from the interfacial water.
Nodirect correlationbetween thedegree ofhydrogen bonding, themeasured RMS rough-
ness and interfacial water layer thickness values could be found. The only trend identified
was between increasing water amounts and the MVD silica layer thickness. This is best illus-
trated in Figure 6.1 showing all four samples over layed at the same 45% relative humidity.
This figure shows that as the thickness of the MVD silica layer increases the interfacial water
layer thickness increases.
Evidence shows that more water is absorbed onto MVD silica layers as opposed to
amounts found on traditionally smooth silicon surfaces as a function of relative humidity. Of
additional complexity, it is believed that the shifting observed in the OH stretching region
corresponding to different degrees of hydrogen bonding are interrelated to the morphology of
the surface. Although, no direct trends between interfacial water structure and RMS rough-
ness as influenced by specific topographical characteristics can be deduced, it is confirmed
that the amount present vary as a function of the thickness of the silica layer.
90
Figure 6.1: ATR-FTIR Spectra of adsorbed water at 45% relative humidity on various silica
layers ranging in thickness from 18 to 65 nm. Bending mode at 1635 cm?1 is constant but
varying degrees of hydrogen bond strength presented.
Figure 6.2: ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on
an oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of ice-like
3119 cm?1 and the position of the O-H bending mode is marked at 1632 cm?1. [Inset View]
Corresponding AFM Image 5?m x 5?m. Dotted line is a scan of pure liquid water on the
MVD silica layer scaled down in size for peak position comparison.
91
Figure 6.3: ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on
an oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of ice-like
3225 cm?1 and the position of the O-H bending mode is marked at 1632 cm?1. [Inset View]
Corresponding AFM Image 5?m x 5?m.
Figure 6.4: ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on
an oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of ice-like
3183 cm?1 and the position of the O-H bending mode is marked at 1632 cm?1. [Inset View]
Corresponding AFM Image 5?m x 5?m.
92
Figure 6.5: ATR-FTIR spectra of adsorbed water on the surface of a MVD silica layer on
an oxidized silicon ATR crystal at ? 45 % relative humidity. The peak positions of ice-like
3107 cm?1 and the position of the O-H bending mode is marked at 1632 cm?1. [Inset View]
Corresponding AFM Image 5?m x 5?m.
6.3 Conclusions
MVD silica surfaces varying in RMS roughness values and different thicknesses are pre-
sented. Water adsorption onto the MVD silica surfaces changes as a consequence of different
morphological characteristics with increasing water thickness as a consequence of increasing
MVD silica layer thickness. All of the MVD silica layers exhibit FTIR characteristics repre-
sentative of an oriented highly hydrogen bound water layer. The amount of water absorbed
onto the MVD silica layers was significantly higher than, what has been previously reported
on smooth silicon oxide surfaces, with measurable variability from surface to surface. The
extent of hydrogen bonding is shown to contrast, suggesting, water layers are more tightly
bound to some samples versus others. It is suspected that the surface morphology is the
contributing factor for the increased water adsorption and the varying degrees of hydrogen
bonding observed. No correlation was confirmed between the degree of hydrogen bonding,
the thickness of the MVD silica layer, and the RMS roughness values. The results suggest
an interdependency between the characteristics under study. It is apparent from this inves-
tigation that additional considerations must be included in order to determine the surface
93
characteristics responsible and possible correlations. As a consequence of this investigation,
it is postulated that the interfacial reactivity associated with the MVD silica surfaces will
vary as a consequence of morphology adding an additional consideration for OTS monolayer
formation on rough silicon oxide surfaces.
94
Chapter 7
Influence of Cleaning Procedures on Interfacial Water Layers on Silicon Oxide Surfaces-
Implications Toward OTS Monolayer Formation
7.1 Introduction
Octadecyltrichlorosilane film adsorption on silicon oxide surfaces is dependent on the
interfacial water layer present under formation conditions. It is known that adsorbed water
has an important role in interfacial reactivity and that the reaction of silicon oxide surfaces
with water vapor leads to hydroxylation, influencing the reactivity and adsorptive properties.
At sufficiently high relative humidities bulk water will condense on the surface forming thin
water films that can alter the electrical and ionic conductivity of the surface, its adhesive and
tribological properties, and the reactivity of the surface toward other gases in the atmosphere
[140]. The properties of these thin water films under ambient conditions are noted as different
from those of bulk water; for instance, the dielectric constant of water at interfaces is about
6, compared to 78 in the case of bulk water [145] and there is a proposed existence of an
?ice like? and a ?liquid like? structure that changes as a function of relative humidity [90].
The degree of hydration necessary for uniform coverage and the role this water layer plays
in surface attachment of OTS molecules remains unclear. In general, attempts at controlling
the surface reactivity on silicon surfaces during OTS film formation consist of maintaining a
restricted environment to limit the amount of water present. This methodology has afforded
mixed results with monolayer characteristics and quality still varying from study to study.
Adding additional complexity, the choice of preparation procedures for silica surfaces is
known toinfluence surface characteristics. Forexample, water adsorptionis dependent on the
cleanliness of a surface- with contaminants (i.e. hydrocarbons) forming a hydrophobic barrier
layers resulting in non-uniform water adsorption. Work done by Donose et al.. examined
95
three commonly applied cleaning methods: the RCA ammonium and hydrogen peroxide wet
cleaning method, a water plasma cleaning method, and a UV/ozone cleaning method and
indicated that clean surfaces were produced with identical hydrophilicity, RMS roughness,
and surface charge [146]. Yet, friction force measurements on the RCA cleaned surface were
found to be several times greater than the other two cleaning methods. It was suggested
that slight deviations in surface roughness generated the observed increase in friction due
to increased water adsorption. Sirghi et al. found that small changes in surface roughness,
drastically, affected the adhesive force measurements measured between a silicon AFM probe
and a silicon oxide surface with surface roughness leading to a higher adhesive response for
water [147]. In addition, studies on glass and quartz surfaces, have revealed changes in the
thickness and structure of bound water dependent on initial surface treatments [148]. Various
techniques have been used to evaluate the structure and thickness of water monolayers as a
function of relative humidity on silicon oxide surfaces [90,94,140,147,149] and even these
studies demonstrate a variety of different preparation procedures with conflicting results.
A multitude of preparation methodologies are identifiable in OTS monolayer literature
and to date no preferred method has been established. Ultimately, researchers use various
methods relying on surface measurements, such as: ellipsometry for measuring initial silicon
oxide thickness, contact angle measurements and surface imaging techniques that examine
surface energy and RMS roughness values to ensure consistency with what is expected of a
uniform oxide film. There has been no examination of the interfacial water layer properties
as influenced by preparation conditions, namely, the thickness, structure and degree of hy-
drogen bonding with the substrate. The importance of the interfacial water layer on OTS
deposition has been established, but variationin monolayer properties exist atsimilar relative
humidities, as does a difference in methodology used to prepare silicon oxide surfaces. This
work is intended to examine any structural differences that may exist in the interfacial water
layer as influenced by preparation methodology on traditional Si(100) in order to ascertain
if changes in surface reactivity occur ultimately influencing OTS monolayer growth.
96
Experimental Details
Materials
Silicon wafers, p-type, with (100) orientation were obtained from University Wafer
(Boston, MA). Tetrachlorosilane were obtained from Gelest, Inc. (Morrisville, PA) and
used as received. Concentrated hydrofluoric acid (HF, 49%), acetone, isopropanol, Sulfu-
ric Acid A298-212 Technical Grade, and Hydrogen Peroxide H325-500(30%) were obtained
from Fisher Scientific (Fair Lawn, NJ). Deionized water (18 M?-cm) was obtained from a
Millipore filtration system.
Methods
Cleaning methods and preparation methodologies for the various samples have been
previously described in Chapter 3.
7.2 Results
The influence of cleaning methodology on interfacial water layers in terms of struc-
ture and thickness at ambient temperatures under increasing levels of relative humidity is
described. Four different cleaning methodologies are applied to silicon FTIR crystals and
evaluated in terms of initial oxide thickness, RMS roughness and contact angle. All cleaning
methodologies evidenced oxide surfaces with a CA<5? indicative of a clean and highly hy-
drophilic surface free of surface contamination with a layer of high density silanol groups on
the surface. The oxide thicknesses ranged from ?1.0 to 2.0 nm. The RMS roughness values
were evaluated by using AFM images of the wafer surface and were comparable across all
cleaning procedures (RMS-? 0.2 nm), but atomic resolution is not achievable with the AFM
system utilized in this investigation.
Four cleaning procedures were selected for this investigation. Three of these consisted of
a stepwise removal of native oxide followed with concurrent regrowth via different oxidation
97
techniques. This included oxidation with an RF Oxygen plasma treatment (HF-Plasma),
UV-Ozone oxidation (HF-UVOzone), and chemical regeneration of oxide with a piranha
solution (HF-Piranha). Detailed information is listed in Chapter 3 for conditions utilized
for oxide regrowth. The fourth condition was a comparison to previously reported data and
involved UV/Ozone oxidation cleaning of the pre-existing native oxide (Native Oxide-UV
Ozone).
To investigate the evolution of adsorbed water, the clean oxidized silicon ATR surfaces
were purged with dry nitrogen, until no differences were noted in the spectra. Each dried sur-
face was utilized as the corresponding background spectra for investigation water adsorption
under varying relative humidities ranging from 20% to 80%. All experiments were run at 27
?0.5 ?C. The adsorption of water onto the oxidized silica surfaces was conducted by varying
the relative flow rates of two nitrogen streams- one is dry and the other is saturated with
water vapor. As described in subsequent chapters, the data analysis method for studying the
evolution of the adsorbed water layer thickness on silicon through ATR-FTIR spectroscopy,
is as described by Asay et al. [90]. The thickness of the adsorbed water layer is determined
by comparing the intensity of the water O-H bending mode absorption located at 1635 cm?1
to the intensity of the same peak for a bulk liquid water sample on the same surface under
study, which is limited by the depth of penetration of the evanescent wave. The thickness
determined in this manner can be converted to the number of monolayers by dividing by
the mean van der Waals diameter of water, which is 2.82?A [90?92]. The position of the O-H
bending mode located at 1635 cm?1 has been shown as constant and is not influenced by
the degree of hydrogen bonding or structure of water present on the ATR surface [93].
The results of water adsorption on the oxidized silicon ATR crystals surfaces are pre-
sented in Figure 7.1 to Figure 7.2. The absorption spectra for bulk water on the oxidized
silica layer is presented, scaled down in size, to allow for a comparison of peak positioning.
The arrows are representative of the 3250 and 3400 cm?1 peaks associated with an ice like
and liquid like structure, respectively and the water adsorption band located at 1635 cm?1
98
represent water bending modes. For comparative purposes, a silicon ATR crystal was cleaned
following the method utilized in the work of Asay et al., by cleaning a native oxide layer
without initial oxide removal and regrowth. This was accomplished by UV/Ozone treatment
in air for 30 minutes and resulted in an ellipsometric oxide thickness of 1.8 nm. The results
presented are in good agreement with Asay and Kims previously reported data.
Asay and Kim describe three distinct growth regions, which occur approximately from
0-20% RH, from 20-60% RH, and from 60-100% RH. The first region is dominated by the
formation of an ice-like water structure, which continues to grow to about 20% RH, after
which point the formation of ice-like water slows, and the adsorption of liquid water begins.
Above 60% RH, no further increase in the ice-like water structure is observed, but the growth
is completely dominated by adsorption of liquid water. At nearly 100% RH, they report an
adsorbed water layer that is approximately 10 monolayers thick (about 5 monolayers at 80%
RH).
Our results in Figure 7.1 (top) are consistent with these observations. The position of
the ice like peak is 3258 cm?1 and the liquid like peak is found to evolve at 3400 cm?1 and
exhibit the same structural evolution as a consequence of increasing relative humidity. The
range of thicknesses obtained from 20-80 % relative humidity are consistent with reported
values revealing an average thickness at 20% RH of ? 0.5 nm increasing to ? 1.5 nm at
80% as illustrated in Figure 7.3. Studies show that a bare silicon wafer covered with a layer
of native oxide (SiO2?x), maintains siloxane rings which are very stable against hydrolysis,
therefore exhibiting properties more hydrophobic in nature. It is suspected that this hy-
drophobic characteristic contributes to the observed structural evolution of the interfacial
water. Studies of hydrophobic surfaces have revealed that water adsorption is dependent
on surface defects with water first nucleating on these spots in a well ordered manner be-
coming more disordered at increasing thicknesses. Native oxide surfaces are known to exist
with defects in the SiOx structure. It is suspected that water condenses at these defects
first allowing for a more ice like bonding character (higher hydrogen bonding) [150] with
99
subsequent layers interacting less with the surface thus display more bulk like properties at
increasing thicknesses.
Major differences are identified between the surfaces initially treated with HF in com-
parison to the UV/Ozone treatment of native oxide. HF etching is a common technique
found in silicon processing literature. Etching the native oxide (and any impurities it con-
tains) of Si(100) in aqueous HF is known to produce a hydrogen-terminated Si surface stable
against native oxidation in air for a few hours [151]. In fact, studies indicate that HF pro-
cessing creates a Si(100) surface that is detectably smoother than the native oxide techniques
resulting in a more homogeneous post oxidation Si/SiO2 interface [152,153]. Unfortunately,
H-terminated silicon surface are active in air and if oxidation is not controlled, surfaces can
degrade morphologically to form nonuniform oxides at room temperature and adsorb con-
taminations [154]. In addition, Grunthaner and Maserjian found that an HF stripped silicon
surface oxidized at a faster rate than a surface where no HF strip was performed [155]
A variety of techniques have been introduced to control oxidation on H-terminated
silicon surface resulting in an Si/SiO2 interface. The sample illustrated in Figure 7.2 (top)
was treated with HF and then RF Plasma exposed for oxide layer growth. This treatment
resulted in an oxide layer 2.0 nm with more adsorbed water as compared to all other cases
under study with an interfacial water thickness ranging from 1.1 nm at 20% RH to a final
thickness of 2.4 nm at 80% RH. A significant advantage of plasma processing stems from
lower oxidation temperatures which minimizes dopant diffusion and the generation of oxide
defects. Unfortunately, the high electric fields present during the processes can cause damage
to the resultant oxide, in particular, a high density of interface traps often result [156].
It is suspected that the cleaning of the surface by oxygen plasma induces charge on the
surface [157], which can affect the adsorbed water layer structure [158].
The sample illustrated in Figure 7.2 (bottom) was treated with HF and then UV/Ozone
exposed for oxide layer growth. This treatment resulted in an oxide layer measured at 2.0
nm. Hydrogen-terminated silicon surfaces, have been found to be readily oxidized by ozone
100
at low temperatures. Ciu et al. found that room temperature ozone treatment results in
the formation of various stoichiometric ratios of oxide films as a consequence of exposure
time with film thicknesses measured above 1.5 nm maintaining more bulk like properties
as represented by Si-O-Si peak frequencies [159]. This HF/Ozone treatment resulted in the
lowest measured thicknesses in the water structural layers as compared to all other cases
with an interfacial water thickness ranging from 0.1 nm at 20% RH to a final thickness of 0.7
nm at 80% RH. This lower level of water adsorption is again contributed to the interfacial
charge associated with the oxide surface as ozone-grown oxides are found to exhibit less
charge trapping as compared to other oxidation methods [160]
Sulfuric-hydrogen peroxide acid mixtures (H2SO4/H2O2 aka piranha treatment) are
known to produce uniform SiOH on the surface of silicon surfaces by oxidizing the con-
taminants present on the surface along with surface oxidation, leading to the formation of a
wet-chemical oxide (SiO2?x) rendering the surface hydrophilic after water rinsing, with the
installation of SiOH groups. For the data presented in Figure 7.1 (bottom) this technique
was used post HF etching. The resultant oxide was measured at 0.8 nm and was the thinest
oxide analyzed. The thickness range for the interfacial water layers was between that found
on the HF/Ozone and HF/Plasma samples ranging from 1.0 nm to 1.3 nm thick at 20% and
80% RH.
As evidenced, surfaces initially treated with HF passivation display similar growth struc-
tures regardless of the oxidation treatment exhibiting ice-like and liquid water peaks growing
together, and at nearly an equal rate, across all relative humidities studied. In contrast, the
different oxidation schemes produce varying relative thicknesses (amounts) of interfacial wa-
ter as a function of relative humidity. This suggests that the interfacial water layer exists in
a similar hydrogen bound state on each HF treated surface and that the initial conditioning
of the silicon surface determines the bonding structure. Previous studies indicate that HF
processing creates a Si(100) surface that is detectably smoother resulting in a more homo-
geneous post oxidation Si/SiO2 interface [152,153]. It is suspected that the HF treatment
101
results in a more uniform oxide film regardless of the oxidation methodology. The reactivity
of this surface as inferred from the ?binding structure? should be equivalent. It is speculated
that varying degrees of charging associated with the oxidation procedure are responsible for
the observed thickness differences. Surface charging can lead to a more ordered water layer
with water molecules dipoles being more oriented in the plane of the sample with the extent
of charging affecting the overall thickness obtained. The native oxide surface exhibited a
different structural evolution, and is postulated that, this is due to the hydrophobic nature
of the oxide in addition to a rougher silicon oxide layer with inherent defects associated with
ozone treatment without removal of the native oxide. From STM experiments ozone expo-
sure on native oxide surfaces were found to etch the surface and thus to affect the surface
structure resulting in a rougher surface [156]. The evolution of interfacial water appears
more consistent with water adsorption on a hydrophobic surface this is speculated to be due
to defects and the chemistry of native SiO2 surface.
7.3 Conclusions
Different surface treatments resulted in different water structural layer characteristics
on oxidized silicon surfaces. Surfaces initially stripped of the native oxide showed similar
structural evolution as influenced by the etchant and initial surface roughness. Different
thicknesses were observed on all HF treated samples as a consequence of charging effects
associated with the oxidation techniques. The native oxide sample showed a different struc-
tural evolution indicating a decrease in the degree of hydrogen bonding as compared to the
other techniques. The thickness of the water structural layers found on the native oxide
layers were consistent with previously reported results. Different interfacial water structures
and thickness have been identified based on different surface treatments. This difference in
absorption suggest varying interfacial reactivity in terms of the surface energy, chemical re-
activity, surface charging and will ultimately influence adsorption of OTS molecules. These
102
findings highlight the importance of the pretreatment procedure and offer an explanation for
why variation in reported results regarding OTS films still exist.
103
Figure 7.1: ATR-FTIR spectra of adsorbed water on the surface of an oxidized silicon ATR
crystal at increasing relative humidities. (Top) Native oxide exposed to a 30 minute UV
ozone treatment with a measured oxide thickness of 1.8 nm. Ice-like and liquid water are
indicated with arrows at 3258 cm?1 and 3400 cm?1, respectively, and the position of the O-H
bending mode is marked at 1635 cm?1. (Bottom) HF Removal of native oxide followed by
Piranha oxidation with an oxide thickness of 0.9 nm. Ice-like and liquid water are indicated
with arrows at 3255 cm?1 and 3400 cm?1, respectively, and the position of the O-H bending
mode is marked at1635cm?1. The direction of increasing relative humidity is from bottomto
top. Spectra correspond to relative humidities of 20%, 40%, 60% and 80% with an accuracy
of ?2.0%
104
Figure 7.2: ATR-FTIR spectra of adsorbed water on the surface of an oxidized silicon ATR
crystal at increasing relative humidities. (Top) HF removal of native oxide followed by RF
oxygen plasma treatment with a measured oxide thickness of 2.0 nm. Ice-like and liquid
water are indicated with arrows at 3257 cm?1 and 3400 cm?1, respectively, and the position
of the O-H bending mode is marked at 1635 cm?1. (Bottom) HF Removal of native oxide
followed by UV Ozone oxidation with an oxide thickness of 2.0 nm. Ice-like and liquid water
are indicated with arrows at 3259 cm?1 and 3400 cm?1, respectively, and the position of the
O-H bending mode is marked at 1635 cm?1. The direction of increasing relative humidity is
from bottom to top. Spectra correspond to relative humidities of 20%, 40%, 60% and 80%
with an accuracy of ?2.0%
105
Figure 7.3: Comparison interfacial water layer thickness as influenced by cleaning method-
ologies on Si(100) with increasing levels of relative humidity.
106
Chapter 8
Overall Conclusions / Contribution to OTS Monolayer Formation
Controlling friction, adhesion and wear on MEMS surfaces is of critical importance
for the proper functioning of MEMS devices. In recent years, many studies have been
dedicated to understanding how OTS monoloyers modulate friction, adhesion and wear in
such devices. It has been deduced that assembly conditions and film composition play
a critical role in the performance of the film as indicated by the films ability to create
well ordered assemblies on smooth surfaces. Research shows that reaction conditions have
dramatic affects on the formation properties of OTS films. There is an identified correlation
between OTS monolayers ability to form densely packed monolayers and the level of friction
and adhesion abatement attainable.
In most studies of friction and adhesion of MEMS interfaces, it is assumed that the
structure of SAMs formed and their properties are comparable to those observed when as-
sembled on flat surfaces. If an OTS monolayer exists in an ideal densely packed state on
smooth surfaces, gauche defects within the structure, contribute to the weakening of lateral
chain-chain interactions reducing film integrity. MEMS surfaces such as those of polycrys-
talline silica deviate from the traditional tribologogical observations on smooth surfaces.
Typically, MEMS surfaces consist of nanoscopic asperities that interact during intentional
or intermittent contact and it is the asperities and the monolayer film on the asperities that
dominate the tribological properties of these surfaces.
Unfortunately, resultant tribological improvements have proven unpredictable as the
monolayers have been translated to MEMS surfaces with topographical asperities. This has
been demonstrated by friction and adhesion studies performed on MEMS test platforms
created from machined polycrystalline silicon.
107
This dissertation evidences a relationship between monolayer adsorption characteristics
and topographical asperities with observed variations in monolayer order resultant from
surface roughness. Results show that the packing structure of OTS monolayers is dependant
on local asperity curvature varying significantly from that observed on flat surfaces. A silica
film produced from the low temperature, vapor-phase hydrolysis of tetrachlorosilane with a
tunable topography was leveraged as a novel investigative platform to reveal the influence
of surface topography on OTS film formation.
An examination of interfacial properties of the MVD silica layer revealed a highly hy-
droxylated surface lacking free OH groups with an ordered water structure that is more
tightly bound due to a higher degree of hydrogen bonding associated with the hydroxylated
surface. In contrast to previously reported results, the mechanism of OTS films growth is
found to be comparable to those formed on traditional silicon oxide surfaces, despite a higher
amount of water present and a lack of free OH groups. As a consequence, it was evidenced,
that the role of free OH groups on the formation mechanisms for OTS film formation are
inconsequential, highlighting the importance of interfacial water on the OTS film formation
process. It is shown that high quality OTS films can be formed on a highly hydrated surface
as evidenced with traditional analytical techniques optimized for smooth silica surfaces.
A dilute etchant technique was developed to provide topographic contrast for AFM
imaging enabling an investigation true monolayer location relative to topographical asper-
ities. The results evidence a preferential monolayer adsorption and variation in molecular
packing density influenced by surface topography. It is shown that the adsorption of OTS
monolayers occurs differently on various surface topographies with preferential adsorption
occurring in the lower regions of the surface leaving exposed topographical asperities. The
extent of coverage to the surface asperities is suspected to depend on the curvature associ-
ated with the asperities. Surfaces with higher curvature support monolayers less laterally
interacted on the peaks. This will result in a lowered tribological characteristics based on
108
asperity-asperity interactions. From these results, it is determined that the LC phase do not
deposit uniformly on the asperities.
In addition, a difference in surface reactivity is observed as a result of different surface
topographies with thicker silica layers maintaining a higher coverage at similar reaction times
and conditions. This observation lead to a detailed investigation of the interfacial water layer
and changes in surface reactivity. Water adsorption onto the MVD silica surfaces is shown
to change as a consequence of different morphological characteristics. All of the MVD silica
layers exhibit FTIRcharacteristics representative ofanoriented highlyhydrogen boundwater
layer. The amount of water absorbed onto the MVD silica layers was significantly higher
than what has been previously reported on smooth silicon oxide surfaces with measurable
variability from surface to surface. The extent of hydrogen bonding is shown to contrast
suggesting water layers are more tightly bound to some samples versus others. It is suspected
that the surface morphology is the contributing factor for the increased water adsorption and
the varying degrees of hydrogen bonding observed. No correlation was confirmed between
the degree of hydrogen bonding and the RMS roughness values. But, the extent of water
present on the surface correlates well with the thickness of the MVD silica layer. The higher
content of interfacial water causes an increase in the rate of adsorption as observed under
similar reaction conditions and increased coverage. As a consequence of this investigation,
it is postulated that the interfacial reactivity varies as a consequence of morphology adding
an additional consideration for OTS monolayer formation on rough silicon oxide surfaces.
Based on these observations it is suspected that OTS monolayers will deposit differently
on every surface investigated and be dependent on the overall profile of the asperities. In
addition, the interfacial reactivity associated with adsorbed water will vary based on the
interfacial topography. It is postulated that this is a contributing factor for why friction
measurements vary from group to group and sample to sample on MEMS devices and war-
rants future considerations when trying to utilize OTS monolayers deposited from the mixed
regime on surfaces.
109
In addition, different surface treatments (etching and oxidation methods) resulted in
different water structural layer characteristics on smooth oxidized silicon surfaces. Surfaces
initially stripped of the native oxide showed similar structural evolution as influenced by the
etchant and initial surface roughness. Different thicknesses were observed on all HF treated
samples as a consequence of charging effects associated with the oxidation techniques. The
native oxide sample showed a different structural evolution indicating a decrease in the de-
gree of hydrogen bonding as compared to the other techniques. The thickness of the water
structural layers found on the native oxide layers were consistent with previously reported
results. Different interfacial water structures and thicknesses have been identified based
on different surface treatments. This difference in absorption suggest varying interfacial
reactivity in terms of the surface energy, chemical reactivity, surface charging and will ulti-
mately influence adsorption of OTS molecules. These findings highlight the importance of
the pretreatment procedure and offer and explanation for why variation in reported results
regarding OTS films still exist on smooth surfaces.
A novel high surface area silica support is introduced capable of sustaining loadings
measurable with TGA analysis. This support exhibits a similar reactivity toward OTS film
decomposition as compared to other high surface area silica materials and sustains higher
loadings. In addition, results suggest that due to the lack of porosity, low degree of surface
curvature, and the increased loadings obtained, that the fibrous silica, provides a more
representative support material for the determination of loading information for comparison
to the films produced on planer substrates. This support has been utilized to determine the
influence of annealing environments on the thermal stability of OTS films. It was shown
that films annealed under an air environment decomposed at lower starting temperatures
and to a greater extent than those annealed under inert atmospheres.
It is concluded that MEMS researchers focused on using OTS monolayers for friction
and adhesion improvement must first understand the consequence of surface topography and
interfacial reactivity on the deposition mechanics and overall film quality. Ultimately, this
110
work indicates that variation in topographical asperities will result in an unpredictable OTS
monolayer deposition dependant on surface topography. This unpredictable film will result
in device to device variation in tribological performance. Future work in the field should be
aimed at controlling surface topography and understanding what characteristics are required
to ensure an optimal film.
In conclusion this work shows that researchers must determine a nominal surface topog-
raphy and interfacial water structure required for successful utilizatation of OTS monolayers
for friction and adhesion improvement on rough surfaces.
111
Chapter 9
Self-Assembled Octadecylphosphonate Monolayers on Metal Oxide Substrates
9.1 Motivation
This research is a portion of a collaborative research effort designed to bridge the sci-
entific and technical areas of micro/nano tribology. The overall goal of this research is
to improve understanding of the molecular origins of friction by investigating microscopic,
multi-asperity interfaces in well-controlled experiments and by comparing results across ve-
locity regimes. More specifically, in order for these comparative efforts to be successful it
is necessary that the interfaces be structurally identical. The intent of this portion of the
collaborative effort was to optimize the coating methodology necessary to produce an opti-
mal densely packed monolayer translatable to various technologically relevant substrates in
an effort to provide an uniform interface comparable across different micro/nano tribological
techniques.
9.2 Background
The principal research tools of nanotribology have been the AFM, the quartz crystal
micro balance (QCM), the surface force apparatus (SFA), and MD simulations, along with
myriad extensions and refinements of these [4,61,120]. Unfortunately, there have been
relatively few attempts to compare measurements of friction between different tribological
instruments. The information revealed about friction is decidedly instrument dependent,
since the contact area, pressures and velocities accessible to a specific instrument affect
the friction of the interface. Well controlled comparisons of friction results from different
techniques are important in order to gain a better understanding of friction in different
112
physical regimes. To date there have been few comparative studies between the techniques
of AFM and SFM. Many more comparative studies of this kind are needed before researchers
in micro/nanotribology can be sure that different techniques reveal consistent information
about frictional affects. To enable this type of comparison it is necessary to utilize the same
interface and substrate materials. SAMs are to be utilized given their ability to form well
defined and uniform interfaces. These interfaces can be tailored to study the resulting effects
on friction and other properties.
9.2.1 Research Objectives
Various surface engineering techniques have been introduced to reduce friction and ad-
hesion and one general approach is to avoid the formation of water layers on the surfaces by
interfacial treatments that provide a stable hydrophobic surface, thereby eliminating cap-
illary forces. To be effective, such chemical modification, must also be stable both under
ambient and elevated temperatures, be hard and wear resistant, and prove conformal on
surfaces with inherent roughness. Much research has been dedicated to the chemical modi-
fication of surfaces with coatings designed to tailor the interfacial properties indicating that
despite the measurement scale, the frictional properties are influenced by molecular tilting,
inter atomic interactions, interface stability, adhesion to substrate, and gauche defect for-
mation. Dissipation of frictional energy in SAMs through the formation of gauche defects
has been thoroughly investigated concluding that increased disorder correlates with higher
gauche defects and an easier dissipation of frictional energy thus stressing the importance of
uniform film structure and tribological improvement.
Phosphonate monolayers are the focus of this investigation as they show significant
promise to define new directions in MEMS technology and have the potential to overcome
the limitations of silane-based SAMs with respect to immersion and thermal stability. Prior
investigation of alkylphosphonic acids on flat Al sheets showed that SAM formation of alkyl
113
phosphonates on Al surfaces relies on the hydroxylation of the oxide (alumina) layer. Gen-
erally, chemisorption of alkyl phosphonic acid occurs by proton dissociation to form an
alkylphosphonate species. Due to the nature of the surface reaction between alkylphospho-
nic acids and OH surface species, this class of self-assembled monolayers, could overcome
some of the plaguing reproducibility issues associated with silane chemistries such as hydra-
tion control, inherent processing difficulties, and polymerization concerns as related to metal
substrates. In addition to offering a physical protective barrier, the monolayers, by virtue of
their inert hydrophobic nature, also provide a degree of chemical protection [80]. The tribo-
logical mechanisms for surfaces coated with phosphonate SAMs have not yet been studied
extensively, although these monolayers have recently begun to attract the attention of tribol-
ogy research groups [25,79]. Friction force measurements with the phosphonate monolayer
coated smooth alumina demonstrated a remarkable improvement in performance when com-
pared to untreated alumina surfaces the coated surfaces exhibited remarkable robustness and
resistance to wear over each experiment.
For comparison purposes, octadecylphosphonic acid monolayers (ODP) were selected as
the SAM of choice for this study enabling a tribological comparison to an already extensively
studied silane counterpart octadecyltrichlorosilane (OTS) with proven efficacy. Although the
bonding mechanism associated with ODP is significantly different than that proposed for
OTS, the hydrophobic properties associated with its 18 carbon segments should provide a
similar tribological improvement assuming a comparable packing efficiency. The phosphonic
acids undergo a condensation reaction with surface bound alumino hydroxyl (-Al-OH)species
to form aluminophosphonate compounds with H2O as a byproduct [81]. An inelastic electron
tunneling spectroscopy study of the adsorption and structure of alkyl phosphonic and alkyl
phosphoric acids suggested that tridentate species were formed with the aluminum oxide
surface [82] but other groups have suggested a bidentate species [161]. It has been shown
that ODP adsorbs preferentially to aluminum surfaces that are hydroxylated in air and that
an increase in the surface hydroxyl density can accelerate adsorption kinetics [83]. Such an
114
acceleration can be explained by the adsorption of the phosphonic acid via surface hydrogen
bonds prior to the condensation reaction, leading to the finally adsorbed phosphonate [84].
After considering several metal oxides, we have chosen to focus on aluminum oxide
as the structural material. Two aluminum oxide substrate materials reactive toward ODP
deposition were identified and evaluated in this work. Based on its commercial availability
as a quartz micro balance electrode and its promise as a component in device fabrication the
first focused on an oxidized aluminum surface. The second support material was aluminum
oxide spheres. These spheres are representative of those to be attached to AFM and SFM
probes through collaborative efforts, but obtained in bulk with a larger distribution in size
as compared to those mounted to the tribology probes. Acquisition and utilization of this
representative material allowed for additional analytical insight before direct translation of
the ODP films to those spheres attached to the tribology probes.
In order to prepare an ideal interface a detailed mechanistic study was performed to
analyze the quality of these films. In the study of monolayers and thin films on surfaces, both
their surface and bulk properties are important. Several analytical methods are introduced
for characterization of the monolayers and thin films produced throughout this work. The
deposition quality was assessed in terms of packing density and uniformity with FTIR, TGA,
ellipsometer, contact angle analysis and TGA.
9.2.2 Results
Sputtered Aluminum Surfaces
The ODP monolayer formation mechanism was investigated on an oxidized sputtered
aluminum surface. The surfaces were cleaned and plasma oxidized as described in Chapter
3. The clean oxidized surfaces were analyzed with AFM and ellipsometry and an initial
oxide thickness was measured at 2.5 nm with a RMS roughness of 4.3 nm. This surface is
illustrated in Figure 9.2 following the plasma oxidation. The sputtered aluminum surface was
compared to the surface of an aluminum coated quartz crystal micro balance and the surface
115
Figure 9.1: Water contact angle (squares) and ellipsometric film thickness (diamonds) vs.
immersion time for ODP films on oxidized aluminum oxide surfaces at room temperature.
roughness and oxide values were found to be similar. Due to the limited availability and
high cost associated with the quartz crystal micro balance substrates initial characterization
experiments focused solely on the sputtered aluminum surface.
Deposition of the ODP monolayers as described in Chapter 3 were performed at room
temperature under atmospheric conditions with film formation times ranging from 5 sec to
4 hours. Contact angle measurements ranged from 99 to 109.9? consistent with previously
reported results obtained on similar aluminum oxide surfaces [83]. Contact angle measure-
ments indicated that no significant changes in contact angle were observable after 1 hour of
deposition. But, ellipsometric measurements indicated that coverage increased as a function
of immersion time with thickness measurements ranging from 18.9 ?A to 25.1 ?A at higher
immersion times. These results are illustrated in Figure 9.1.
AFM was performed on each substrate before and after deposition, generally, not enough
contrast was visible between the phosphonate film and the underlying substrate with features
undiscernable from the underlying surface topography. For illustration a monolayer formed
after 4 hours of deposition on the sputtered aluminum substrate with a contact angle of
109.9? is illustrated in Figure 9.3 for comparative purposes with Figure 9.2.
116
Figure 9.2: AFM image of a 5 ?m x 5 ?m plasma cleaned MVD aluminum wafer with RMS
roughness of 4.3 nm and oxide thickness of 2.5 nm.
Figure 9.3: AFM image of a 5 ?m x 5 ?m 240 minute deposition of ODP on plasma cleaned
MVD aluminum wafer with RMS roughness of 4.67 nm, contact angle 109.0? and ODP
thickness of thickness of 2.5 nm.
117
Figure 9.4: Representative FTIR spectra of ODPmonolayer formationon ?plasma activated?
aluminum surface as a function of immersion time at room temperature.
Although AFM provides information on the microscopic structure of OTS film growth, it
does not reveal molecular structure and conformation changes within the film and, as demon-
strated, can not provide visual contrast on surfaces with topographical roughness. To obtain
structural information infrared reflection absorption spectroscopy (IR-RAS) was performed
using a Perkin Elmer Spectrum 5000 spectrometer equipped with a liquid nitrogen-cooled
MCT detector. The reflection spectra were collected using a Perkin Elmer Specular re-
flectance accessory, mounted in the sample chamber, set to reflect light at an angle of 83?.
A ZnSe wire grid polarizer was mounted on the accessory and was aligned to allow only
p-polarized radiation to reflect from the surface. The background spectra for ODP films
deposited on the oxidized aluminum surface was a clean aluminum surface with a measured
oxide layer thickness of 2.5 nm. FTIR analysis was performed on the films adsorbed on the
oxidized aluminum surface as a function of immersion time.
118
The results presented in Figure 9.4 are the FTIR spectra for films formed on the oxidized
aluminum layer as a function of immersion times at room temperature. In general, mean
orientation, packing density and interfacial bond formation of the self assembled monolayer
can be analyzed in terms of polarized IR-RAS analysis given that films are produced on a
reflective underlying metal substrate. The prominent feature of these spectrum is the pres-
ence of symmetric and asymmetric methyl CH3 and methylene CH2 -stretching vibrations.
As shown in previous studies the methylene stretches are the most commonly monitored- as
the frequency of the gives an indication of orientation and the packing density for long-chain
SAMs.
Figure 9.4shows two types of changes in the FTIR spectra as the film formation time was
increased. The first observation is that the position of the ?s CH2 mode is decreasing with
increasing adsorption time to a final value of 2917 cm?1 after 4 hours of ODP adsorption. A
similar trend is seen with the ?a CH2 mode with values decreasing from 2853 cm?1 to 2850
cm?1. This shift in wavenumber for the two methylene stretching modes is indicative of a
more ordered film. For completely disordered structures, the frequency of the asymmetric
CH2 stretching is close to that of a liquid alkane (? 2924 cm?1. For well-ordered SAMs,
the frequency is expected is close to that of a crystalline alkane (? 2915-2918 cm?1). This
indicates that as the surface coverage increases, the order in the monolayer gradually increase
(with the wavenumber for ?s CH2 and ?a CH2 modes decreasing) becoming progressively
more ordered. An additional observation is a consistent broadness of the peaks associated
with both the symmetric and asymmetric CH2 modes, which will be detailed later, in this
document.
A second change appears in the intensities of the CH2-stretching vibrations, which
decrease as a function of immersion time. With increasing time of adsorption the CH2 band
intensities decrease while the CH3 band intensities remain almost constant, in response to the
selection rules associated with polarized IR-RAS, indicating that the alkyl chain orients itself
with its long axis more normal to the aluminum surface. An optimal rectangular adsorbed
119
molecule with no tilt will have no intensity due to the fact that CH2 groups would not have
a transition dipole moment in the direction normal to the investigated surface necessary for
signal.
Ellipsometric data in combination with the infrared data suggest that the ODP mono-
layer adsorption mechanism initially consists of molecules randomly distributed on the sub-
strate. As the surface coverage increases, the order of the monolayer gradually increases
consistent with a decrease in tilt angles of the chains as indicated by a decreasing IR band
intensity data (Figure 9.4) and are congruent with increasing ellipsometric film thickness
values. Final film thickness values are measured at ? 2.5 nm, but the ellipsometric model
utilized forthickness extraction assumes the the film refractive index remains constant during
the self-assembly process and is equal to the fully formed monolayer. As indicated with ellip-
sometric measurements in previous reports on OTS monolayers, this is not a valid assumption
and data only give representative trend information and not exact measurements [162]. The
surface roughness associated with the sputtered aluminum substrates adds complexity to the
exactness of the ellipsometric values. Investigations performed by Pellerite et al. on H16PA
and H22PA show a range of thickness values between 1.7 nm and 2.7 nm, respectively, on an
aluminum oxide surface with similar roughness values and the data from this study are in
agreement with this range [163].
Tillman et al. derived a means to estimate the molecular orientation in terms of inclina-
tion angle from surface normal with IR-RAS. In accordance with the surface selection rule,
IR-RAS spectra only contain vibrational absorptions with species oriented in the p-plane
thus perpendicular to the surface. Non-oriented random hydrocarbon chains exhibit high ?a
CH2/?a CH3 ratio because the C-H stretch of the methylene groups has a dipole moment in
the p-plane [12,164].
120
The mean inclination angle can be derived from the ?a CH2/?a CH3 peak ratio, and was
calculated according to the following equation (taken from ref. [162]):
ICH2
ICH3 =
2
3
n?cos2(??)
cos2(?? ?55?) (9.1)
Where n represents the methylene/methyl ratio, 23 is the ratio of hydrogen in a methylene
group to a methyl group and the angle of inclination calculated by (90?-??). The calculation
for OTS films produced in that work gave a value of ICH2/ICH3 equal to 0.9 with a calculated
angle of inclination of (90?-??) = 15?, this value was corroborated to orientational results
observed with ATR spectroscopy and ellipsometric measurements, consistent with was ap-
proximated as vertical, extended alkyl chain axes. It is important to note that subsequent
research on monolayers produced at reduced temperatures result in a more tightly packed
monolayer with reduced tilt angles and actually represent optimal all trans orientations while
the monolayers presented by Tillman are evidenced as a mixture of ordered and disordered
phases.
The ICH2/ICH3 ratios measured for the phosphonate monolayers formed as a function
of immersion time are shown in Figure 9.5. The ICH2/ICH3 ratio decrease as a function of
immersion time, indicative of alkyl chain orientation, approaching the surface perpendicular,
removing the ?a CH2 from the p-plane. At formation times of 4 hours the angle of inclination
was calculated at 16? ?0.8?) suggesting a similar packing arrangement as found in OTS
monolayers formed at room temperature.
Figure 9.5: Calculated inclination angles and ICH2/ICH3 ratios as a function of immersion
time
121
Literature discusses three forms of arrangement possible for phosphonic acid adsorption
mono-,di, and tridentate for binding the phosphonate headgroup to metal oxide surfaces.
However, our IR data alone do not allow a direct determination of the preferred binding mode
in the system studied here. But, calculations suggest an evolution from tridentate (of about
35?)atlow immersion times tolower binding modes (ofabout16?)as film formationproceeds.
This indicates a decrease in alkyl chain tilt angles and points to a more organized film.
As evidenced by the broadening affect previously mentioned, the agreement in calculated
angle of inclination to OTS monolayers produced at room temperature, it is suspected that
a small amount of disorder is still present in our ODP films even after high immersion
times. Deposition of OTS films at room temperature at increased immersion times show
a dimunition of the peak broadening affect on smooth surfaces. Given the influence of
the interfacial water on OTS adsorption mechanisms, molecules are more free to rearrange
to thermodynamically stable conformations. But, friction force microscopy has shown that
ordered phase and a disordered phase are still present in OTS films. In conjunction with film
order studies presented previously of OTS on surfaces with varying degrees of roughness, the
surface roughness associated with the sputtered alumina will have a higher degree of disorder
as a consequence of surface features. In addition in the low frequency regions of the spectra
we note peaks associated with P=O groups with vibrations in the range from 1180-1220
cm?1indicative of incomplete cross linking.
It is speculated that the growth mechanism associated with the phosphonate film under
study contribute to varying bonding structures that may limit complete cross linking. In
addition, consideration must be given for the influence of the nanoscopic curvature associated
with this surface, as curvature has been shown to influence the order of self assembled films.
It is suspected that the orientation could improve if studies were performed on a planer
atomically smooth substrate. The utilization of a smooth substrate has been proposed for
phase II of this investigation but is not part of this body of work.
122
Spherical surface
In the case of spherical surfaces, methods such as ellipsometry, contact angle analy-
sis, and IR-RAS are not applicable. ATR-FTIR spectroscopy was used to investigate the
state of order for the alkane chains. However, ATR-FTIR measurements are useful only
for monitoring trans/gauche sequences in the alkyl chains (order), and can not be used to
measure lateral interaction packing. TGA analysis is utilized to obtain this information by
measuring weight loss as a function of temperature under a nitrogen atmosphere. A electron
micrograph of the ODP spheres used for this research is shown in Figure 9.6 and consists of
mainly spherical particles with an average diameter 75 ?m.
Figure 9.6: Electron micrograph of spherical alumina powder with a distribution in size
ranging from 75?m to 100?m utilized for ODP deposition and subsequent FTIR analysis
and TGA analysis.
Research has shown that the decomposition mechanism is similar to that identified for
OTS monolayer, but the temperatures are elevated [165]. The thermal degradation of the
monolayers occurs via destruction of the hydrocarbon chains with phosphorus remaining on
the surface after the pyrolysis.
Spherical alumina powder was cleaned and coated with ODP monolayers following the
methods indicated in Chapter 3. ODP depositions were performed as a function of differ-
ent immersion times with data indicating an increase in weight loss observed at increasing
immersion. The highest immersion time was 4 hours, consistent with results from planer
123
investigations. A typical TGA weight loss curve for the monolayers of ODP supported on
the alumina spheres is shown in Figure 9.8. The most prominent feature was the dramatic
weight loss in the temperature region from ?200 to ?700 ?C, which was not present in the
TGA of bare alumina and was assigned to the degradation of the surface adsorbed organic
species. The onset of the weight loss was ? 250-300 ?C and the temperature of the maximal
rate of the weight loss (TMAX) was ? 400-600 ?C. To obtain an estimate of the extent of
coverage of OTS on the surface, 100 % coverage was calculated assuming a maximum pos-
sible value of 5 molecules/nm2, and using 0.2 m2/g as the surface area of alumina spheres.
Therefore, the percent coverage for the spherical alumina was 0.042%, and lower than the
amount calculated, suggesting an incomplete monolayer and non uniform coverage. We point
out, however, that the weight loss (from 200 to 700 ?C) was 15-20% lower than the weight of
the monolayers as determined from calculations based on the an all trans film arrangement
and surface area available. But, the proposed decomposition mechanism in this range indi-
cates that only part of grafted molecules degraded into volatile products while a portion of
the monolayers (?15-20% by TGA curves in weight) remain on the surface (until 800 ?C).
Re-calculating the loading percentages assuming that the inorganic species remain on the
surface resulted in 99.4% loading effecacy.
ATR-FTIR analysis was performed at increasing immersion times. Figure 9.7 shows
two types of changes in the FTIR spectra as the film formation time was increased. The first
observation is that the position of the ?s CH2 mode is decreasing with increasing adsorption
time to a final value of 2917 cm?1 after 4 hours of ODP adsorption. A similar trend is
seen with the ?aCH2 mode with values decreasing from 2853 cm?1 to 2850 cm?1. A second
observation is the increase in intensity associated with these modes. These observations
indicate that as the surface coverage increases, the order in the monolayer gradually increase
(with the wavenumber for ?sCH2 and ?aCH2 modes decreasing) becoming progressively more
ordered and are consistent with what was observed on the planer substrates. ATR-FTIR
does not allow for a direct determination of the degree of order present.
124
Figure 9.7: Representative FTIR spectra of ODPmonolayer formationon ?plasma activated?
aluminum oxide spheres as a function of immersion time at room temperature.
Monolayer films were translated to quartz crystal micro balance crystals and ODP
spheres attached to AFM and SPM probes based on the conclusions found in this research.
A holder was designed for this effort as illustrated in Figure 9.9 to allow for deposition to
performed at the same time to various surfaces. Collaborative results are under way and
will determine the efforts that follow.
9.2.3 Conclusions to Phosphonate Work
Octadecylphosphonic acid monolayers (ODP) were investigated for a tribological com-
parison across different measurement regimes. Two aluminum oxide substrate materials re-
active toward ODP deposition were identified and evaluated. A detailed mechanistic study
was performed to analyze the quality of these films. Results show that ODP evolves from a
tridentate bonding structure to a lower bonding configuration at increased immersion times
125
Figure 9.8: TGA weight loss as a function of temperature under nitrogen at a ramp rate of
10 ?C/min comparison between two different immersion times. Scan of uncoated alumina
spheres are removed from baseline.
126
Figure 9.9: Photograph of holder designed for holding AFM Cantilevers, SPM Probes, and
Quartz Crystal Micro balance Assemblies for simultaneous phosphonate monolayer deposi-
tion.
with increasing film order. As produced on rough oxidized aluminum surfaces, they maintain
a small degree of disorder. Films adsorbed on the spherical particles show an increase in
order as a function of immersion time and maintain a lower degree of disorder more indica-
tive of an all trans orientation. Films were produced under identical conditions and sent for
collaborative tribological measurement.
127
Part II
Interfacial Engineering of a Catalytic Fischer Tropsch
MicroReactor
128
Chapter 10
Motivation
While the Fischer-Tropsch reaction system is rich in character, there is no fundamental
kinetic model to describe the reaction, and therefore reactor design becomes an empirical
exercise. It is generally understood that operating FT reactions at low pressure and high
temperature will favor methane whereas high pressure and lower temperature will favor
wax products. Conventional reaction engineering approaches (isothermal reactors, isobaric
reactors, plug flow or tubular reactors) have been used extensively and successfully to char-
acterize this reaction. However, due to lack of fundamental knowledge about the reaction
details, it is not always clear how or why alteration of key process parameters will affect
product distribution and selectivity. It is known that both process pressure and tempera-
ture influence selectivity and product distribution. We intend to explore the effects of time
varying pressure and temperature as well as space varying temperature and pressure in order
to learn if or how these factors can be tailored to provide a targeted product and selectiv-
ity. The fundamental goal of this project is to leverage microreactor technology combined
with a micro fibrous catalyst support system (MCSS) to probe the effects of high gradient
(both temporal and spatial) process conditions on reactive systems to advance fundamental
knowledge in interfacial control over reactor operation. As a platform for this effort, Fischer-
Tropsch synthesis is investigated over silica supported cobalt catalyst in a nickel wire micro
fibrous matrix. The principal objectives of this work include (1) identifying a microreactor
capable of performing the FT reaction, while being easily tunable for operating parameters
such as pressure gradients, temperature profiles, and reactor geometries (2) developing an
active catalyst material that can be incorporated into the micro reactor system easily and
controllably (3) systematically probing of the operating parameters in the micro reactor
129
to leverage the reduction in size as compared to conventional systems utilizing the advan-
tages in improved heat and mass transfer (4) Utilization of the micro reactor to investigate
Fischer-Tropsch synthesis weighing conversion and activity verses information learned from
the third objective. This is an extensive investigation that will not be fully envisioned by
the end of this dissertation. As a first chapter for this research, this work will only focus on
understanding the effect space varying pressure as influenced by the MCSS material and its
subsequent influence on product distribution and selectivities as related to Fischer Tropsch
synthesis. These experiments will be studied under isothermal conditions. The next phase
of this investigation will include an analysis of the influence of microreactor geometry as a
means to control the pressure gradient within the reactor.
10.1 Background Research
10.1.1 Introduction to MicroReactors
Microreactors are miniaturized chemical reaction systems, which contain reaction paths
with at least on dimension in the range of tens of microns to approximately 1 mm in size. The
small reactor dimensions lead to a relatively large surface area-to-volume ratio and increased
driving forces for heat and mass transport. Therefore, microreactors are especially suited for
fast reactions with a large heat effect, where they allow for nearly isothermal conditions at
high reactant concentrations, or for operation in the explosive regime, which is not possible
in large-scale equipment [166].
Short characteristic length scales intensify heat and mass transport because the con-
duction and diffusion lengths are short as compared to those of large scale systems. A high
heat transfer rate enables a direct control of the process temperature, e.g. fast preheat-
ing to the process temperature or cooling and quenching of reaction products to prevent
the formation of unwanted side products. The high heat transfer rate in combination with
small overall equipment size enables fast start-up and shutdown of the entire process. Fast
start-up is a benefit for on demand production and fast shutdown will increase safety in case
130
of failure. Down scaling a high temperature unit operation, can increase the temperature
gradient between the inside and the outside of the unit proportionally. On the other hand,
a large temperature gradient also gives a high driving force for heat transfer. Short diffusion
lengths improve mass transport limited process rates and a large surface area to volume ratio
enables a more efficient use of exchange surfaces for e.g. heterogeneous catalytic reactions
and membrane processes.
The throughput of the process can be increased per unit volume, by making more
efficient use of these exchange surfaces. Due to the small length scales, Reynolds numbers in
micro systems are low, in the order of 1-1000 and therefore a laminar flow pattern will exist
inside micro channels. Laminar flow can complicate mixing of different media. The small
volume and precise geometry of micro structures can give a short and well-defined residence
time. In this way, the residence time of the chemical process can be optimized, resulting in
higher yield and higher reaction product selectivity compared to large scale systems.
Next to these process-yield and selectivity enhancing features, micro-chemical systems
are inherently safe. The high heat removal capabilities prevent hot spot formation. Runaway
chain reactions like flames and explosions are averted. The short diffusion length scales
promote the adsorption of free gas phase radicals at the walls and accordingly terminate
these homogeneous reactions. Moreover, in case of failure, the small hold-up of reactants
will give less waste and low environmental impact. The influence of erosion and corrosion on
construction materials will have alargerimpact onamicro system compared toamacroscopic
system. The surface area to volume ratio and inherently the hydrodynamic properties of a
micro system will change more drastically for a certain diameter change by erosion and or
corrosion. The construction materials will need to be highly erosion and corrosion resistant
or will need a protective coating. Although these materials are often more expensive and/or
harder to utilize in a high precision fabrication process, less material is needed to build a
micro system. Micro systems can therefore enable the use of exotic and costly materials.
131
Another advantage of micro-chemical-systems is the scale-up by scale-out or numbering-
up the system. The throughput of one single channel might be low, but by using more
channels in parallel, production can easily be increased. Using micro structures with the
same dimensions, no scale-up research is necessary, saving time and money on research.
However, to ensure a good flow distribution between parallel micro-systems or inside a stack
of channels, the relative difference between the channel diameters should be kept small and
also a flow distributor will be necessary. Extensive reviews on micro-chemical-systems can
be found in the books of Ehrfeld et al. [167], Hessel et al. [168], and Lowe et al. [169].
It is important to note that micro reactors do not affect the overall reaction mechanisms
and kinetics. The main impact in micro reactor design is to focus on intensifying mass and
heat transport as well as improving flow patterns. A decrease in the linear dimensions causes
a difference in the temperature, concentration, density, or pressure gradients. Consequently,
the driving forces for heat transfer, mass transport, or diffusional flux per unit volume
increase when using micro reactors.
10.1.2 Catalyst Considerations
For an optimal heterogeneous catalytic reactor design, many variables must be analyzed
to maximize reactor performance in terms of yield and selectivity. To this end, an integrated
approach should begin at the micro-level by an optimal design of the catalyst itself by con-
sidering features such as catalyst size, shape and uniform distribution of the active material.
The active materials used as catalysts are often expensive metals, and in order to be utilized
effectively, they are dispersed on large surface or supports. The reactant diffuses from the
bulk fluid within the porous network of the support, reacts at the active catalyst site, and the
product diffuses out. The transport resistance of the porous support alters the concentration
of chemical species at the catalyst site as compared to the bulk fluid. The consequence of
concentration gradients is that reactions occur at different rates, depending on position of
132
the catalyst site within the porous support. As far as the size of the catalyst pellet is con-
cerned, it should be realized that although the use smaller size of the catalyst pellet would
be optimal, it is often the pressure-drop considerations that outweigh other consideration
in view of the fact that the pressure drop considerably increases as the size of the catalyst
in the heterogeneous reactor is decreased. Therefore, any optimal design should explicitly
take into consideration of the issue of the pressure-drop since most of the reactor designs
used so far are the traditional catalyst particulates packed into microreactors that normally
suffer from poor intra particle mass / heat transfer, low contacting efficiency, high pressure
drop, mechanical attrition, and catalyst clumping in a way [170]. Our research intends to
overcome many of these limitations by focusing on the incorporation of an MCSS catalyst
that has been identified as a unique catalytic system which will be further expanded on and
investigated in this work.
10.1.3 Introduction to Fischer Tropsch Synthesis
Chemistry
In Fischer-Tropsch synthesis, liquid hydrocarbon fuels are produced from lighter gases.
This process was first put to large-scale industrial use by Germany during World War II.
There continues tobeintense academic andcommercial interest in improving Fischer-Tropsch
synthesis because it offers a source of liquid hydrocarbon fuels as opposed to the increasingly
costly process of extracting oil from the ground. Furthermore, Fischer-Tropsch synthesis
could make a significant beneficial environmental impact by capturing methane and other
greenhouse gases that might otherwise be released from remote oil wells.
A problem with Fischer-Tropsch synthesis is that it is difficult to control the products
resulting from the synthesis. FT synthesis is a surface polymerization reaction. The reac-
tants, CO and H2, adsorb and dissociate at the surface of the catalyst and react to form
chain initiator CH3, methylene CH2 monomer and water. The hydrocarbons are formed by
CH2 insertion into metal-alkyl bonds and subsequent dehydrogenation or hydrogenation to
133
an -olefin or paraffin, respectively. The pioneering work by Fischer and Tropsch in the 1920.s
led to the realization that hydrocarbon chain formation proceeds via the stepwise addition
of one C-atom at the time. Detailed product analysis studies indicate that the reaction
produces a vast array of hydrocarbons and oxygenates over a large boiling range. The major
reaction products at high pressure operation are linear paraffins, linear 1-olefins, and lin-
ear 1-alcohols. At lower pressures, the selectivity for mono-methyl branched hydrocarbons
and internal-olefins increases. The formation of aromatics is only observed at higher tem-
peratures on Fe-based catalysts and is not observed on Co catalysts. The Fischer-Tropsch
reaction is represented in the following equations,
nCO +2n+1H2 ? CnH2n+2 +nH2O (10.1)
nCO +2nH2 ? CnH2n +nH2O (10.2)
Equation 10.1, relates to the production of paraffins and Equation 10.2 to that of olefins.
Alcoholic products can also be formed either as by-products or as main product depending
on the catalytically active metal and the pressure. A whole range of products with differ-
ent lengths of the carbon chain can be obtained. Two important side reactions generally
also proceed during the process they are the water gas shift conversion reaction and the
Boudouard reaction. The water gas shift conversion involves reaction of carbon monoxide
with water, which is formed during the reaction (Equation 10.1 and 10.2), to hydrogen and
carbon dioxide.
CO +H2O ? CO2 +H2 (10.3)
The water gas shift conversion can be favorable for the Fischer-Tropsch synthesis start-
ing with CO-rich synthesis gas as is obtained from gasification of coal or heavy oil fractions
134
through partial oxidation (H2/CO molar ratio approximately 1). The water gas shift conver-
sion raises the H2/CO ratio. In contrast, synthesis gas produced from natural gas initially
already possesses a high H2/CO ratio. Hence, the hydrogen-to-carbon monoxide ratio can
become undesirably elevated due to the water gas shift conversion. The second side reaction
mentioned above is the Boudouard reaction. This reaction occurs preferably on the active
metal surface and involves the conversion of carbon monoxide into carbon dioxide leaving
carbon behind on the surface (Equation 10.4)
2?CO ? CO2 +C (10.4)
The carbon deposition can eventually lead to deactivation by, for example, blocking of
active sites, disintegration of the catalyst bodies or plugging of the reactor by formation of
very strong carbon fibrils.
Anderson shows that a polymerization-like process effectively describes the product
distribution of the Fischer-Tropsch synthesis. This results in the so-called Anderson-Schulz-
Flory (ASF) product distribution:
Wn = n(1??)2?n?1 (10.5)
where Wn represents the weight fraction of all hydrocarbon products with carbon num-
ber n, and ? is chain growth probability, which is the rate of growth per the sum of growth
and termination rates [171]. Similar molecular weight distributions were already observed
during polycondensation by Schulz and free-radical polymerization by Flory [Hinderman et
al.., 1993]. The entire product spectrum is characterized by a single parameter, i.e. the chain
growth probability defined as:
? =
i=n+1summationdisplay
?
?i/
i=nsummationdisplay
?
?i = rpr
p +rt
(10.6)
135
where rp and rt are the rates of chain propagation and termination, respectively, and
?i is the mole fraction of the product spectrum containing i carbon atoms. Equation 10.5
implies that the Fischer-Tropsch reaction is not selective toward a single reaction product
or a specific carbon range, with methane as only exception. Methane can be produced
with 100 % selectivity. The Fischer-Tropsch synthesis, however, can selectively produce
one type of reaction product, i.e. the selectivity towards 1-olefins or paraffins can be opti-
mized. Wax formation is favored at low temperature, high pressure, low H2/CO feed ratios,
and large residence times. Mass transport issues have a strong influence on the degree to
which readsorption modifies the product distribution. [172] An illustration of the distribu-
tion of products is provided in Figure 10.1. The most important aspects for FT reactor
development are the high reaction heats and the large number of products (gas, liquid and
waxeous hydrocarbons). An optimal design of a commercial scale reactor requires detailed
information of the hydrodynamics, reaction kinetics, catalytic system and FT chemistry.
Kinetic information is crucial for reliable design and scale up of commercial Fischer-Tropsch
processes [173].
10.1.4 Catalysts for Fischer Tropsch synthesis
On investigating the Fischer Tropsch systems employed for catalytic synthesis of hydro-
carbons, it is clear that cobalt and iron are the most commonly used due to there overall
reactivity and cost [174] [175]. Iron catalysts for the Fischer-Tropsch synthesis generally
consist of precipitated iron, which is promoted with potassium and copper to obtain a high
activity and selectivity, and with Al2O3 and SiO2 added as structural stabilizers. Typically
for these relatively cheap Fe-based catalysts, Fe-carbide appears to be the active phase for
Fischer-Tropsch synthesis. Fe-oxides are also formed and these are active toward the water-
gas shift reaction. A high water-gas shift activity causes these catalysts to be flexible toward
the H2/CO feed ratio of the synthesis gas. This allows the utilization of a large variety of
feed stocks. Iron catalysts can use synthesis gas with a H2/CO ratio below 2, because excess
136
of CO is converted with water to carbon dioxide and hydrogen in the water gas shift (WGS)
reaction. However, the water-gas-shift activity of the catalyst also results in a low carbon
efficiency of the gas-to-liquid process.
Cobalt catalysts are usually supported on an inert support due to the higher cobalt price
and better catalyst stability. The active phase is metallic cobalt; the water-gas shift activity
of Co-based catalysts is low and water is the main oxygen containing reaction product.
Cobalt catalysts are suited for producing high yields of long-chain alkanes in FT synthesis.
Davis et al.. stated thatcobaltcatalysts exhibit two distinct advantages forsome applications
[176]. The first being a lack of the WGS activity allowing one to reject the oxygen in CO
as water rather than CO2 such that carbon efficiency of the cobalt catalyst is double the
efficiency of the iron catalyst at high conversion levels. The second advantage of the cobalt
catalyst is it can be added to a support that can provide robustness that iron catalyst does
not have in unsupported state. An inert support is a support that does not, by itself, react in
a Fischer-Tropsch synthesis, although it may interact with the catalytically active layer, and
the shape and configuration of the inert support may play an important role in controlling
the reaction. Examples of inert support materials include alumina and silica.
Cobalt was selected as the material of interest for this study. Preparation of cobalt
supported catalysts involves several essential steps. Catalysts are commonly synthesized
using aqueous impregnation with cobalt nitrate and promoters, followed by oxidative and
reductive pretreatments. The activity and selectivity of cobalt supported catalyst toward
different hydrocarbons is strongly dependent on preparation procedures [177] . These prepa-
ration procedures are known to contribute to the final properties associated with the active
sites of the catalyst. Literature shows that for cobalt particles larger than 7-8 nm, Fischer
Tropsch activity is a function of the number of surface metal sites [178]. Girardon has shown
that controlling catalyst calcination process can influence cobalt particle size, reducibility,
and catalytic properties [179].
137
It has been generally accepted that on cobalt catalysts the active phase for FT synthesis
is cobalt metal and that the activities of cobalt catalysts depend on the number of active sites
on the surface, cobalt particle size, and cobalt species [180]. The Fischer-Tropsch process is
critically dependent on the effectiveness of the cobalt catalyst and there has been a significant
effort in the optimization in the design of catalysts which optimize metal crystallite size,
control dispersion, improve reducibility and increase overall stability. The interaction of
cobalt with the metal oxide support and its dispersion characteristics have been shown to
affect the electronic density as well as the structure of the metal crystallites and are assumed
to play a role in the hydrocarbon selectivity and the catalyst activity.
Most of the industrial cobalt-based catalytic systems used in the FTS reaction are sup-
ported on Al2O3, TiO2 or SiO2 [181] [182]. Silica is commonly used as the support material
due to its high surface area, porosity, stability, weak metal-support interaction and its in-
ertness [183]. Iglesia et al. showed that FT synthesis rates were proportional to metal
dispersion and were independent of support [174]. Others have indicated that catalytic
supports significantly influence the extent of metal reduction, morphology, adsorption and
catalytic properties of the active phase especially in well dispersed catalytic systems [184].
The problem lies in the fact that synthesis of highly dispersed cobalt catalysts requires strong
interaction between the support and the cobalt precursor, but in turn, such strong interac-
tions generally lower reducibility of such precursors [185]. A weak cobalt support interaction
in silica-supported catalyst promotes high cobalt reducibility but, can favor agglomeration
of support particles [170].
The porous structure of catalytic supports can strongly influence the catalytic behavior
of many catalysts in a large number of reactions. Khodakov et al. showed that an increase in
the pore size of the supports lead to an increase in Co3O4 particle size and offers an improved
reducibility of the intermediary CoO particles to metallic cobalt [186]. Narrow pore cobalt
impregnated catalysts were shown to exhibit a decreased reducibility due to small cobalt
intermediates this affect was correlated to lower Fischer Tropsch reaction rates and higher
138
methane selectivities. Catalytic support structures can also contribute to catalytic poisoning
due to the nature of some of the Fischer Tropsch reaction products such as wax and carbon
deposits, which can block the catalyst pores making them unavailable for reaction molecules.
Given the complexities of using a cobalt catalyst much experimentation was performed on
the cobalt catalyst system and will be discussed further in 10.3.2
10.2 Tunable Fischer Tropsch Microreactor Realization
One of the fundamental goals for this project was the realization of a microreactor ca-
pable of performing the FT reaction, while being easily tunable for operating parameters
such as pressure gradients, temperature profiles, and reactor geometries. Various materials
have been explored for microreactors, e.g., stainless steel because of its robustness and good
thermal conductivity, ceramic materials for high temperature application, polymers for bi-
ological compatibility and silicon due to its mature fabrication infrastructure, low cost and
mass production capability. Our initial efforts focused on the fabrication of micro reactors
utilizing silicon wafers. Both wet chemical etching and plasma etching of single crystalline
silicon have proven suitable for mass fabrication of various micromechanical components. As
such, much information was available on the details of silicon processing. A draw back is
that the equipment required is relatively costly, and there is a need for processing within a
clean room. But, an advantage exists for batch processing because several structures can be
realized in parallel on one wafer. A variety of simple geometric structures like grooves, chan-
nels or membranes can be realized by lithographic and thin film processing on a silicon wafer.
Silicon micro machining also benefits from existing bonding processes, either thermally or
anodically, which can be used for assembly of multiple layers. We successfully fabricated
multiple reactor geometries in silicon. Subsequent sealing of the reactor also proved success-
ful via anodic bonding. The problems that we encountered dealt more with the attachment
of the interconnections to and from the reactor. We were unsuccessful in creating a suitable
connection that was gas tight. Various arrangements were tried and alternate connection
139
assemblies were manufactured. Other issues encountered included successful incorporation of
the catalytic support within the reactor. It proved very difficult to incorporate the membrane
uniformly within the reactor bed. Due to limited success we chose to investigate alternative
fabrication schemes and materials.
Metal microreactor
Realization of an all metal micro reactor presented an opportunity for a robust micro
reactor with the potential for gas tight connections, sustainability of high temperatures and
the ability of high pressure operation. This metal microreactor design also afforded a means
of successfully manipulating the micro reactor geometry to create pressure gradients and
temperature profiles. We manufactured the micro reactors utilizing metal sheets that could
be compression sealed to eliminate gasket materials that inevitable caused leaks. These
sheets could be manipulated by etching, cutting or stamping features into the metals.
The design and development of micro reactors have been driven in part by fabrication
constraints and innovations to overcome leaks. For our application specific attention has
to be applied to the interconnection techniques associated with gas phase reactions. It was
determined that we were in need of a reactor capable of sustaining 450 ?C for extended
periods of time and that seals needed to be leak tight, withstand hydrocarbon products and
maintain a max pressure of 80 psi. The other issues that needed to be addressed were the
ease in fabrication, the ability to manipulate flow path designs to test for pressure gradients
and temperature gradients, as well as create interconnections to the inlet gases and outlet
product gases.
A preliminary micro reactor was successfully fabricated and utilized in the work that
follows. A photograph of this microreactor is in Figure 10.22. This micro reactor is a simple
channel geometry that consists of a single groove within which the catalyst membrane can be
compression sealed. The pressure gradient across this reactor can be manipulated by chang-
ing the available volume within the membrane (compressing the membrane). Metal sheets
140
of differing thicknesses were utilized and examined in terms of pressure drop and residence
time distribution determination.which will ultimately control how much compression is ap-
plied to the membrane. In addition metal sheets were obtained with an alternative reactor
geometry which will modify the pressure gradient across the reactor while maintaining the
same reactor volume and catalyst loading. This is more of a cone type geometry and will be
illustrated in sections that follow.
10.3 Development of catalyst material for incorporation into the micro reactor
system
10.3.1 Micro fibrous Catalyst Support System (MCSS)
Although the small dimensions of the microreactor reduce thermal and mass gradients,
filling the microreactor with a catalyst of a reduced particle size can dramatically increase
the pressure drop in the reactor. For a given bed length, a packed-bed with 60 ?m catalyst
particles has a pressure drop ? 275X larger than a bed with 1 mm particles, and ? 27500X
larger than a bed with 1 cm particles. In studies with the moderately fast phosgene reaction
in the axial flow microreactor, gas flow rates of only 4.5 sccm with 53-71 ?m catalyst yielded
a large pressure drop of 0.4 atm [187]. In general, a large quantity of catalyst is desirable
for catalyst testing with microreactors to average out variances between catalyst particles.
However, shortening the length of the catalyst bed is necessary to reduce pressure drop. In
this work, we present a single channel microreactor for catalyst testing that uses a Micro fi-
brous Catalyst Support System which provides large void volume, an entirely open structure,
large surface-to-volume ratio, high permeability, high thermal conductivity, and unique form
factors instead of thin-films or coatings. This way, results are relevant to catalyst prepara-
tion techniques, catalyst-support interactions, and other issues in industrial catalysis. The
Micro fibrous Catalyst Support System enables the use of practical flow rates and catalyst
quantities while minimizing pressure drop. A comparison of the flow path of gases through a
141
microreactor with an MCSS catalyst vs a standard particle catalyst is illustrated in Figures
10.2 and 10.3. The microreactor design, fabrication, and characterization are presented.
Fabrication Methodology
The fabrication of the Micro fibrous Catalyst Support System was performed by a wet-
lay process. The base of the slurry is a cellulose pulp suspension. To this, an adjustable
volume percent of nickel microfibers (typical diameter 3.0 ?m) are added and mixed followed
by the desired volume percent ofthe silica support particles. This slurry is diluted and formed
into hand-sheets using a typical paper forming technique where the majority of the slurry
is poured over a fine screen allowing most of the free water to drain, leaving behind a dense
mesh of cellulose and nickel fibers entangling the silica support particles. The hand-sheet is
then air dried and oxidized in air at approximately 400 ?C to remove the cellulose, leaving
behind the nickel entrapped particles. A subsequent hydrogen reduction at 900 ?C sinter the
micro fibrous nickel network around the silica particles and produces a robust micro fibrous
network. The micro fibrous network can be tailored in both void age and support loading,
and can be mechanically formed post-fabrication by cutting, compressing or even welding.
The finished micro fibrous media has tunable transport properties including effective thermal
conductivity, mean distance between fibers and pressure drop. A SEM image of one example
of this material is shown in Figure 10.8
10.3.2 Qualification of Micro fibrous Catalyst Support System (MCSS)
The Fischer-Tropsch process is critically dependent on the effectiveness of the cobalt
catalyst and there has been a significant effort in the optimization in the design of catalysts
which optimize metal crystallite size, control dispersion, improve reducibility and increase
overall stability. The interaction of cobalt with the metal oxide support and its dispersion
characteristics have been shown to affect the electronic density as well as the structure of
the metal crystallites and are assumed to play a role in the hydrocarbon selectivity and
142
the catalyst activity. Due to the nature of our microreactor it was impossible to incorpo-
rate an existing catalyst into our system. Work was needed to identify an active material
that was capable of being utilized in a microreactor. Significant preliminary investigations
were required as it was necessary to start from scratch and examine all the steps in the
preparation procedure. This effort prompted a significant investigation of the impregnation
procedure, calcination procedure, and reduction procedures necessary to produce a viable
catalyst material. Reduction of supported Co3O4 species in hydrogen has been a subject of a
large number of investigations. Reduction of Co3O4 has been shown to produce CoO within
the temperature range of 200-300 ?C. The amount of metallic cobalt produced in the second
reduction step and the temperature range associated with the reduction is controlled by the
ease of CoO reduction to metallic Cobalt. A final qualification was performed by utilizing
our optimized catalyst material in an existing macro scale FT reactor.
Calcination and Reduction Mechanism for Cobalt Nitrate Hexahydrate
Preliminary analysis involved understanding the reaction mechanisms for the calcination
and reduction of cobalt nitrate hexahydrate. The reaction mechanisms as indicated below
were found in the literature review.
Calcination Mechanism
CoNO3 ?6(H2O) Air? 6H2O +NO +NO2 + 13Co3O4 (10.7)
Reduction Mechanism
Co3O4 +H2 ? H2O +3CoO(stepI) (10.8)
CoO +H2 ? H2O +Co(stepII) (10.9)
143
It was determined that due to the nature of both reactions that TGA-MS could be
utilized to examine the calcination and reduction mechanisms. Runs were performed on
cobalt nitrate hexahydrate to measure the extent of both calcination and reduction. TGA-
MS allowed for a measurement of weight loss as a function of reaction temperature. From
this analysis we were able to determine the extent of calcination and reduction and identify
the temperature at which these events occurred. The calcination results are shown in Figure
10.4. The results were in good agreement with the values calculated based on stoichiometry.
The reduction mechanism as shown in Figure 10.5 are also in agreement with the calculated
values. Two distinct peaks were identifiable for the reduction mechanism which allowed for a
correlation of reduction amount and reduction temperatures. It can be seen from Figure 10.5
that step 1 occurs at 283 ?C and step 2 of the reduction occurs between 320-400 ?C The mass
spectral data allowed identification of the evolved gases as the reactions proceeded. These
gases were in agreement with the mechanisms proposed in the literature. Quantitatively,
the mass losses at steps I and II are 6.9% and 20.7% respectively, while the mass loss at
step II is approximately three times that at step I. This is in agreement with the reduction
mechanism shown above. The water evolved from the reduction steps trends with the weight
loss events and validates that the weight loss is due to the reduction mechanism proposed.
Investigation of the Reduction Properties for Silica Supported Cobalt as Influ-
enced by Pore Size
Initial investigations focused on the utilization of different silica support particles. As
the interaction with the metal oxide support and its dispersion characteristics are of great
importance it was necessary to understand the interactions that occurred between the silica
support and the cobalt precursor. Co/SiO2 catalysts were prepared by aqueous cobalt nitrate
impregnations of silica with different pore sizes to study the effect of the pore size on the
reduction profile. Cobalt nitrate hexahydrate impregnated silica catalysts were calcinated in
air at 350 C for 2 hours. Investigations revealed this calcination procedure regardless of pore
144
size consistently provided a high degree of conversion to Co3O4. An example is illustrated
in Figure 10.6. Four different sized silica powders were utilized for impregnation of a cobalt
loading of 20 wt%. The sample information and identifications are given as follows:
1. Silica Gel 60A Pore Size, 43-63 micron overall size
2. Silica Gel 150A Pore Size, 150-250 micron overall size
3. Silica Gel 250A Pore Size, 60-250 microns overall size
4. Silica Gel 150A Pore Size 75-150 micron overall size
Figure 10.7 represents the weight loss associated with the four catalysts of different
silica pore size. Again, two distinct weight loss events associated with Co3O4 reduction are
illustrated. The results showed that the reduction proceeded in a two step process with
the first reduction step occurring between 200 ?C and 320 ?C. All of the catalysts showed
similar reduction percentages for the first step of reduction. This was as expected as the
first reduction event is expected to occur easily. The second reduction step occurred between
320 ?C and 700 ?C depending on the pore size. Quantitatively, the mass loss at step II is
expected to be approximately three times that at step I to be in agreement with the reduction
mechanism shown above. It became evident during the investigation, that the catalysts with
the larger pore sizes were more easily reducible to metallic cobalt. Based on the calculations
full reduction weight loss is calculated to be 6.8%. The catalyst with the smallest pore size
6 nm showed the least amount of reduction at 5.0%. Sample 4 was choosen as our starting
material given the extent of reduction and the particle size range. Future analysis will include
a more in depth investigation of particle size as influenced by our reduction procedure.
Impregnation Procedure for MCSS
The procedure consisted of creating the nickel micro fibrous silica support via a standard
wet paper-lay process. A micro fibrous carrier consisting of a 2.0-3.0 vol. % of 3 ?m (dia)
145
Ni fibers was utilized to entrap 20-30 vol. % of 75-150 ?m diameter silica particulates with
a pore size of 6nm. This process was followed by an oxidative treatment at a temperature
of 400 ?C to remove binders associated with the paper lay process and was followed by a
hydrogen reduction at 900 ?C that allowed the micro fibrous nickel network to be sintered
together around the silica powder to form a more rigorous support. The sintered support
was solution impregnated with a 15 wt.% cobalt nitrate hexahydrate and water solution.
After completely immersing the sintered support in the 15 wt% cobalt nitrate hexahydrate
solution, excess solution was removed via blotting relying on capillary forces to remove the
excess material that was not incorporated within the silica pores. The material was again
subjected to an oxidative treatment as previously described that removed the nitrate portion
of the cobalt precursor and produced a cobalt oxide that could be subsequently reduced to
the active metal prior to its utilization as a FT catalyst.
Catalyst Characterization
SEM analysis was performed to characterize the surface features of the impregnated
MCSS. Figure 10.8 on the right is a SEM image of the fabricated micro fibrous support
just prior to impregnation. It can be seen that the silica particulates are of various shapes
and are anchored within the nickel matrix. Figure 10.8 on the left is a SEM image of an
individual silica particle used for comparison purposes. Figure 10.9 right is an SEM image
of the micro fibrous entrapped silica supported cobalt catalyst with 75-150 ?m silica pore
size prior to reduction. It is interesting to note, that the cobalt catalyst which was expected
to be only associated with the silica support was actually distributed uniformly throughout
the micro fibrous matrix. An individual particle was investigated in Figure 10.9 and shows
that the cobalt catalyst also coated the surface of the particles.
The fact that the cobalt catalyst was distributed non uniformly throughout the micro
fibrous matrix was an intriguing result. But, due to this fact there were questions on the
actual loading percentages associated with the cobalt. The loading that was assumed was
146
not accurate. It has been stated in the literature that when doing solution impregnation
the active catalyst will diffuse into the pores of the silica support and remain. Subsequent
reduction would reduce the cobalt oxide to active metal cobalt within the pores. Given
the interaction of the impregnating solution and the support an alternative impregnation
method was necessary.
Additional catalyst samples were produced that utilized a more rigorous attempt to
remove the excess catalytic solution. These attempt included a vacuum impregnation step.
The catalyst was exposed to excess solution of the cobalt nitrate solution previously de-
scribed. The catalyst membrane was placed securely onto a vacuum plate and excess solvent
was removed via vacuum. This procedure was repeated 3 times and the weight gain achieved
was in agreement with what was calculated after an overnight drying procedure at 90 ?C.
This procedure was adopted as the standard procedure for all subsequent MCSS fabrication.
Additional analysis utilizing optical microscopy confirmed that the impregnation solution
was no longer present in the network and only accumulated on the silica particle surface and
on in its? pores.
10.3.3 Catalyst Support Manufacturing Procedure
The procedure consisted of creating the nickel micro fibrous silica support via a standard
wet paper-lay process. A micro fibrous carrier consisting of a 2.0-3.0 vol. percent of 3
?m (dia) Ni fibers was utilized to entrap 20-30 vol. percent of 75-150 ?m diameter silica
particulates with a pore size of 6nm. This process was followed by an oxidative treatment
at a temperature of 400 ?C to remove binders associated with the paper lay process and was
followed by a hydrogen reduction at 900 ?C that allowed the micro fibrous nickel network to
be sintered together around the silica powder to form a more rigorous support. The sintered
support was solution impregnated with a 15 wt% percent cobalt nitrate hexahydrate and
water solution. After completely immersing the sintered support in the 15 wt% percent
cobalt nitrate hexahydrate solution, excess solution was removed via vacuum filtration to
147
remove the excess material that was not incorporated within the silica pores. This procedure
was repeated 3 times and the support was dried in an oven at 90 ?C overnight. The material
was again subjected to an oxidative treatment that removed the nitrate portion of the cobalt
precursor and produced a cobalt oxide that could be subsequently reduced to the active metal
prior to its utilization as a FT catalyst.
10.3.4 Utilization of Catalyst Support in Standard FT Reactor
Initial experiments were performed to determine the applicability of the MCSS catalyst
for Fischer Tropsch synthesis. A preliminary catalyst was prepared following the procedure
detailed inSection 10.3.3andwas incorporatedinto an existing macroscale gasphase reactor.
A standard sample was also manufactured that consisted of a nickel micro fibrous silica
support that was void of the reactive catalyst, which was analyzed in order to determine the
reactivity ofthe support material. The procedure used to produce this standard was the same
as listed in Section 10.3.3, but without the subsequent cobalt nitrate loading. It should be
noted that this catalyst is being testing in a high temperature reactor that has proven to be
repeatable successful in prior runs. The macro scale experiment utilized the catalyst powder
4 (15nm Pore size,100nm Avg particle size). This material showed promise as a potential
Fischer Tropsch catalyst. It is important to note that the information learned from these
experiments will not be directly transferable to our microreactor due to the different reaction
pressures utilized for our system.
Two runs were made on the Wilmore 202 gas phase reactor utilizing the micro-fibrous
material for FTS. The first was a blank run without the reactive catalyst. The second used
a 15 % Cobalt loading on a silica supported nickel mesh. The Fischer-Tropsch Synthesis
reaction was performed in a stainless-steel reactor (di-.5 inch, l = 2in). The reactor was
loaded with 1 g of catalyst (1.7 g of overall micro-fibrous material). Prior to the catalytic
experiments, the catalysts were reduced with hydrogen in situ at 300 ?C. The reduction time
was 4 hours. After the reduction step, the temperature was lowered to 250 ?C under H2
148
and the reactant gas mixture (H2:CO:N2, v/v 63:33:1, N2 was the internal standard) was
introduced at a total flow rate of 50 SCCM). The pressure was slowly increased to 250psi.
Once the reaction temperature was achieved the reaction was led to proceed during a 24
hour period. Periodic sampling and temperature adjustments were made during this period
to monitor there effects on catalytic activity. During reaction, the reactor e?uent passed
through a cold trap at 10 ?C to collect both lighter products (water, alcohol, and hydrocar-
bons) and wax products. The e?uent gases were depressurized and analyzed periodically by
on-line gas chromatography.
A Varian 3800 gas chromatographic system equipped with a TCD detector was devel-
oped for the rapid on-line analysis of light Fischer-Tropsch products. A Haycep DB100/120
(part number 2836PC) packed column separates the most common C1-C5 hydrocarbons.
The packed column is used for the separation of methane, carbon monoxide, carbon dioxide,
water and methanol. Retention characteristics for the analysis on the packed column are well
characterized. The total analysis cycle is approximately 40 minutes. The heavier Fischer-
Tropsch products are collected in a cold trap and are analyzed off-line. These products are
injected into a liquid chromatograph Varian 3300 with a capillary tube (DB-5/ Part number
125-5032) that can resolve hydrocarbons from C6-C30 utilizing a FID detector.
For the first experiment no liquid product was present in the cold trap. Analysis showed
that minimal activity was associated with the nickel-based micro fibrous material void of
catalyst indicating that any activity associated with the catalyst enhanced micro fibrous
material would be due to the cobalt catalyst. The second experiment was performed with
the 15 % Cobalt loading on a silica supported nickel mesh. The data in table 10.1 show a
combination of results for the combination of the gas (analyzed via the TCD) and the liquid
products collected (analyzed via the FID). The CH4 and CO2 selectivities are high, but that
is to be expected with gas phase operation. Some interesting observations are that the olefin
content of the liquid product was negligible and that the alpha value was quite high (79%),
149
although it should be noted that this owes most to the low temperature period. As such,
this material shows considerable promise for gas-phase Fischer Tropsch synthesis.
150
Date Reaction Temp (?C) CO Conv H2 Conv CH4 Sel. CO2 Sel.
2/08/2008 214?C 100 94 58 24
2/08/2008 214?C 100 93 58 25
2/08/2008 204?C 100 90 56 26
2/08/2008 195?C 99 93 39 20
2/08/2008 194?C 40 49 13 1
2/08/2008 204?C 100 90 42 37
Table 10.1: Macro scale conversion data for cobalt impregnated silica supported nickel cat-
alyst.
10.4 Systematic Probing of the Operating Parameters
The initial goals of this project include a complete analysis of the operating parameters
associated with our designed microreactor. This included an analysis of the pressure drop
through the reactor as influenced by the MCSS compression. Calculations were performed
to characterize the flow characteristics within the reactor. Extended analysis included an
analysis of the residence time distribution as influenced by compression.
10.4.1 Phase 1. Investigation of the Effect of Pressure Drop
- Investigate Pressure Drop as a function of (MCSS) compression
- Characterize the Residence Time Distribution as a function of compression
One of the main problems in using Micro reactors for heterogeneously catalyzed gas-
phase reactions is the introduction of the catalyst in the reaction zone. The straightforward
way is to fill the micro channels with catalyst powder. The size of the particles introduced
generally ranges between 35 and 75 ?m in diameter, leading to a high pressure drop during
the passage of fluid. In addition, each channel must be packed identically to avoid mis
distribution, which is known to broaden residence time distribution while diminishing reactor
performance. Another approach is create micro reactors with catalytically-active walls. The
specific surfaceareais increased by chemical treatment ofthe channel walls orby theircoating
with a porous layer. The porous layer can serve as a catalyst or a support for a catalytic
151
phase. The main limitation of catalytic-wall micro reactors is the thickness of the porous
layer. Since the majority of micro reactors are used for fast, highly exothermic reactions, the
layer should be < 1-2 ?m in order to avoid mass/heat transfer limitations. Therefore, the
total mass of the catalyst is too small for achieving high performance preferred for a unit of
the reactor volume. In our micro reactor a micro fibrous catalyst support system (MCSS)
with a structured catalytic bed consisting of entrapped particles in a sintered metal network
whose void volume can be varied. Each (MCSS) possesses a 3-dimensional regular micro
structure, resulting in a low pressure drop during the passage of reacting gases at higher
void age. Our initial system investigations have included pressure drop investigation as a
function of MCSS compression.
Pressure Drop Investigation Each of these experiments utilized the standard MCSS cat-
alyst that incorporated the 15nm pore size (75-150?m)particle size silica. These experiments
were performed without the addition of the catalyst and represent the effects of solely the
support system. The calculated ratio by weight for the silica to nickel fibers is 37.5 wt%
Nickel / 62.5 wt% Si. Three different arrangements of the MCSS micro reactor were tested.
(1) With compression to 1016 ?m and (2) with compression to 508 ?m and (3) with compres-
sion to 127 ?m. It should be noted that compression to the 127 micron completely blocked
flow and was not included in the analysis. The pressure drop across the microreactor was
measured with a differential pressure gauge placed at the reactor inlet and outlet. Calcu-
lations utilizing a modified reynolds number as suggested by the Taterchuk group revealed
that the flow is in the laminar flow regime regardless of compression or void age these results
are shown in Figure 10.12. This calculation is elaborated on in the powerpoint presentation
and will be subsequently included.
A linear dependence on flow rate and pressure drop for both compressions has been
found. The results were in good agreement with Darcy?s Law and are presented in Figure
10.11. Darcy?s law (Equation 10.10) for the pressure drop in a laminar flow was fitted to the
152
experimental data by a least-squares method,
?P
L =
?Q
kA
bracketleftbiggPa
m
bracketrightbigg
(10.10)
With the surface, A= f(compression), L= .0381m, ? = 1.98E-05 Pa/sec,P = Pa
The MCSS membranes consisted of identical loadings of silica at a 62.5 wt% loading.
The difference in pressure drop caused by compression was very significant and is shown to
be at least three orders of magnitude greater with a doubling of compression. This indicated
that compression could significantly influence the residence distribution time by forcing a
higher velocity to overcome the pressure differences.
10.4.2 Residence Time of Distribution
The RTD was obtained from measured response curves to the step change in the re-
actor inlet from flow of He to O2. The responses were first obtained for the installation
only (without reactor). The measurements were then done with the empty reactor without
MCSS. Finally, reactor + MCSS were characterized. The F[t] function was obtained after
a normalization of the concentration response. These three functions, F[t], are shown in
Figure 10.13 and Figure 10.14.
The axial dispersion in the micro reactor is however more visible when the function,
F[t], is transformed into a function, E(t), as presented in Figures 10.15 and 10.16. The E(t)
curve was calculated from the following equation:
E(t) = dFdt (10.11)
For the case indicated in Figures 10.13 and 10.14 at a flow rate of 5.0 ml/min the MCSS
with the lower compression (1016 ?m) increased its residence time only slightly over the
empty reactor. But, upon compression to the 508 ?m range the residence time for the gas
in the MCSS was much greater than the residence time of the empty reactor.
153
Further studies were performed with a variable flow rates as illustrated in Figures 10.17
and 10.19. Figure 10.18 represents the data obtained from the 1016 ?m compression at
varying flow rates from 5.0 ml/min to 25.0 ml/min. There is a transition between 5.0 ml/min
and 25.0 ml/min where the residence time is less affected by the MCSS and it appears that
the effect of frictional drag becomes less inhibitive of flow or the compressibility of the gas
starts to become more of an issue. Based on the information shown in Figure 10.18 as the
flow rate is increased the dispersion becomes decreased at this compression.
Figure 10.19 represents the data obtained from the 508 ?m compression at varying flow
rates from 5.0 ml/min to 25.0 ml/min. For this compression between 5.0 ml/min and 25.0
ml/min the residence time is greatly influenced by the MCSS and it appears that the effect
of frictional drag becomes more of an issue. Based on the information shown is Figure 10.20
as the flow rate is increased the dispersion becomes decreased at this compression but to a
much lesser extent than that at the lower compression.
Data has been collected for the 1016 compression as a function of temperature and
appears to trend as expected. The RTD shown at 5.0 ml/min show that the distribution
becomes more narrow as a function of increased temperature and is shown in Figure 10.21.
Work has not completed on the lower compression material as a function of temperature.
There was a drastic differences in residence time based on compressibility coupled with
the pressure drop.
154
10.5 Microreactor Utilization for Fischer Tropsch Synthesis
As directed by committee during the preliminary exam, the focus of this work became
limited. Suggestions were to discontinue the pressure drop measurements, RTD studies, and
catalyst investigations shifting focus to simple experimental observations. Namely,
? Goal 1. Determine if the System Works.
? Experiments suggested:
? Utilization of only one catalyst loading (same calcination procedure as with the
macro reactor testing)
? Limit catalyst compression to one setting (1014 ?m)
? Measure product distribution as a F(reactor temperature) and F(flow rates) to
determine system limitations and outcomes.
? Goal 2. Repeat experiments with a different bed geometry and compare selectivities.
10.6 Results and Conclusions
Experiments were carried out in the microreactor system as illustrated in Figure 10.22
under isothermal conditions. The micro fibrous catalyst support system (MCSS) is ?sand-
wiched? by two separated heat transfer channels, which are designed to maintain a high
heat transfer coefficient. The MCSS described above is snugly inserted in the microreactor
providing the total length of the catalyst bed of 2.5 inches. The reactants can be preheated
to a desired temperature in the upstream portion of the channel before entering the catalyst
bed. The individual channel of the microreactor has the dimension of: 2.5? (x-direction)
X variable (y-dir) X .25? (z-dir). The channel is filled with the MCSS and there is no gap
between the catalyst pieces and the reactor walls. A thermocouple is placed on the outlet
side just above the MCSS.
155
To begin the catalyst was activated in a 3% hydrogen / helium blend at 350 ?C in situ
overnight, followed by the introduction of the syngas feed with H2/CO ratio of 2. Before
measurements were performed the system was equilibrated for 24-hours at a given flow
rate. Initial settings ranged from 5.0 ml/min to 0.1 ml/min begining with the highest
settings and gradually stepped down to the lower values. From this initial test- a set flow
rate of 0.1 ml/min was required to obtain measurable conversion. Noncondensed gases are
analyzed using an on-line gas chromatography to determine CO conversion and light product
selectivity. All process parameters such as temperature, pressure, flowrates were recorded.
At this point in time selectivity was only for CO2 and methane with limited conver-
sion. While maintaining the 0.1 ml/min flow rate experiments were performed with 24 hour
equilibration with a reduction in temperature at 10 ?Cintervals. Once the temperature was
reduced to 280 ?C measurable conversion was not observed. From this information it was
determined that the catalyst was not active and the conversion noted was due to thermal
effects.
A new catalyst material was reduced ex-situ in the 3% hydrogen / helium blend at
350 ?C at a higher flow rate of 100 ml/min (consistent with TGA analysis) and introduced
into the microreactor system in an attempt to make the catalyst more reactive. Again, the
catalyst was treated in a 3% hydrogen / helium blend at 350 ?C overnight at 5.0 ml/min and
followed by the introduction of the syngas feed with H2/CO ratio of 2. Before measurements
were performed the system was equilibrated for 24 hours at a given flow rate. Initial settings
ranged from 5.0 ml/min to 0.1 ml/min as the experiment before begining with the highest
settings and gradually stepped down to the lower values. From this initial test- a set flow rate
of 0.1 to 2.0 ml/min was required to obtain measurable conversion. The lower flow rates gave
higher conversions and better selectivities toward C4 products as measured with GC. The
best conversion values were obtained at 280 ?C a temperature which was much higher than
what was expected. This high temperature could be indicative of reduced thermal contact
with the MCSS membrane. This microreactor system is a low pressure system relying on only
156
the pressure drop accross the reactor to aid in conversion. Results obtained for conversion
were very poor and no products above C5 were observed.
To close the loop on the experiments a different reactor geometry was analyzed (cone
type) with a higher built in pressure drop. As no RTD distribution or pressure drop mea-
surements were performed on this reactor it is difficult to determine the true influence of this
geometry. The experiments were repeated with the ex?situ reduction of the catalyst. The
catalyst amount was the same by mass even though a different geometry was placed into the
reactor. The same experimental scheme as mentioned previously was performed stepping
down flow rates and temperatures for comparative purposes. The results of the experiments
were similar to what was observed previously and not direct trend data was obtained.
Work was discontinued on this project.
157
Figure 10.1: Typical Fischer Tropsch product distribution based on chain growth probability.
158
Figure 10.2: Representation of gas flow through a micro fibrous catalytic support system.
Figure 10.3: Representation of gas flow through a standard packed bed reactor.
159
Figure 10.4: Calcination of unsupported cobalt nitrate powder in 100 ml/min air and a ramp
rate of 10 ?C/min.
Figure 10.5: Reduction of unsupported Co3O4 with a 100 ml/min 3% H2/ Helium flow and
ramp rate of 10 ?C/min.
160
Figure 10.6: Calcination of silica-supported cobalt nitrate powder in 100 ml/min Air and a
ramp rate of 10 ?C/min.
Figure 10.7: Water evolution during the reduction of Co3O4 with a 100 ml/min 3% H2/
Helium flow and ramp rate of 10 ?C/min as a function of pore size.
161
Figure 10.8: [Left] MCSS Sheet prior to solution impregnation. [Right] Individual particle
from MCSS Sheet.
Figure 10.9: [Left] MCSS Sheet that has been impregnated and dried by blotting method
and subsequently calcinated [Right] Individual particle after calcination.
162
Figure 10.10: Illustration of the microreactor system fabricated for this work.
Figure 10.11: Measured pressure drop across the microreactor as a function of compression
as fitted by Darcy?s law.
163
Figure 10.12: Reynolds number calculations in a MCSS as a function of variable compression,
void volume, and volumetric flow rate.
164
Figure 10.13: F(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 1016 ?m
compression.
165
Figure 10.14: F(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 508 ?m
compression.
Figure 10.15: E(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 1016 ?m
compression.
166
Figure 10.16: E(t) Break-through curve He ? O2 at 5.0 ml/min, T= 298 K at 1016 ?m
compression.
Figure 10.17: F(t) Break-through curve He ? O2 at variable flow rates, T= 298 K at 1016
?m compression.
167
Figure 10.18: E(t) Break-through curve He ? O2 at variable flow rates, T= 298K at 1016
?m compression.
Figure 10.19: F(t) Break-through curve He ? O2 at variable flow rates, T= 298K at 508
?mcompression.
168
Figure 10.20: E(t) Break-through curve He ? O2 at variable flow rates, T= 298K at 508
?mcompression.
Figure 10.21: Residence time distribution as a function of temperature at a set flow rate of
5.0 ml/min.
169
Figure 10.22: Photograph of microreactor fabricated for this work.
170
Part III
Interfacial Engineering of Microfluidic Devices
171
Chapter 11
Surface Engineering of Poly(dimethylsiloxane) Microfluidic Devices
11.1 Motivation
Polydimethylsiloxane (PDMS) has become a material of choice for the fabrication of mi-
crofluidic devices. It is inert, bio-compatible and maintains a transparency that has proven
convenient for visualization and detection. The flexibility of PDMS is suitable for the inte-
gration of flow control devices such as, micro valves and micro pumps, that can be easily
incorporated due to PDMS ability to be manipulated via soft lithography [188]. Despite
its advantages, the use of PDMS is limited because of poor control of its surface proper-
ties. Surface chemistry is of great importance in microfluidic devices due to the high surface
area-to-volume ratio. There has been much research dedicated to modification of the sur-
face properties of PDMS to overcome the limitations associated with microfluidic devices,
such as; protein absorption, solvent incompatibility, and surface wet ability. Water vapor
diffusion in PDMS also limit many applications that require increased temperatures and is
particularly important as related to PCR Applications [189]. It is the limitation associated
with temperature cycling as related to PCR investigations that prompted this work.
A Molecular Vapor Deposition (MVD) silica layer has been shown to deposit onto
PDMS surfaces resulting in a stable reactive surface capable of sustaining much of the same
chemistries associated with glass surfaces [190]. Due to its low-temperature deposition con-
dition, controlled coverage and process tunability it is expected that this silica layer could
also be capable of improving the water barrier properties associated with PDMS microfluidic
devices. Investigations are performed to characterize the water diffusion as influenced by an
added silica layer. Prior to this investigation, PDMS devices are custom fabricated and a
172
baseline testing method was established for water diffusion measurements. In addition, a de-
tailed investigation of water vapor diffusion as influenced by curing conditions is illustrated
as this information proved vital and unattainable current literature.
11.2 Background Research
There has been significant research aimed at miniaturized and fully integrated lab-
on-chip systems for applications, such as drug screening, cellular assays, protein analysis,
and point-of-care diagnostics. PDMS an inexpensive elastomeric polymer has emerged as a
promising material for lab-on-chip applications [191]. The ease in fabrication with both rapid
prototyping and mass production techniques as well as lower cost relative to silicon and glass
makes it particularly attractive for the development of microfluidic applications [192]. These
systems offer a drastically reduced liquid sample and reagent volume, and an increase the
speed of analysis [192,193]. In addition, these miniaturized lab-on-chip systems, also known
as micro-fluidic devices, offer the potential for parallel analysis of multiple samples on a single
chip. Within these devices there must be precise micro-fluidic control via the integration
of pumps, valves, mixers, separation units, reactors and detectors. Recent developments in
miniaturization technology have already had a great impact on the biomedical research field.
The MEMS-based polymerase chain reaction (PCR) chamber is one of the great progresses.
PCR is awell-characterized method forthe selectively identical replication ofDNAmolecules.
By an enzymatic amplification process, the concentration of a DNA species is nearly doubled
through a cycling between three different temperatures. In this way, the DNA concentration
can be multiplied more than a million times by about 30 cycles of temperature.
PDMS is a silicon-based organic polymer consisting of repeated [SiO(CH3)2] units. The
PDMS used for our micro structure application is a RTV-615 silicone rubber, (i.e. Room
Temperature Vulcanizing) silicone composed of two components: a base and a curing agent.
Upon curing, the siloxane base oligomers containing vinyl groups are linked to the hydrogen
173
containing curing agent oligomers through a hydrosilylation reaction, forming Si-CH2-CH2-
Si linkages. The curing agent contains a platinum-based catalyst that is needed to catalyze
the hydrosilylation reaction. As seen from the reaction diagram in Figure 11.1, the multiple
reaction sites on both the base and curing agent oligomers allows for 3-D cross linking.
The ability of PDMS to serve as a useful substrate for microfluidic fabrication extends
from advantages such as high gas permeability, good optical transparency, mold ability,
non toxicity or bio compatibility, low curing temperature, and ease of sealing with other
materials [194]. It can be bonded to a variety of materials including glass, quartz, silicon,
polystyrene, and polyethylene after oxygen plasma oxidation [195]. Although PDMS has
many advantages, there are some physical properties that limit its broad application. PDMS
is hydrophobic, and small hydrophobic molecules, biopolymers, and cells can all irreversibly
adsorb to the surface [196]. Water vapor diffusion in PDMS has also been shown to limit
many applications that require increased temperatures.
There have been numerous investigations of water vapor diffusion in PDMS as related
to membrane applications. The techniques involved have ranged from simple gravimetric
techniques, quartz crystal micro balance measurements, spectroscopic systems based on va-
por phase absorption of IR, and various iterations and improvements on vapor permeation
cells [197,198]. Recognition of the problem of water vapor diffusion as applied to microfluidic
devices has prompted some studies on ways to eliminate or reduce the diffusion of water.
Lounaci et al. placed a PDMS microdevice in a controlled humidity chamber to control dehy-
dration by manipulating the partialpressure ofwater around the microdevice thus controlling
the driving force [199]. Other researchers have investigated techniques to deposit either inter-
nal barrier layers or external barrier layers. These investigations have incorporated solution
coating methods, sol-gel techniques and CVD deposition of various materials ranging from
oxide layers to different polymeric materials to produce barrier layers [189,194,200?202].
Although these investigations have provided some improvement, microfludic devices are still
174
limited to applications that require reduced temperatures or are designed in such a way that
water vapor diffusion is controlled.
11.3 Investigation of Water Vapor Diffusivity in PDMS Devices as Influenced
by Curing Conditions
One production step often overlooked and under-examined is the choice of the curing
scheme as related to vapor diffusion properties within a microfluidic device. Despite the
manufacturers? recommended curing time for PDMS, many unique protocols exist for the
production of microfluidic devices. Many of the protocols have focused on ways of optimizing
valving processes, modifying overall microfluidic device reactivity, or on improving inherent
bonding difficulties in PDMS [203]. In fact, there are not many research groups who perform
the same manufacturing protocol for microfluidic devices. While many curing techniques
have been optimized and widely studied, it is hard to ascertain if one method has an ad-
vantage over another in terms of vapor diffusion properties. For example, the recommended
cure time for RTV 615 is 24 hours at room temperature. But research indicates that PDMS
curing can be accelerated by applications of heat for short periods and that to much heat
will create a stiff PDMS structure which can prove detrimental to pumping processes [204].
It has also been shown that PDMS can be made more stable by thermally aging microfluidic
devices [205]. Because chemical cross linking can alter the composition of a polymer, and
can consequently increase or decrease molecular permeability, the choice of a curing regimen
could drastically influence the cross linking in PDMS and have strong implications toward
water vapor diffusivity [206].
This research focuses on a simple gravimetric approach that utilizes TGA-MS to inves-
tigate the steady state diffusion of water vapor from within a PDMS device as a function
temperature. PDMS test devices are created by utilizing two different curing mechanisms.
The first being a 24-hr room temperature cure as recommended by the manufacturer and
a second that relies on a brief exposure (1hr to 100 ?C) followed by a 24 hour exposure to
175
room temperature. The intent of this work is to measure and identify any differences in the
steady state diffusivity measurements of water vapor as a influence by curing conditions.
The choice of temperature cycling ranges from ambient to 75 ?C(lowest maximum) to 95
?C(highest maximum) both to expedite the measurement process and to mimic the condi-
tions encountered in PCR measurements. This work was necessary to establish a baseline for
diffusivity measurements as influenced by curing conditions and provides a means to decou-
ple any changes in vapor diffusion as influenced by barrier layers and the various techniques
utilized to incorporate the barrier layers. This work was necessary to provide a baseline for
diffusivity measurements and provides a means to monitor our subsequent investigations on
surface modification which will be explained in more detail in a later section.
11.4 Experimental Details
11.4.1 Materials
Silicon wafers, p-type, with (100) orientation were obtained from University Wafer
(Boston, MA). GE Silicon RTV 615a and RTV 615b were purchased from Allied Electronics
and used as received. Deuterated Water was purchased from Acros Organics packaged in
0.75 ml ampules with a molecular weight of 20.02 g/mol. Each ampule was opened just prior
to use and the D2O was utilized within 10 minutes of opening.
11.4.2 Wafer preparation and cleaning
Both silicon wafers and diced silicon samples are used in this work. Silicon samples are
prepared by cutting the silicon wafers into 8mm ? 8mm squares with a dicing saw. Silicon
wafers anddiced silicon aresonicated in acetone forten minutes andthen againin isopropanol
forten minutes andthen dried under astream ofnitrogen. Samples areetched forten minutes
in concentrated HF to remove the native oxide layer, rinsed in copious amounts of deionized
water, and then dried under a stream of nitrogen. Samples are then loaded into a custom-
built vacuum deposition system, the design of which is based on a previously described
176
system [88], for oxygen plasma treatment. Briefly, the vacuum deposition system consists
of a rotary vane pump (base pressure 0.4 Pa) which is coupled to a glass reaction chamber
that contains perforated electrodes which can be biased to create an in situ, capacitively-
coupled radio frequency (RF) plasma (13.56 MHz). MKS Baratron capacitance manometers
are used to monitor the system pressure up to 1.3?104 Pa. After introducing the samples
into the vacuum system, the system is evacuated to a pressure of less than 2.6 Pa. Oxygen
gas is then allowed to flow through the system, and an oxygen background is established
by multiple pump-purge cycles with oxygen gas. Then, the oxygen pressure is allowed to
stabilize around 33.3 Pa, at which point the chamber is isolated from the pump. An RF
plasma is struck at 50 W forward power, and the samples are exposed to the plasma for five
minutes. This treatment grows an oxide layer about 2 nm thick on the silicon surface. The
samples are removed from the vacuum deposition system and etched again for ten minutes
in concentrated HF, rinsed in copious amounts of deionized water, and dried under a stream
of nitrogen. An additional oxygen plasma treatment is employed to re-grow an oxide layer.
Treating the samples by this method ?peels off? the uppermost layers of the crystal surface,
revealing a clean, flat surface. This iterative cleaning method is repeated until inspection
by contact angle and AFM indicate that a clean (contact angle < 5?), flat (RMS roughness
?0.2 nm) surface has been obtained, but usually two iterations suffice.
11.4.3 Teflon Mold Fabrication
A teflon mold was fabricated and incorporated onto a silicon wafer. This mold consisted
of a 4.0 inch diameter machined teflon disc (1/4?thick) consisting of multiple holes with an
outer hole diameter of 1/4?. The teflon ring was attached to the silicon wafer surface and
an inner teflon cylinder with and outer diameter of 1/8 ? was centered in each of the holes.
The teflon disc and center cylinders were secured by applying a light layer of PDMS to both
the disc and the cylinders. An illustration of the mold is shown in Figure 11.2. The mold
177
and silica unit was inserted into an aluminum pan for containment and placed in the furnace
at 100 ?C for 1-hour prior to utilization.
11.4.4 PDMS Preparation
GE RTV 615 Parts (a and b) were mixed according to the manufactures? instructions in
a ratio of 10:1 by weight. The components were stirred for 3 minutes and then the mixture
was pored over the teflon mold described in 11.2. Enough PDMS was applied to ensure that
the entire mold would be completely submerged within the PDMS solution. The PDMS was
also poured onto into an ZnSe ATR accessory to be utilized for FTIR analysis. The PDMS
filled mold and ATR accessory were placed in the vacuum chamber previously described in
section 11.4.2. The PDMS mixture was degassed under vacuum until the pressure was less
than 10 mtorr.
Clean diced silicon squares were placed over each of the holes on the mold. Pressure
was applied until the silica wafer laid flat against the mold. The silica wafer was initially
applied at an angle which ensured that there would be no air entrapment between the silica
and the diffusion cell. The PDMS mold structure and ATR accessory were cured either by
placing in an oven for 1hr at (100 ?C) and followed by room temperature for 24 hours or
by leaving at room temperature for 24 hours. After curing the mold the silica squares are
removed exposing a clean smooth surface. The wafer is then removed from the mold which
exposes the individual diffusion cells which are still attached to the wafer. These cells are
then removed by a gentle peeling and the center teflon cylinders are removed by applying a
light pressure with a center punch.
11.4.5 Deuterated Water Loading and Plasma Treatment for Device Sealing
The PDMS Diffusion cells are placed on a clean piece of silicon and placed into the
vacuum system. Placing the device on a clean piece of silicon allows for isolation of the
plasma to the exposed surface (top). After introducing the samples into the vacuum system,
178
the system is evacuated to a pressure of less than 2.6 Pa. The samples are exposed to a 300
mtorr O2 plasma for 30 seconds. The samples are immediately removed and a clean glass
cover slip (6 mm diameter) is sealed on the plasma treated side of the sample. The sample is
inverted (glass side down) and reintroduced into the vacuum chamber where it is exposed to
a second round of plasma following the aforementioned procedure. Afterward, Deuterated
water is syringe loaded in to the center chamber. The chamber is then sealed with a second
clean glass cover slip. This addition of water and the final seal all occur within 5 minutes of
removal of the diffusion cell from the plasma chamber. The PDMS that was molded into the
ATR accessory is removed from the accessory and contact angle measurements are performed
it is then placed upside down on a glass slide. The samples are exposed to a 300 mtorr O2
plasma for 30 seconds. Upon removal from the vacuum system the PDMS is measured via
contact angle analysis then placed back into the ATR accessory (plasma side down) and
pressure clamped down into the accessory in order to maintain a consistent contact on the
ZeSe crystal to allow for reproducibility in the FTIR measurements.
11.5 Analytical Techniques
11.5.1 Contact Angle Analysis
Water contact angle measurements are obtained by the sessile drop method on a Ram?e-
Hartmodel 200automatedgoniometer (Ram?e-Hart, Inc. MountainLakes, NJ)using DROPim-
age Standard software. Measurement error for this technique is ?2.0?. The samples that
were prepared for ATR/FTIR analysis were characterized for hydrophilic activity before and
after after plasma treatments using contact angle measurements. Deionized water droplets
were delivered on the PDMS surface by a calibrated syringe. The samples are then blown
dry with dry nitrogen. Results showed that the PDMS as cured all had contact angles of
around 120? which dropped to 5? after plasma treatment.
179
11.5.2 Fourier transform infrared (FTIR) spectroscopy
The PDMS films before and after plasma treatment were characterized using FTIR spec-
troscopy (PerkinElmer spectrum 2000 spectrometer) equipped with a deuterated triglycine
sulfate detector. The samples were analyzed using horizontal attenuated total reflection
(HATR) accessory, which consists of a steel plate with an attenuated total reflection (ATR)
element sealed to it that rests on top of an optical alignment box. The optical alignment box
contains two planar mirrors into focal mirrors which can be adjusted to focus the infrared
beam on the incident face of the ATR Crystal. All spectra are recorded at room temperature
at 4.0cm?1 resolution with 284 co added scans. In this study, trapezoidal ZnSe and ATR
elements are used (50 x 20 x 2 mm3, 45? face angle) enabling acquisition of spectral data
from 4000 to 640 cm?1.
11.5.3 Thermogravimetric Analysis and Mass Spectroscopy
A schematic of the TGA Module used for this analysis is illustrated in Figure 11.3.
The water filled PDMS diffusion cells are loaded into a TA Instruments Q-5000 IR. This
system consists of an isolated thermo balance, an IR furnace, quartz sample chamber, purge
inlets and outlets, and a capillary connection to a mass spectrometer. The thermo balance
is maintained at 40 ?C in a well insulated, gas purged chamber isolated from the furnace by
a water cooled plate. Evolved gas products are analyzed by a quadrapole mass spectrometer
(Pfieffer Thermostar) via a heated transfer line. The technical specifications for this unit can
be found in more detail on the manufacturers? website, but include a weighing accuracy of
?0.1% and a weighing precision of?.01%. The furnace consists of four symmetrically placed
IR lamps with a quartz lined sample compartment that is a chemically inert and resistant
to adsorption of off-gas products.
180
11.6 Results and Discussion
A gravimetric method was utilized for a direct determination of steady state water va-
por diffusivity as influenced by curing methods and exposure temperature. Sample cells
fabricated from PDMS, as illustrated in Figure11.2, were loaded with liquid D2O and sealed
following the methodology in section 11.4.5. Weight loss was monitored with TGA in con-
junction with mass spectral observation of D2O vapor evolution with respect to time. The
advantage of this method, is its simplicity, allowing for the continuous recording of sample
weight change as a function of time and vapor activity. In addition, the arrangement allowed
for water to be trapped inside the test device, allowing for a more representative comparison
to the diffusion mechanism associated with microfluidic devices as opposed to membrane
measurements.
Water transport can be described as a three-step process, starting with the absorption
of water molecules to the interior surface of the polymer, diffusion through the polymer and
as the third step, external surface desorption.
The transport phenomena is usually characterized by three quantities: diffusivity D
(m2/s), solubility S (m3/(m3 ? Pa)) and permeability P (m2/(s ? Pa)). Solubility is the
molecular concentration C in the polymer near its surface divided by the partial pressure p
(Pa) of the molecules in the nearby gaseous phase, i.e. according to Henry?s law: C = Sp.
Permeability is the product of diffusivity and solubility P = DS [207]. The solubility limit
of water in PDMS, which has been estimated in various experiments to be between 0.1 and
0.2 wt% at 300 K [208,209].
The linear relation between the flux (J, which is the amount ofmolecules going through a
unit area within the polymer matrix during a unit time) and the gradient of the concentration
of these molecules (?C) is related by the first Ficks law: The permanent flux through the
PDMS device is given by:
J = D?A?(?Cx ) (11.1)
181
where D is the diffusion coefficient and C and x are the concentration within the fabri-
cated system and external to the system and the wall thickness respectively. The PDMS test
device design restricts diffusion to one dimension (radial direction) due to the utilization of
glass seals on the top and bottom.
The diffusivity values were calculated based on assuming steady state. This assumption
was validated by the linear weight loss measured for our test device. Water transport as
limited to the radial direction is modeled with.
?C
?t =
1
r
?
?r(rD
?C
?r ) (11.2)
Steady State Assumed
0 = ??r(r?C?r ) (11.3)
Thus simplified to
Q = 2piDt(C2 ?C1)logb
a
(11.4)
With Q = mass measured from TGA weight loss (mass difference in linear regime) (g)/ unit
length in (m), D (m3/sec), b and a = internal and external radius (m), t = time for the
weight loss interval (sec) and C = concentration (g/m3).
The driving force of water transport is the difference between the partial pressures of
water vapor on the interior and exterior sides. If the water vapor?s partial pressure on both
sides are different, a gradient of water concentration is established within the device. A
schematic representation of the TGA system is shown in Figure 11.3. The external portion
of the cell was maintained at a zero concentration by constantly sweeping a dry helium
through the TGA-MS module. The balance assembly is isolated from the furnace and dry
helium was continuously flowed through the module at 10 ml/min such that positive pressure
was maintained (no moisture regression into balance chamber). Dry Helium was flowed into
182
the purge lines into the purge lines at a flow rate of 20.0 ml/min. The IR furnace is ramped
at 10 ?C/min to the selected temperature (75, 85, 95 ?C) where it is held and weight loss
is monitored with time. D2O resulting from vapor transport from within the PDMS Cell
is monitored by mass spectroscopy. The experiment is allowed to proceed until there is
no longer evident weight loss and the D2O concentration falls back to a zero baseline. An
example of the data evolved is indicated in Figure 11.5, showing both weight loss and water
evolution as a function of time at a temperature of 95 ?C. The initial unsteady state start
(ramping from ambient to temperature of interest) of vapor transport was ignored during
calculations. The diffusivity calculations were performed on the data found in the linear
region of weight loss. This linear region can be identified in Figure 11.5 following an initial
transient period. The internal concentration was calculated based on the vapor pressure
associated with water as a function of hold temperatures. The rate of vaporization was
calculated and compared to the flux from the cell and was found to be 10000X higher than
the diffusion rate from within the cell, thus indicating the rate controlling step was the vapor
diffusion through the PDMS membrane.
The composition and morphology of a polymer play a significant role in determining
transport properties. The factors having the greatest influence fall into two categories: those
relating to the chemical composition and structure such as the degree of cross linking, and
the degree of saturations. As well as those that involve heterogeneities in the polymer,
such as orientation, crystallinity, and the presence of plasticizers and fillers [206]. The
PDMS material presented lacks plasticizers or fillers and the test devices were formed in the
absence of shearing such that orientational effects as assumed inconsequential. It has also
been shown that when cured at room temperature that no preferential transport pathways
exist in PDMS as commonly noted in heterogeneous polymers.
The data in Figure 11.6 show that regardless of the curing procedure as temperature
increases water diffusivity increases. This increase is due to a higher chain mobility within
the PDMS network as influence by increased thermal energy. The diffusion mechanism in
183
PDMS is described by the polymer?s ability to continually provide opportunities for perme-
ants to progress through randomly generated voids ?holes? [209]. Conformations of polymer
chains change successively by the thermal motion, resulting in the fluctuation of the occupa-
tion volume of the chains and the formation of holes, which are effective transport channels
allowing molecules to permeate easily. The plots in Figure 11.6 of the logarithm of the diffu-
sion coefficient vs. the inverse absolute temperature follows a linear relationship confirming
that no preferential transport pathways exist in PDMS despite the differing curing regimes.
In addition, the PDMS cell that was cured at a higher temperature had lower diffusivity
values across all temperatures investigated. This result is consistent with increased chemical
cross linking due to the higher curing temperature. This increase in cross linking has been
shown contribute to a lowered chain segment mobility resulting from an increasing extent of
saturation [210]. The decreased segment mobility results from conformational changes being
limited by steric hinderences and ionic repulsions. These changes often limit the path of
diffusion for molecules forcing a more tortuous path (often longer) due to a limited number
of ?holes? formed and thus lowering diffusivity.
Previously reported experimental water vapor diffusivity values at 300 K show a wide
divergence. The equilibrium sorption and permeation studies of Barrie and Machin obtain a
value of D = 4.0 x 10?5 cm2/s. While Favre et al. obtained a diffusion constant of D ? 104
[209]. For higher concentrations, aggregation effects can be significant in determining the
transport coefficients for water in PDMS. Watson and Baron whose values is most often cited
indicate a value of D = 2.0 x 10?5 cm2 ruling out the theory of water cluster formation. The
values reported herein range from D = 1.0 x 104 cm2 to 9.7 x 104 cm2 at between 348-368K
and are in agreement with previously reported ranges.
The activation energy determines the magnitude of diffusion. The contribution of tem-
perature as a driving force can generally be evaluated from determination of activation
energy over a temperature range [211]. Temperature dependent diffusion coefficients have
184
been used to calculate the activation energies for the diffusion processes by using the equa-
tion D = D0e?Ed/RT. Slope of the Arhenious plot of ln D vs. 1/T (T in kelvin) Figure 11.7
provides the activation energies (Ed). If we look at the activation energies for diffusion they
can generally be explained by considering both the chemical affinity with PDMS and the
molecular size of the penetrants. Barrie and Platt found that clustering of water molecules
into dimers and trimers can lead to a 3-5-fold reduction in the measured diffusion constants.
We feel this mechanism of clustering has been prevented with the PDMS test cells kept
at high temperature, and thus diffusive species are expected to be only water monomers.
Activation energies for the two different curing conditions (1) RT Cure vs (2) 100 ?C(1hr)
followed by 24 hr are calculated at 13.9 kJ and 9.8 kJ respectively. Watson and Baron calcu-
lated that the energy to generate voids of sufficient size to accommodate the water molecules
within the PDMS (cured at 30 ?C for 24 hrs) is 14.0 kJ/mol which is in close agreement
with our experimental value for the RT Cure [212]. The activation energy for the higher
cure condition was slightly lower and it is suspected that cross linking limits the available
free volume ?holes? such that less energy is required to start the diffusion process but, due
to the tortuous path, the overall diffusivity is reduced as compared to the room temperature
cure. In addition, the PDMS cured at the lower temperature will have a higher level of
unsaturated bonds contributing to more hydrogen bonding with the diffusing molecules as
compared to the higher cured case resulting in a higher activation energy.
FTIR analysis was performed to provide a comparison between the two curing regimes.
No differences were noted as illustrated in Figure 11.4 in the FTIR spectra of the PDMS cells
regardless of the curing procedure utilized. The data presented represent the PDMS devices
post plasma treatment as it is known that plasma treatment causes surface activation and
some modification of surface reactive groups.
185
11.7 Conclusion
It has been shown that the choice of curing scheme directly affects the water vapor
diffusivity from within a PDMS cell. PDMS test devices cured at a higher temperature
indicate a lower diffusivity and lower activation energy. This indicated that while less energy
was needed to initiate diffusion a lower diffusivity value was obtained. It is suggested that
increased cross linking contributes to a lowered chain segment mobility limiting the path
of diffusion for molecules, forcing a more tortuous path due to a limited number of ?holes?
formed. FTIR analysis showed no obvious structural differences between the two curing
temperatures. Because cross linking is shown to decrease molecular permeability, the choice
of a curing regimen should be investigated by individual groups to ascertain the starting
point for diffusivity improvements. This information will prove vital to discriminate between
barrier layer improvement as compared to structural rearrangements caused by heating steps
required for their attachment.
Figure 11.1: Reaction diagram of platinum catalyzed PDMS curing. The R group is either
CH3 or H.
186
Figure 11.2: (A) Illustration of mold platform used for fabrication of the PDMS test devices
and (B) PDMS diffusion cell.
Figure 11.3: Representation of the gravimetric system utilized for this work.
187
Figure 11.4: Representative FTIR spectra of PDMS (1) 24-hr room temperature cure after
30-sec oxygen plasma (2)1hr at 100 ?C followed by 24 hr room temperature cure after 30 sec
oxygen plasma
Figure 11.5: Example of data collected from TGA-MS experiments. Both weight loss and
ion current are measured as a function of time. Ramp was 10 ?C/min to 95 ?C and held
until until no observable weight loss.
188
Figure 11.6: Diffusion coefficients as a function of temperature.
Figure 11.7: Data used to calculate activation energy.
189
11.8 Investigation of MVD silica Layers for Water Vapor Barrier Improvement
Prior Work on Utilizing Silica layers on Polymer Surfaces- Implications toward
utilization on PDMS
Research has shown thin transparent SiOx barrier coatings deposited on polymer sub-
strates exhibit a substantially reduced oxygen and water vapor permeability while main-
taining stability in terms of reactivity, time and temperature. Such coatings possess highly
desirable properties, such as transparency, recycle ability, and microwave use as demon-
strated in their use for food packaging and biomedical applications. One of the benefits of
these coatings lies in the flexibility by which they can be deposited on polymer surfaces. Thus
far, sputtering, electron beam deposition, and plasma-enhanced chemical vapor deposition
(PECVD) have all been utilized successfully to produce SiOx coatings on polymer sub-
strates [213?216]. Work done by Erlat et al. on PECVD SiOx on polymer films has revealed
a correlation between SiOx morphology (including defects) and barrier performance [217].
Their group utilized phase-imaging atomic force microscopy and energy-filtered transmis-
sion electron microscopy to compare morphological and permeation results to identify some
of the physical factors governing water vapor permeation through SiOx-modified polymers.
The focus of their research effort was on PET films. This research intend on depositing a
Molecular Vapor Deposition (MVD) silica layer onto PDMS surfaces resulting in a stable
reactive surface capable of sustaining much of the same chemistries associated with glass
surfaces. Due to its controlled coverage and process tunability it is expected that this silica
layer can provide the water barrier properties necessary for PDMS microfluidic devices. The
surface roughness, surface activation procedure, coating thickness and uniformity of SiOx
coatings have been shown to be contributing factors affecting barrier performance on al-
ternative polymer substrates. This work intends on exploring these factors on water vapor
barrier improvement as related to MVD silica layers deposited on PDMS.
190
11.8.1 Results and Discussion
Different morphologies of silica layers were produced by controlling the design space
associated with precursor dosing pressures. The morphologies and thicknesses were charac-
terized via AFM and ellipsometry, and subsequent investigations when viable were performed
on the TGA-MS to characterize the water diffusion as a function of morphology and thick-
ness. If we consider a two layer structure, the total permeability is governed by the less
permeable layer. In the case of an inorganic coating on a permeable polymer film, the per-
meability depends mostly on the permeability coefficient of the coating and on its thickness.
The MVD silica coatings that were produced proved to be full of defects regardless of the
deposition conditions. In addition the coating was susceptible to delamination if exposed to
any type of compressive stress regardless of the surface activation procedure. This delamina-
tion occurred with gentle manipulation of the PDMS test cell and device sealing. Different
thicknesses produced similar failures. A PDMS film with and without coating is illustrated
in Figure 11.8. We were unable to produce a uniform film as deposited on PDMS surfaces
regardless of the deposition conditions utilized. As a result, work was discontinued on this
project.
191
Figure 11.8: PDMS film (top), silica coated PDMS film (bottom).
192
Bibliography
[1] Seong H. Kim, David B. Asay, and Micheal T. Dugger. Nanotribology and MEMS.
nanotoday, 2:22?29, 2007.
[2] C. Carraro, O. W. Yauw, M. M. Sung, and R. Maboudian. Observation of three
growth mechanisms in self-assembled monolayers. Journal Of Physical Chemistry B,
102(23):4441?4445, June 1998.
[3] D. Dowson. History of Tribology. Professional Engineering Publishing, London, 1998.
[4] J. Krim. Friction at macroscopic and microscopic length scales. American Journal of
Physics, 70(9):890?897, September 2002.
[5] A Hariri, J W Zu, and R Ben Mrad. Modeling of dry stiction in micro electro-
mechanical systems MEMS. J. Micromech. Microeng., 16:1195?1206, 2006.
[6] R Maboudian and C Carraro. Surface chemistry and tribology of MEMS. Annual
Review of Physical Chemistry, 55:35?54, 2004.
[7] M. J. Madou. Fundamentals of Microfabrication. CRC Press, New York, 2nd edition
edition, 2002.
[8] R. Maboudian and R. T. Howe. Critical review: Adhesion in surface micromechanical
structures. Journal of Vacuum Science and Technology B, 15:1?20, 1997.
[9] Y. Ando. The effect of relative humidity on friction and pull-off forces measured on
submicron-size asperity arrays. Wear, 238(1):12?19, March 2000.
[10] Y. Ando. Relation between friction and plastic deformation examined by using periodic
asperity arrays fabricated on an AFM multipurpose cantilever. Tribology Letters,
15(2):115?125, August 2003.
[11] U. Srinivasan, M. R. Houston, R. T. Howe, and R. Maboudian. Alkyltrichlorosilane-
based self assembled monolayer films for stiction reduction in silicon micromachines.
Journal of Microelectromechanical Systems, 7(2):252?260, 1998.
[12] Abraham Ulman. An Introduction toUltrathin Organic Films: from Langmuir-Blogett
to Self Assembly. Academic Press, 1991.
[13] A. Ulman. Formation and structure of self-assembled monolayers. Chemical Reviews,
96 (4):1533?1554, 1996.
193
[14] S. R. Wasserman, Y. T. Tao, and G. M. Whitesides. Structure and reactivity of alkyl-
siloxane monolayers formed by reaction of alkyltrichlorosilanes on silicon substrates.
Langmuir, 5(4):1074?1087, 1989.
[15] S. R. Wasserman, G. M. Whitesides, I. M. Tidswell, B. M. Ocko, P. S. Pershan, and
J. D. Axe. The structure of self-assembled monolayers of alkylsiloxanes on silicon ?
a comparison of results from ellipsometry and low-angle x-ray reflectivity. Journal of
the American Chemical Society, 111(15):5852?5861, July 1989.
[16] I. M. Tidswell, T. A. Rabedeau, P. S. Pershan, S.D. Kosowsky, J. P. Folkers, and G. M.
Whitesides. J. Chem. Phys, 95:2854, 1991.
[17] DL Angst and GW Simmons. Moisture absorption characteristics of organosiloxane
self-assembled monolayers. Langmuir, 7(10):2236?2242, OCT 1991.
[18] M. Vallant, H. Brunner, U Mayer, H Hoffmann, and T. Leitner. J. Phys. Chem. B,
102:7190, 1998.
[19] A. N. Parikh, D. L. Allara, I. B. Azouz, and F. Rondelez. An intrinsic relationship be-
tween molecular structure in self-assembled n-alkylsiloxane monolayers and deposition
temperature. Journal of Physical Chemistry, 98(31):7577?7590, August 1994.
[20] D. K. Schwartz. Mechanisms and kinetics of self-assembled monolayer formation.
Annual Review Of Physical Chemistry, 52:107?137, 2001.
[21] T. Balgar, R. Bautista, N. Hartmann, and E. Hasselbrink. An AFM study of the
growth kinetics of the self-assembled octadecylsiloxane monolayer on oxidized silicon.
Surface Science, 532:963?969, June 2003.
[22] M. Goldmann, Davidovits J.V., and Silberzan P. Kinetics of self-assembled silane
monolayers at various temperatures: evidence of 2d foam. Thin Solid Films, pages
166?171, 1998.
[23] T. Wink, S.J. Zuilen, and A. Bult. Self-assembled monolayers for biosensors. Analyst,
122:43R, 1997.
[24] N.K. Chaki and K. Vijayamohanan. Self-assembled monolayers as a tunable platform
for biosensor applications. Biosens. Bioelectron., 17:1, 2002.
[25] N. S. Tambe and B. Bhushan. Nanotribological characterization of self-assembled
monolayers deposited on silicon and aluminium substrates. Nanotechnology,
16(9):1549?1558, September 2005.
[26] F.J. Arregui, I.R. Matias, and Cooper K.L. Fabrication of microgratings on the ends
of standard optical fibers by the electrostatic self-assembly monolayer process. Opt.
Lett, 26:131, 2001.
[27] A Hozumi, H Inagai, Y Yokogawa, and T Kameyama. Molecular-scale growth of silicon
oxide on polymer substrate through vacuum ultraviolet light-assisted photooxidation
of organosilane precursor. Thin Solid Films, 437(1-2):89?94, AUG 1 2003.
194
[28] R Haneda, H. Nishihara, and K.J. Aramaki. Chemical modification of an alkanethiol
self-assembled layer to prevent corrosion of copper. J. Electrochem. Soc., 144:1215,
1997.
[29] S. A. Mirji, S. B. Halligudi, D. P. Sawant, N. E. Jacob, K. R. Patil, A. B. Gaikwad,
and S. D. Pradhan. Adsorption of octadecyltrichlorosilane on mesoporous SBA-15.
Applied Surface Science, 252(12):4097?4103, April 2006.
[30] C. Rill, A. Glaser, J. Foisner, H. Hoffmann, and G. Friedbacher. Temperature-
dependent formation of octadecylsiloxane self-assembled monolayers on mica as studied
by atomic force microscopy. Langmuir, 21(14):6289?6295, JUL 5 2005.
[31] C.R. Kessel and S. Granick. Formation and characterization of a highly ordered and
well-anchored alkysilane monolayer on mica by self-assembly. Langmuir, 7:532?538,
1991.
[32] Lauren N. Mitchon and J. M. White. Growth and analysis of octadecylsiloxane mono-
layers on al2o3 (0001). Langmuir, 22(15):6549?6554, JUL 18 2006.
[33] M. A. Baker and J. Li. The influence of an OTS self-assembled monolayer on the
wear-resistant properties of polysilicon based MEMS. Surface And Interface Analysis,
38(4):863?867, April 2006.
[34] Touraj Manifar, Asad Rezaee, Mehdi Sheikhzadeh, and Silvia Mittler. Formation of
uniform self-assembly monolayers by choosing the right solvent: OTS on silicon wafer,
a case study. Applied Surface Science, 254:4611?4619, 2008.
[35] B. Wang, Z. Abdulali-Kanji, E. Dodwell, J. H. Horton, and R. D. Oleschuk. Surface
characterization using chemical force microscopy and the flow performance of modified
polydimethylsiloxane for microfluidic device applications. Electrophoresis, 24(9):1442?
1450, May 2003.
[36] Sunhyung Lee, T. Ishizaki, N. Saito, and O. Takai. Effect of reaction temperature
on growth of organosilane self-assembled monolayers. Japanese Journal Of Applied
Physics, 47 No. 8:6442?6447, 2008.
[37] M.E. McGovern, K.M.R. Kallery, and M. Thompson. Role of solvent on the silanization
of glass with octadecylsiloxane. Langmuir, 10(10):3607?3614, OCT 1994.
[38] J.D. Legrange, J.L. Markham, and C.R. Kurkjian. Effects of surface hydration on the
deposition of silane monolayers on silica. Langmuir, 9(7):1749?1753, JUL 1993.
[39] N. Rozlosnik, M. C.Gerstenberg, and N.B.Larsen. Effect ofsolvents andconcentration
on the formation of a self-assembled monolayer of octadecylsiloxane on silicon (001).
Langmuir, 19(4):1182?1188, February 2003.
[40] M. M. Sung, C. Carraro, O. W. Yauw, Y. Kim, and R. Maboudian. Reversible liquid-
liquid transitions in the early stages of monolayer self-assembly. Journal Of Physical
Chemistry B, 104(7):1556?1559, February 2000.
195
[41] J. B. Brzoska, I. Benazouz, and F. Rondelez. Sialnization of solid substrates - a step
toward reproducibility. Langmuir, 10(11):4367?4373, November 1994.
[42] D. L. Allara, A. N. Parikh, and F. Rondelez. Evidence for a unique chain organization
in long-chain silane monolayers deposited on two widely different solid substrates.
Langmuir, 11(7):2357?2360, July 1995.
[43] R.W. Wang, J. Guo, G. Baran, and S.L. Wunder. Characterization of the state of order
of octadecylsilane chains on fumed silica. Langmuir, 16(2):568?576, JAN 25 2000.
[44] C.P. Tripp and M.L. Hair. An infrared study of the reaction of octadecyltrichlorosilane
with silica. Langmuir, 8(4):1120?1126, APR 1992.
[45] Brian G.Bush, F.W.Delrio, Opatkiewicz Justin, R.Maboudian, andC. Carraro. Effect
of formation temperature and roughness on surface potential of octadecytrichlorosilane
self-assembled monolayer on silicon surfaces. J. Phys. Chem. A., 111:12339?12343,
2007.
[46] Sun Hyung Lee, Nagahiro Saito, and Osamu Takai. The importance of precursor
molecules symmetry in the formation of self-assembled monolayers. Japanese Journal
of Applied Physics, 46 No.3A:1118?1123, 2007.
[47] Bo H. Lee and M. M. Sung. Effect of island size on the paking density in the early
stages of alkylsilane-based monolayer self assembly. Bull. Korean Chem. Soc, 26:127?
130, 2005.
[48] G.J. Kluth, M.M. Sung, and R. Maboudian. Thermal behavior of alkylsiloxane self-
assembled monolayers on the oxidized si(100) surface. Langmuir, 13(14):3775?3780,
JUL 9 1997.
[49] H.K. Kim, J.P. Lee, C.R. Park, H.T. Kwak, and M.M. Sung. Thermal decomposition
of alkylsiloxane self-assembled monolayers in air. Journal of Physical Chemistry B.,
107(18):4348?4351, MAY 8 2003.
[50] M. R. Houston, R. T. Howe, and R. Maboudian. Self-assembled monolayer films
as durable anti-stiction coatings for polysilicon microstructures. Technical Digest,
Proceedings of the 1996 Solid-State Sensor and Actuator Workshop, Hilton Head,
pages 42?47, 1996.
[51] M. Calistri-Yeh, E.J. Kramer, R. Sharma, W. Zhao, M.H. Rafailovich, J. Sokolov, and
J.D. Brock. Thermal stability of self-assembled monolayers from alkylchlorosilanes.
Langmuir, 12:2747?2755, 1996.
[52] K. C. Popat, R. W. Robert, and T. A. Desai. Characterization of vapor deposited
thin silane films on silicon substrates for biomedical microdevices. Surface & Coatings
Technology, 154(2-3):253?261, May 2002.
196
[53] M. Maccarini, M. Himmelhaus, S. Stoycheva, and M. Grunze. Characterization and
stability of hydrophobic surfaces in water. Applied Surface Science, 252 (5):1941?1946,
2005.
[54] Elizabeth E. Parker. Microelectromechanical systems (MEMS) tribology in microflu-
idic environments. Master?s thesis, University of California at Berkeley, 2004.
[55] A. Anderson and W. Robert Ashurst. Investigation of a vapor-deposited thin silica film:
Morphological and spectral characterization. Langmuir, 10.1021/la800591n, 2008.
[56] K Iimura and T. Kato. Colloid Science A, 171:241, 2000.
[57] Maik J. Geerken, Thomas S. van Zanten, Rob G.H. Lammertink, and Matthias
Wessling. Chemical and thermal stability of alkylsilane based coatings for membrane
emulsification. Advanced Engineering Materials, 6 No. 9:749?754, 2004.
[58] J Krim. QCM tribology studies of thin adsorbed films. nanotoday, 2(5):28?43, 2007.
[59] M. T. McDermott, J. B. D. Green, and M. D. Porter. Scanning force microscopic
exploration of the lubrication capabilities of n-alkanethiolate monolayers chemisorbed
at gold structural basis of microscopic friction and wear. Langmuir, 13(9):2504?2510,
April 1997.
[60] G. J. Leggett. Friction force microscopy of self-assembled monolayers: probing molecu-
lar organisation at the nanometre scale. Analytica Chimica Acta, 479(1):17?38, March
2003.
[61] B. Bhushan and S. Sundararajan. Micro/nanoscale friction and wear mechanisms
of thin films using atomic force and friction force microscopy. Acta Materiala,
46(11):3793?3804, July 1998.
[62] W. R. Ashurst, C. Carraro, and R. Maboudian. Vapor phase anti-stiction coatings
for MEMS. IEEE Transactions on Device and Materials Reliability, 3(4):173?178,
December 2003.
[63] S. Miller, J. J. Sniegowski, G. LaVigne, and P. J. McWorther. Friction in surface
micromachined microengines. Proc. SPIE Smart Electronics and MEMS, 2722:197?
204, 1996.
[64] S. L. Miller, M. S. Rodgers, G. La Vigne, J. J. Sniegowski, P. Clews, D. M. Tanner,
and K. A. Peterson. Failure modes in surface micromachined microelectromechanical
actuation systems. Microelectronics Reliability, 39(8):1229?1237, August 1999.
[65] J. Barriga, B. Coto, and B. Fernandez. Molecular dynamics study of optimal packing
structure of ots self-assembled monolayers on sio2 surfaces. Tribology International,
40(6):960?966, JUN 2007.
[66] M. Salmeron. Generation of defects in model lubricant monolayers and their contribu-
tion to energy dissipation in friction. Tribology Letters, 10 (2):69?79, 2001.
197
[67] R. ?Astr?om, R. Mutikainen, H. Kuisma, and A. H. Hakola. Effect of alkylsilane coating
on sliding wear of silica-silicon contacts with small amplitude motion. Wear, 253(7-
8):739?745, October 2002.
[68] J. N. Ding, P. L. Wong, and J. C. Yang. Friction and fracture properties of polysilicon
coated with self-assembled monolayers. Wear, 260(1-2):209?214, January 2006.
[69] T. Kasai, B. Bhushan, G. Kulik, L. Barbieri, and P. Hoffmann. Micro/nanotribological
study ofperfluorosilane SAMs forantistiction and low wear. Journal ofVacuum Science
and Technology B, 23(3):995?1003, 2005.
[70] M. J. Brukman, G. O. Marco, T. D. Dunbar, L. D. Boardman, and R. W. Carpick.
Nanotribological properties of alkanephosphonic acid self-assembled monolayers on alu-
minum oxide: Effects of fluorination and substrate crystallinity. Langmuir, 22(9):3988?
3998, April 2006.
[71] B. D. Beake and G. J. Leggett. Friction and adhesion of mixed self-assembled mono-
layers studied by chemical force microscopy. Physical Chemistry Chemical Physics,
1(14):3345?3350, July 1999.
[72] E. E. Flater, W. R. Ashurst, and R. W. Carpick. Nanotribology of octadecyltrichlorosi-
lane monolayers and silicon: Self-mated versus unmated interfaces and local packing
density effects. Langmuir, 23(18):9242?9252, August 2007.
[73] J. D. Kiely, J. E. Houston, J. A. Mulder, R. P. Hsung, and X. Y. Zhu. Adhesion,
deformationand frictionforself-assembled monolayers onAu andSisurfaces. Tribology
Letters, 7(2-3):103?107, 1999.
[74] J. T. Woodward, A. Ulman, and D. K. Schwartz. Self-assembled monolayer growth of
octadecylphosphonic acid on mica. Langmuir, 12(15):3626?3629, July 1996.
[75] Nikolaj Gadegaard, Xinyong Chen, Frank J.M. Rutten, and R. Alexander, Morgan.
High-energy electron beam lithography of octadecylphosphonic acid monolayers on
aluminum. Langmuir, 24 (5):2057?2063, 2008.
[76] Brett M. Silverman, Kristen A. Wieghaus, and Jeffery Schwartz. Comparative proper-
ties of siloxane vs phosphonate monolayers on a key titanium alloy. Langmuir, 21:225?
228, 2005.
[77] Irina Gouzman, Manish Dubey, Michael D. Carolus, Jeffrey Schwartz, and Steven L.
Bernasek. Monolayer vs. multilayer self-assembled alkylphosphonate films: X-ray pho-
toelectron spectroscopy studies. Surface Science, 600:773?781, 2006.
[78] A. Kanta, R. Sedev, and J. Ralston. The formation and stability of self-assembled
monolayers of octadecylphosphonic acid on titania. Colloids and Surfaces A:
Physiochem. Eng. Aspects, 291:51?58, 2006.
198
[79] B. Bhushan, M. Cichomski, E. Hoque, and J.A. DeRose. Nanotribological characteriza-
tion of perfluoroalkylphosphonate self-assembled monolayers deposited on aluminum-
coated silicon substrates. Microsystem Technologies, 12:588?596, 2006.
[80] Alan Berman, Suzi Steinberg, Samuel Campbell, Avi Ulman, and Jacob Israelachivili.
Controlled microtribology of a metal oxide surface. Tribology Letters, 4:43?48, 1998.
[81] E. Hoque, J.A. DeRose, P. Hoffman, B. Bhushan, and H.J. Mathieu. Alkylperfluorosi-
lane self-assembled monolayers on aluminum: A comparison with alkylphosphonate
self-assembled monolayers. J. Phys. Chem. C., 111(10):3956?3962, 2007.
[82] R.D. Ramsier, P.N. Henriksen, and A.N. Gent. Adsorption of phosphorus acids on
alumina. Surf. Sci., 203:72, 1988.
[83] Ioannis L. Liakos, Eoghan McAlpine, Xinyong Chen, Roger Newman, and R. Alexan-
der, Morgan. Assembly of octadecyl phosphonic acid on the a-al2o3 (0 0 0 1) surface
of air annealed alumina: Evidence for termination dependent adsorption. Applied
Surface Science, 255:3276?3282, 2008.
[84] M. Giza, P. Thissen, and G. Grundmeier. Adsorption kinetics of organophosphonic
acids on plasma-modified oxide-covered aluminum surfaces. Langmuir, 24 (16):8688?
8694, 2008.
[85] R.D. Peters, P.F. Nealey, J.N. Crain, and F.J. Himpsel. A near edge x-ray absorption
fine structure spectroscopy investigation of the structure of self-assembled films of
octadecyltrichlorosilane. Langmuir, 18(4):1250?1256, 2002.
[86] N.J. Harrick. Internal Reflection Spectroscopy. Interscience Publishers, 1967.
[87] H.D. Downing and D. Williams. Journal of Geophysical Research, 80:1656?1661, 1975.
[88] W. R. Ashurst, C. Carraro, R. Maboudian, and W. Frey. Wafer level anti-stiction
coatings for MEMS. Sensors And Actuators A-Physical, 104(3):213?221, May 2003.
[89] S. Szunerits and R. Boukherroub. Preparation and characterization of thin films of siox
on gold substrates for surface plasmon resonance studies. Langmuir, 22(4):1660?1663,
February 2006.
[90] D. B. Asay and S. H. Kim. Evolution of the adsorbed water layer structure on silicon
oxide at room temperature. Journal of Physical Chemistry B, 109(35):16760?16763,
September 2005.
[91] G.E. Ewing. Thin film water. Journal of Physical Chemistry B, 108(41):15953?15961,
2004.
[92] J.R. Scherer. Advances in Infrared and Raman Spectroscopy. Heyden: London, 1978.
[93] A. Anderson and W. Robert Ashurst. Interfacial water structure on a highly hydrox-
ylated silica film. Langmuir, 25 (19):11549?11554, 2009.
199
[94] Anna Barnette, D. B. Asay, and Seong H. Kim. Average molecular orientations in
the adorbed water layers on silicon oxide in ambient conditions. Physical Chemistry
Chemical Physics, 10:4981?4986, 2008.
[95] R. S. McDonald. Surface functionality of amorphous silica by infrared spectroscopy.
Journal Of Physical Chemistry, 62(10):1168?1178, 1958.
[96] K.A. Becraft and G.L. Richmond. In situ vibrational spectroscopic studies of the
CaF2/H2O interface. Langmuir, 17:7721?7724, 2001.
[97] Quan Du, Eric Freysz, and Y. Ron Shen. Surface vibrational spectroscopic studies of
hydrogen bonding and hydrophobicity. Science, 264(5160):826?828, 1994.
[98] M.R. Yalamanchili, A.A. Atia, and J.D. Miller. Analysis of interfacial water at a
hydrophilic silicon surface by in-situ ftir/internal reflection spectroscopy. Langmuir,
12:4176?4184, 1996.
[99] Yubo Fan, Xin Chen, Lijiang Yang, Paul S. Cremer, and Yi Qin Gao. On the structure
of water at the aqueous/air interface. J. Phys. Chem. B, 113:11672?11679, 2009.
[100] Zheng Yang, Qifeng Li, and Keng C. Chou. Structures of water molecules at the
interfaces of aqueous salt solutions and silica: Cation effects. J. Phys. Chem. C,
113:8201?8205, 2009.
[101] V.M. Gun?ko, V.V. Turov, V.M. Bogatyrev, V.I. Zarko, R. Leboda, E.V. Goncharuk,
and A. A. Chuiko. Unusual properties of water at hydrophilic/hydrophobic interfaces.
Advances in Colloid and Interface Science, 118:125?172, 2005.
[102] G.E. Ewing. Ambient thin film water on insulator surfaces. Chem. Rev., 106:1511?
1526, 2006.
[103] Yuen Ron Shen and Victor Ostroverkhov. Sum-frequency vibrational spectroscopy
on water interfaces: Polar orientation of water molecules at interfaces. Chem. Rev.,
106:1140?1154, 2006.
[104] Dimitrios Argyris, David R. Cole, and Alberto Striolo. Hydration structure on crys-
talline silica substrates. Langmuir, 25 (14):8025?8035, 2009.
[105] Victor Ostroverkhov, Glen A. Waychunas, and Y.R. Shen. New information on water
interfacial structure revealed by phase-sensitive surface spectroscopy. Physical Review
Letters, 94:046102, 2005.
[106] M. Antognozzi, A.D.L. Humphris, and M.J. Miles. Observation of molecular layering
in a confined water film and study of the layers viscoelastic properties. Applied Physics
Letters, 78 (3):300?302, 2001.
[107] Matthew Goertz, J. E. Houston, and X.-Y. Zue. Hydrophilicity and the viscosity of
interfacial water. Langmuir, 23:5491?5497, 2007.
200
[108] Tai-De Li, Jianping Gao, Robert Szoszkiewicz, Uzi Landman, and Elisa Riedo. Struc-
tured and viscous water in subnanometer gaps. Physical Review, 75:115415, 2007.
[109] Santiago Romero-Vargas Castrillon, Nicolas Giovambattista, Ilhan Aksay, and
Pablo G. Debenedetti. Effect of surface polarity on the structure and dynamics of
water in nanoscale confinement. J. Phys. Chem. B, 113:1438?1446, 2009.
[110] O.P. Khatri, C.D. Bain, and S.K. Biswas. Effects of chain length and heat treatment
on the nanotribology of alkylsilane monolayers self-assembled on a rough aluminum
surface. Journal of Physical Chemistry B, 109 (49):23405?23414, 2005.
[111] H.G. Hong, D.D. Sackett, and T.E. Mallouk. Chem. Mater., 3:521?527, 1991.
[112] RW Wang and SL Wunder. Thermal stability of octadecylsilane monolayers on sil-
ica: Curvature and free volume effects. JOURNAL OF PHYSICAL CHEMISTRY B,
105(1):173?181, JAN 11 2001.
[113] G. Carson and S. Granick. Self-assembly of octadecyltrichlorosilane monolayers on
mica. Journal of Materials Research, 5 (8):1745, 1990.
[114] S. A. Henck. Lubrication of Digital Micromirror DevicesTM. Tribology Letters,
3(3):239?247, 1997.
[115] K. C. Eapen, S. T. Patton, S. A. Smallwood, B. S. Phillips, and J. S. Zabinski.
MEMS lubricants based on bound and mobile phases of hydrocarbon compounds: Film
deposition and performance evaluation. Journal of Microelectromechanical Systems,
14(5):954?960, October 2005.
[116] K. C. Eapen, S. A. Smallwood, and J. S. Zabinski. Lubrication of MEMS under
vacuum. Surface and Coatings Technology, 201(6):2960?2969, December 2006.
[117] N. Gunda, S. K. Jha, and S. A. Sastri. Anti-stiction coatings for mems devices.
Advanced Materials & Processes, 162(9):27?28, September 2004.
[118] R. Maboudian, W. R. Ashurst, and C. Carraro. Self-assembled monolayers as anti-
stiction coatings for MEMS: Characteristics and recent developments. Sensors and
Actuators A-Physical, 82(1-3):219?223, May 2000.
[119] M. O. Robbins and D. Smith Elizabeth. Connecting molecular-scale and macroscopic
tribology. Langmuir, 12:4543?4547, 1996.
[120] R. W. Carpick and M. Salmeron. Scratching the surface: Fundamental investigations
of tribology with atomic force microscopy. Chemical Reviews, 97(4):1163?1194, June
1997.
[121] I.L. Singer. Friction and energy dissipation at the atomic scale: A review. J. Vac. Sci.
Technol. A, 12 (5):2605?2616, 1994.
[122] B. Bhushan, J. N. Israelachvill, and U. Landman. Nanotribology: Friction, wear and
lubrication at the atomic scale. Nature, 374:607?616, 1995.
201
[123] B. Bhushan. Micro/nanotribology: State of the art andits applications. In B. Bhushan,
editor, Tribology Issues and Opportunities in MEMS, pages 229?246, Dordrecht,
Netherlands, 1998. Kluwer Academic Publishers. Invited.
[124] Yue-an Cheng, Bin Zheng, Po-hsiang Chuang, and Shuchen Hsieh. Solvent effects
on molecular packing and tribological properties of octadecyltrichlorosilane films on
silicon. Langmuir, 2010 DOI: 10.102/la904656y.
[125] J.Q. Ma, C.J. Pang, Y.F. Mo, and M.W. Bai. Preparation and tribological properties
of multiply-alkylated cyclopentane (mac)-octadecyltrichlorosilane (ots) double-layer on
silicon. Wear, 262:1000?1007, 2007.
[126] Wang Xin, Zhang Xiangjun, Ahmed Imad, and Wen Shizhu. Friction properties of
ots sams and silicon surface under water lubrication. Journal of Wuhan University of
Technology-Mater. Sci, pages 273?277, 2009 DOI 10.1007/s11595-009-2273-3.
[127] Sili Ren, Shengrong Yang, Yapu Zhao, Jinfang Zhou, Tao Xu, and Weimin Liu. Friction
and wear studies of octadecyltrichlorosilane sam on silicon. Tribology Letters, 13:233?
239, 2002.
[128] Masuko. Tribological characteristics of self-assembled monolayer with siloxane bonding
to Si surface. Tribology International, 40:1587?1596, 2007.
[129] Chang Xu, Ryan L. Jones, and James D Batteas. Dynamic variations in adhesion
of self-assembled monolayers on nanoasperities probed by atomic force microscopy.
Scanning, 30:106?117, 2008.
[130] S. A. Kulkarni, S. A. Mirji, A. B. Mandale, and K. P. Vijayamohanan. Thermal
stability of self-assembled octadecyltrichlorosilane monolayers on planar and curved
silica surfaces. Thin Solid Films, 496(2):420?425, 2006. Thin Solid Films.
[131] Ryan L Jones, Nicole C. Pearsall, and James D. Batteas. Disorder in alkylsilane
monolayers assembled on surfaces with nanoscopic curvature. J. Phys. Chem. C.,
113:4507?4514, 2009.
[132] K Kojio, A Takahara, and T Kajiyama. Formation mechanism of n-
octadecyltrichlorosilane monolayer prepared at the air/water interface. Colloids and
Surfaces A- Physichochemical and Engineering Aspects, 169(1-3):295?306, SEP 10
2000.
[133] I. Doudevski, W. A. Hayes, and D. K. Schwartz. Submonolayer island nucleation and
growth kinetics during self-assembled monolayer formation. Physical Review Letters,
81(22):4927?4930, November 1998.
[134] H. Hoffman, U. Mayer, H. Brunner, and A. Krischanitz. Reflection-absorption infrared-
spectroscopy of self-assembled monolayers on gold and silicon surfaces. Vibrational
Spectroscopy, 8(2):151?157, JAN 1995.
202
[135] Y. Hussain, J. Krim, and C. Grant. OTS adsorption: A dynamic QCM study. Colloids
and Surfaces A- Physiochemical and Engineering Aspects, 262(1-3):81?86, JUL 15
2005.
[136] Aimee Poda, A. Anderson, and Robert W. Ashurst. Ots monolayer formation on a
highly hydrated silica layer. Thin Solid Films, pages DOI: APSUSC?D?10?00983,
2010.
[137] T. Komeda. Self-assembled-monolayer film islands as a self-patterned-mask for SiO2
thickness measurement with atomic force microscopy. Appl. Phys. Lett., 70:3398?3400,
1997.
[138] A. Anderson and W. R. Ashurst. Investigation of a vapor-deposited thin silica film:
Morphological and spectral characterization. Langmuir, 24:7947?7954, 2008.
[139] Mika Harjiu, Tapio Mantyla, Kalle Vaha-Heikkila, and Vesa-Pekka Lehto. Water ad-
sorption on plasma sprayed transition metal oxide. Applied Surface Science, 249:115?
126, 2005.
[140] Hendrik Bluhm. Photoelectron spectroscopy of surfaces under humid
conditions. Journal of Electron Spectroscopy and Related Phenomena,
doi:10.1016/j.elspec.2009.08.006, 2009.
[141] Jianwei Wang, Andrey G. Kalinicher, and R. James Kirkpatrick. Asymmetric hydrogen
bonding and orientational ordering of water at hydrophobic and hydrophilic surfaces:
A comparison of water/vapor, water/talc, and water/mica interfaces. J. Phys. Chem.
C., 113:11077?11085, 2009.
[142] Dongyan Xu, Yongsheng Leng, Yunfei Chen, and Deyu Li. Water structure near
charged (100) and (111) silicon surfaces. Applied Physics Letters, 94:201901, 2009.
[143] S. Trakhtenberg, R. Naaman, and S. R. Cohen. Effect of the substrate morphology on
the structure of adsorbed ice. J. Phys. Chem. B., 101:5172?5176, 1997.
[144] Yinon Rudich, Ilan Benjamin, R. Naaman, Elan Thomas, and S. Trakhtenberg. Wet-
ting of hydrophobic organic surfaces and its implications to organic aerosols in the
atmosphere. J. Phys. Chem. A, 104:5238?5245, 2000.
[145] G.E. Jr. Brown and G.A. Parks. Int. Geol. Rev., 43:963, 2001.
[146] Bogdan C. Donose, Elena Taran Ivan, Ivan U. Vakarelski, Hiroyuki Shinto, and Ko Hi-
gashitani. Effects of cleaning procedures of silica wafers on their friction characteristics.
Journal of Colloid and Interface Science, 299:233?237, 2006.
[147] Lucel Sirghi, Nobuyuki Nakagiri, Katsumi Sugisaki, and Hiroyuki Sugimuri. Effect of
sample topography on adhesive force in atomic force spectrosopy measurements in air.
Langmuir, 16:7796?7800, 2000.
203
[148] Ann Louise Sumner, Erik J. Menke, Yael Dubowski, J. T. Newberg, Reginald M.
Penner, and B. J. Finlayson-Pitts. The nature of water on surfaces of laboratory
systems and implications for heterogeneous chemistry in the troposphere. Phys. Chem.
Chem. Phys., 6:604?613, 2004.
[149] D. B. Asay, AL Barnette, and SH Kim. Effects of surface chemistry on structure
and thermodynamics of water layers at solid-vapor interfaces. Journal of Physical
Chemistry C, 113 (6):2128?2133, 2009.
[150] D. Dai, J. Peters, and G.E. Ewing. Water adsorption and dissociation on nacl surfaces.
J. Phys Chem, 99:10299?10304, 1995.
[151] Kastsuyuki Utani and Sadao Adachi. Hf-treated (111), (110) and (100) si suraces
studied by spectroscopic ellipsometry. Jpn. J. Appl. Phys., 32 Part 1, No. 8:3572?
3576, 1993.
[152] K.T. Queeny, H. Fukidome, E.E. Chaban, and Y.J. Chabal. In-situ ftir studies of
reactions at the silicon/liquid interface: Wet chemical etching of ultrathin sio2 on
si(100). J. Phys. Chem. B., 105:3903?3907, 2001.
[153] A. Munkholm and Brennan P. Goodbread. Wafer cleaning influence on the rough of
the si/sio2 interface. Mater. Res. Soc. Symp. Proc., Symposium O, 1995.
[154] T. Maeda, Akira Kurokawa, and K. Sakamoto. Atomic force microscopy observation
of layer-by-layer growth of ultrathin silicon dioxide by ozone gas at room temperature.
J. Vac. Sci Technology B., 19(2):589?592, 2001.
[155] F.J. Grunthaner and J. Maserjian. IEEE Trans. Nuci., NS-24:2108, 1977.
[156] Tetsuya Narushima, Masahiro Kitajima, and Akiko N. Itakura. Etching-enhanced
surface stress relaxation during initial ozone oxidation. Surf. Sci., 601:1384?1388, 2007.
[157] A. W. Flounders, S.A. Bell, and D.W. Hess. Journal of the Electrochemical Society,
140:1414?1424, 1993.
[158] S. Chen and S. Jiang. An new avenue to nonfouling materials. Advanced Materials,
20(2):335?338, 2008.
[159] Zhenjiang Ciu, Jonathan M. Madsen, and Christos G. Takoudis. Rapid thermal oxi-
dation of silicon in ozone. Journal of Applied Physics, 87 (11):8181?8186, 2000.
[160] Akira Kurokawa, Ken Nakamura, and Shingo Ichimura. Reduction of the interfacial si
displacement of ultrathin sio2 on si(100) formed by atmospheric-pressure ozone. Appl.
Phys. Lett., 76 (4):493, 2000.
[161] P. Thissen, Markus Valtiner, and G. Grundmeier. Stability of phosphonic acid self-
assembled monolayers on amorphous and single-crystalline aluminum oxide surfaces in
aqueous solution. Langmuir, 26(1):156?164, 2009.
204
[162] Nolan Tillman, A. Ulman, Jay S. Schildkraut, and Thomas L Penner. Incorporation of
phenoxy groups in self-assembled monolayers of trichlorosilane derivatives: Effects on
film thickness, wettability, and molecular orientation. J. American Chemical Society,
I 10 6:135?144, 1988.
[163] Mark J. Pellerite, T. D. Dunbar, L. D. Boardman, and Erika J Wood. Effects of flu-
orination on self-assembled monolayer formation from alkanephoshonic acids on alu-
minum: Kinetics and structure. J. Phys. Chem. B., 107:11726?11736, 2003.
[164] C. Bram, C. Jung, and M. Stratmann. Fresenius J. Anal. Chem., 358:108, 1997.
[165] Jeffery McElwee, Roy Helmy, and A. Y. Fadeev. Thermal stability of organic mono-
layers chemically grafted to minerals. Journal of Colloid and Interface Science, pages
551?556, 2005.
[166] S. McGovern, G. Harish, and C.S. Pai. Multiphase flow regimes for hydrogenation in
a catalyst-trap microreactor. Chemical Engineering Journal, 135S:S229?S236, 2008.
[167] W. Ehrfeld, V.Hessel, and H. Lowe. Microreactors-New technology for Modern
Chemistry. Wiley-VCH, 2000.
[168] V. Hessel, S. Herdt, and H. Lowe. Chemical micro process engineering: fundamentals,
modeling, and reactions. Wiley-VCH, 2004.
[169] H. Lowe, V. Hessel, and S. Hart. Chemical micro process engineering. Wiley-VCH,
2004.
[170] Ye Liu, Hong Wang, Jianfeng Li, and Yong Lu. Monolithic microfibrous nickel catalyst
co-modified with ceria and alumina for miniature hydrogen production via ammonia
decomposition. Applied Catalysis A: General, 328:77?82, 2007.
[171] New York, editor. The Fischer-Tropsch Synthesis. Academic Press, 1984.
[172] J.C. Bertole, Gabor Kiss, and CA Mims. The effect of surface active carbon on hydro-
carbon selectivity in the cobalt catalyzed fischertropsch synthesis. Journal of Catalysis,
223:309?318, 2004.
[173] G.P. Van Der Laan and A.A.C.M. Beenackers. Kinetics and selectivity of the fischer-
tropsch synthesis: A literature review. Catal. Rev.-Sci. Eng., 41:255?318, 1999.
[174] Enrique Iglesia, S.C. Reyes, and Stuart L. Madon, R.J. adn Soled. Selectivity con-
trol and catalyst design in the fischer-tropsch synthesis - sites, pellets, and reactors.
Advances in Catalysis, 39:221?302, 1993.
[175] B.W. Wojciechowski. The reaction mechanism of catalytic cracking: Quantifying ac-
tivity, selectivity, and catalyst decay. Catalysis Reviews-Science and Engineering, 40
(3):209?328, 1998.
[176] Burtron Davis. Fischer-tropsch synthesis: Comparison of performances of iron and
cobalt catalysts. Ind. Eng. Chem. Res., 46:8938?8945, 2007.
205
[177] A.Y. Khodakov, J Lynch, D Bazin, B Rebours, N. Zanier, and B Moisson. Reducibility
of cobalt species in silica supported fischer-tropsch catalysts. Journal of Catalysis, 168
(1):16?25, 1997.
[178] J.S. Girardon, E. Quinet, Anne Griboval-Constant, and P.A. Chernavskii. Cobalt
dispersion, reducibility, and surface sites in promoted silica-supported fischer-tropsch
catalysts. Journal of Catalysis, 248 (2):143?157, 2007.
[179] J.-S. Girardon, A Constant-Griboval, A Gengembre, P.A. Chernavskii, and A.Y. Kho-
dakov a. Optimization of the pretreatment procedure in the design of cobalt silica
supported fischertropsch catalysts. Catalysis Today, 106:161?165, 2005.
[180] Guozhu Bian, Takehisa Mochizuki, Norimasa Fujishita, and Hiderou Nomato. Activa-
tion and catalytic behavior of several Co/SiO2 catalysts for fischer-tropsch synthesis.
Energy and Fuels, 17:799?803, 2003.
[181] P. Dutta, Nimir O. Elbashir, A. Manivannan, and C.B. Roberts. Characterization of
fischertropsch cobalt-based catalytic systems Co/SiO2 and Co/Al2O3 by x-ray diffrac-
tion and magnetic measurements. Catalysis Letters, 98 No. 4:203?210, 2004.
[182] R.L. Espinoza, A.P. Steynberg, and A.C. Jager. Low temperature fischer tropsch
synthesis from a sasol perspective. Catalysis A: General, 186:13?26, 1999.
[183] D.C. Song and J. L. Li. Effect of catalyst pore size on the catalytic performance of
silica supported cobalt fischer tropsch catalysts. Journal of Molecular Catalysis A:
Chemical, 247 (1-2):206?212, 2006.
[184] A.Y. Khodakov, Anne Griboval-Constant, Rafeh Bechara, and Vladimir L.
Zholobenko. Pore size effects in fischer tropsch synthesis over cobalt-supported meso-
porous silicas. Journal of Catalysis, 206:230?241, 2002.
[185] Y Zhang, Y Liu, G.H. Yang, S.L. Sun, and N. Tsubaki. Effects of impregnation solvent
on Co/SiO2 catalyst for fischer-tropsch synthesis: A highly active and stable catalyst
with bimodal sized cobalt particles. Applied Catalysis A-General, 321 (1):79?85, 2007.
[186] A.Y. Khodakov, Anne Griboval-Constant, Rafeh Bechara, and F. Villian. Pore-size
control of cobalt dispersion and reducibility in mesoporous silicas. Journal of Physical
Chemistry B, 105 (40):9805?9811, 2001.
[187] Sameer K Ajmera, K.F. Losey, Klavs F. Jensen, and M.A. Schmidt. Microfabricated
packed-bed reactor for distributed chemical synthesis: The heterogeneous gas phase
production of phosgene as a model chemistry. AIChE Journal, 2001.
[188] Noo Li Jeon and Daniel T. Chiu. Design and fabrication of integrated passive
valves and pumps for flexible polymer 3 dimensional microfluidic systems. BioMedical
Microdevices, 4 (2):117?121, 2002.
206
[189] Ranjit Prakash and Karen V.I.S. Kaler. An integrated genetic analysis microfluidic
platform with valves and a PCR chip reusability method to avoid contamination.
Microfluid Nanofluid, 3:177?187, 2007.
[190] A. Anderson and W. R. Ashurst. Investigation of a vapor-deposited thin silica film as a
novel substrate for in situ fourier transform infrared spectroscopy. Journal of Vacuum
Science and Technology?A, 2008.
[191] T. Fujii. PDMS-based microfluidic devices for biomedical applications. microelectronic
engineering 61-62 (2002) 907-914. Microelectronic Engineering, 61-62:907?914, 2002.
[192] J. Cooper McDonald, D. C. Duffy, Janelle R. Anderson, Daniel T. Chui, Hongkai Wu,
Olivier J. Schueller, and G. M. Whitesides. Fabrication of microfluidic systems in
poly(dimethylsiloxane). Electrophoresis, 21:27?40, 2000.
[193] N Koo. Improved mold fabrication for the definition of high quality nanopatterns
by soflt uv-naoimprint lithography using diluted pdms material. Microelectronic
Engineering, 84:904?908, 2007.
[194] Gregory T. Roman and C. T. Culbertson. Surface engineering of
poly(dimethylsiloxane) microfluidic devices using transition metal sol-gel chem-
istry. Langmuir, 22:4445?4451, 2006.
[195] D. C. Duffy, J.C. McDonald, J.A. Schueller, and G. M. Whitesides. Rapid prototyping
of microfluidic systems in poly(dimethylsiloxane). Anal. Chem., 70:4974, 1998.
[196] Spencer L. Roach, Helen Song, and Rustem F. Ismagilov. Controlling nonspecific pro-
tein adsorption in a plug-based microfluidic system by controlling interfacial chemistry
using fluorous-phase surfactants. Anal. Chem., 77 (3):785?796, 2005.
[197] Susanta Banerjee, Ram Asrey, and Chhaya Saxena. Kinetics of diffusion of water and
dimethylmethylphosphonate through poly(dimethylsiloxane) membrane using coated
quartz piezoelectric sensor. J. Polym. Sci., 57:1483, 1995.
[198] Phillip Gibson and Donald Rivin. Measurement of water vapor diffusion through
polymer films and fabric/membrane laminates using a diode laser spectroscop. Polym.
Plast. Technol. Eng., 38(2):221?239, 1999.
[199] M. Lounanci, P. Rigolet, C. Abraham, M. Le Berre, and Y. Chen. Microfluidic device
for protein crystallization under controlled humidity. Microelectronic Engineering,
84:1758?1761, 2007.
[200] J.-B. Orhan, V.K. Parashar, J. Flueckiger, and M.A.M. Gijs. Internal modification of
poly(dimethylsiloxane) microchannels with a borosilicate glass coating. Langmuir, 24
(16):9154?9161, 2008.
[201] Hyungsuk Lee and Junghyun Cho. Parylene-pdms bilayer coatings for microelectronic
and mems packaging. Mater. Res. Soc. Symp. Proc., 969, 2007.
207
[202] Dhananjay Bodas and Chantal Khan-Malek. Formation of more stable hydrophilic
surfaces of PDMS by plasma and chemical treatments. Microelectronic Engineering,
83:1277?1279, 2006.
[203] Mark A. Eddings, Micheal A. Johnson, and Bruce K. Gale. Determining the optimal
PDMS-PDMS bonding technique for microfluidic devices. J. Micromech. Microeng.,
18:067001, 2008.
[204] M.A. Unger, H.-P. Chou, T Thorsen, A. Scherer, and S.R. Quake. Monolithic mi-
crofabricated valves and pumps by multilayer soft lithography. Science, 288:113?116,
2000.
[205] David T. Eddington, John P. Puccinelli, and D. J. Beebe. Thermal aging and reduced
hydrophobi recovery of polydimethylsiloxane. Sensors and Actuators B, 114:170?172,
2006.
[206] R.A. Fava. Method of Experimental Physics. Academic Press, 1980.
[207] Zoltan Filus, Tibor Ajtai, Zoltan L. Horvath, Zoltan Bozoki, Gabor Pap, Tibor Nagy,
and Gabor Szabo. A novel apparatus based on a photoacoustic gas detection system for
measuring permeation parameters of polymer samples. Polymer Testing, 26:606?613,
2007.
[208] M. Rezac, T. John, and P.H. Pfromm. J. Appl. Polym. Sci., 65:1983, 1997.
[209] E. Favre, P Scheatzel, Q.T Nguyen, Clement R., and J. Neel. Sorption, diffusion
and vapor permeation of various penetrants through a dense poly(dimethylsiloxane)
membranes- a transport analysis. J. Membrane Science, 92:169?184, 1994.
[210] Ahmed Ismail, Gary S. Grest, David R. Heine, and Mark J. Stevens. Interfa-
cial structure and dynamics of siloxane systems: PDMS-vapor and PDMS-water.
Macromolecules, 42:3186?3194, 2009.
[211] B. Chowdhury. Treatment of anomalous vapor diffusion in polymers. Journal of
Thermal Analysis and Calorimetry, 64:557?570, 2001.
[212] J.M. Watson and M.G. Baron. The behaviour of water in poly(dimethylsiloxane). J.
Membrane Sci., 110:47?57, 1996.
[213] Adam Abate, Daeyeon Lee, Thao Do, and Christian Holtze. Glass coating for PDMS
microfluidic channels by solgel methods. Lab Chip, 8:516?518, 2008.
[214] Arrelaine A. Dameron, Stephen D. Davidson, and Beau B. Burton. Gas diffusion
barriers on polymers using multilayers fabricated by Al2O3 and rapid SiO2 atomic
layer deposition. J. Phys. Chem. C, 112:4573?4580, 2008.
[215] Singh B., Bouchet J., Rochat G., and Leterrier Y. Ultra-thin hybrid organic/inorganic
gas barrier coatings on polymers. Surface & Coatings Technology, 201:7107?7114,
2007.
208
[216] Meng. Meng Ouyang, Peter P. Klemchuk, and Jeffrey T. Koberstein. Exploring the
effectiveness of SiOx coatings in protecting polymers against photo-oxidation. Polymer
Degradation and Stability, 70:217?228, 2000.
[217] A.G. Erlat and R.J. Spontak. SiOx gas barrier coatings on polymer substrates: Mor-
phology and gas transport considerations. J. Phys. Chem. B., 103:6047?6055, 1999.
209