Morphology and Properties of Anti-Corrosion Organosilane Films

Home , Silane

UNIVERSITY OF CINCINNATI

Date:______

I, ______, hereby submit this work as part of the requirements for the degree of: in:

It is entitled:

This work and its defense approved by:

Chair: ______

Morphology and Properties of Anti-Corrosion Organosilane Films

A dissertation submitted to the

Division of Research and Advanced Studies

of the University of Cincinnati

in partial fulfillment of the

requirements for the degree of

DOCTOR OF PHILOSOPHY

in the Department of Chemical and Materials Engineering

of the College of Engineering

2006

Guirong Pan

B.S., Tongji University, P. R. China 1999

M.S. University of Cincinnati, Ohio, 2003

Committee Chair: Dr. Dale W. Schaefer Abstract

Although it is known that certain organosilanes can dramatically improve the corrosion resistance when deposited on metals, the origin of this effect and its dependence on film characteristics are not fully understood. In this work, the morphology and structure of the silane films, as well as their response to water exposure, are studied mainly by neutron reflectivity.

Hydrothermal conditioning and solvent swelling are used to challenge the films. The silanes studied include bis-[triethoxysilylpropyl]tetrasulfide (bis-sulfur) and bis-

[trimethoxysilylpropyl]amine (bis-amino) as well as mixed silane films. Initial studies were done on films spin-coated on silicon wafer substrates from 1% solutions and cured at 80 °C. Here the focus is the effect of the bridging group on the morphology and water-barrier properties of the films. Subsequent work addresses the same systems deposited on aluminum substrates, films cured at 180 °C and films of larger thickness. The goal is to clarify the relationship between silane molecular structure, processing variables, morphology and water-barrier properties of films while developing a database for optimizing the performance in anti-corrosion applications.

Bridging group is the key factor that controls the morphology and water-barrier properties of silane films. Bis-sulfur silane is not as condensed as bis-amino silane, but it swells less in water

because of the hydrophobic nature of bridging group. By contrast, bis-amino film is more

hydrophilic since the secondary amine group hydrogen bonds with water.

The bulk mixed silane film swells with water to an extent that is slightly less than that of both

components weighted by their volume fraction. But, based on the enhanced shrinkage that occurs

upon water conditioning of the mixed film, condensation is accelerated in the mixed silane film.

Bis-amino silane, may act as a catalyst in the hydrolysis of bis-sulfur silane leading to more

ii silanols groups in the solution, which in turn will improve the wettability of the solution. This effect might explain the superior performance of the mixed film compared to pure bis-sulfur silane film.

Processing variables also impact the water-barrier performance. For bis-sulfur silane films, both larger thickness and higher cure temperature are critical for effective water-barrier properties.

iii iv

Acknowledgements

My deepest appreciation goes to my advisor, Dr. Dale W. Schaefer. I thank him for supervision, recommendations and suggestions during all these years involved in this challenging project. It is my great pleasure and honor to be Dr. Schaefer’s student.

I extend my deep appreciation to Dr. Wim Van Ooij, for all his guidance and insights. I thank him for spending time serving as my committee.

Sincere appreciation and gratitude are also to Dr. Greg Beaucage and Dr. Ray Y.

Lin for their participation as my committee and their valuable insights and suggestions.

Specially, I would like to thank Dr. Mike Kent at Sandia National Lab. Dr. Kent gave me enormous help with neutron reflectivity testing and data analysis. His guidance and advice is greatly appreciated.

This project is funded by the Strategic Environmental Research and

Development Program (www.serdp.org).

This work benefited the use of Surface Profile Analysis Reflectometer (SPEAR) at Lujan Neutron Scattering Center at Los Alamos National Laboratory, NG7 reflectometer at National Institute of Standards and Technology (NIST) as well as

POSYII at Argonne National lab. I thank Jaraslaw Majewski, Erik Watkins, Sushil Satija,

v Young-Soo Seo, Rick Goyette and Jan Ilavsky for their effort in collecting the reflectivity data.

I thank Professor James Boerio for use of the ellipsometer and AFM.

I thank my lab mates (past and present: Kumar, Tingtai, Jian, Kim, Dazhi,

Chetan, Kevin, Ryan, Yimin, Peng), members of SERDP group (Trilok, Chetan, Akshay) for their advice and help.

I appreciate the support provided by the dedicated and enthusiastic staff members of Chemical and Materials Engineering Department and Chemistry

Department.

I would like to thank my dear friend Barbara M. Linder for providing her home for me to stay, and for her encouragement and support.

To my friends Quanyan, Dan Wu, Li Guo, Bin Zheng, Feng Gao, Xuandong, Li

Yuan, Shu Zheng who made my years at Cincinnati a pleasant experience.

Special thanks to my husband Zhengsheng Li, for being there for me, for his encouragement, love and understanding throughout my studies. I am grateful to my family in China for their love, support and inspiration in all my endeavors. This work is dedicated to each and every of them.

This dissertation is dedicated to

My dear husband, parents and sister, for their love, caring, and support

through the past years.

vii Table of Contents

ABSTRACT ...... II

ACKNOWLEDGEMENTS...... V

TABLE OF CONTENTS...... 1

LIST OF FIGURES...... 7

LIST OF TABLES ...... 15

LIST OF ABBREVIATIONS...... 16

CHAPTER 1. INTRODUCTION...... 18

1.1 Research Significance and Objectives...... 18

1.2 Dissertation outline ...... 20

CHAPTER 2. LITERATURE REVIEW ...... 23

2.1. Corrosion Protection of Metals by Organosilane films...... 23

2.1.1 Chemical structure of organosilanes...... 23

2.1.2. Silane solution chemistry: hydrolysis and condensation ...... 24

2.1.3 Processing Variables of Silanes in Solution ...... 28

2.1.3.1 pH effect:...... 28

2.1.3.2 Silane Concentration ...... 29

1 2.1.3.3 Hydrolysis Time...... 30

2.1.3.4 The effect of metal substrate ...... 32

2.1.3.5 The Effect of Curing Step ...... 34

2.1.4 Structure Property Relationships in Corrosion-Inhibiting Films...... 35

2.1.4.1 Interaction of silane with substrate...... 35

2.1.4.2 Interaction of water with silane...... 40

2.2. Neutron reflectivity...... 41

2.2.1 History of Neutron (X-ray) reflectivity ...... 41

2.2.2 Theory of Neutron (X-ray) reflectivity...... 43

2.2.2.1 Refractive Index ...... 43

2.2.2.2 Snell’s law and Fresnel’s law...... 45

2.2.2.3 Reflectivity from a system with one interface...... 48

2.2.2.4 Reflectivity from a system with two parallel interfaces...... 50

2.2.2.5 Analysis of Reflectivity...... 51

2.2.2.6 Non perfect layers and Practical problems:...... 54

2.2.3 Instrumentation for reflectivity...... 59

2.2.4 Application of neutron reflectivity in the study of silane films...... 61

CHAPTER 3. EXPERIMENTAL ...... 64

3.1. Sample Preparation ...... 64

3.1.1 Materials ...... 64

3.1.2 Procedure:...... 65

3.1.2.1 Silane solution preparation...... 65

3.1.2.2 Silane film formation ...... 66

3.2 Characterization of Silane films ...... 67

2 3.2.1 X-ray reflectivity ...... 67

3.2.2 Neutron reflectivity...... 68

3.2.2.1 Test procedure...... 68

3.2.2.2 Data acquisition and analysis ...... 70

3.2.3 Variable angle spectroscopic ellipsometry ...... 74

3.2.4 SEM...... 76

3.2.5 Contact angle ...... 76

3.2.6 Atomic force Microscopy...... 77

CHAPTER 4. NEUTRON REFLECTIVITY INVESTIGATION OF BIS-AMINO SILANE

FILMS ...... 78

4.1 Introduction...... 78

4.2: Results and Discussion ...... 80

4.2.1 Bis-amino silane under d-NB swelling...... 80

4.2.2 Bis-amino silane under hydrothermal conditioning...... 87

4.2.2.1 Exposure to D2O at room temperature for 14h...... 87

4.2.2.2 Exposure to D2O at 80 °C for 14 h...... 90

4.3 Conclusions...... 93

CHAPTER 5. WATER BARRIER PROPERTIES OF BIS-SULFUR AND MIXED

SILANE FILMS ...... 94

5.1. Introduction...... 94

5.2. Results and Discussion...... 96

3 5.2.1 Reflectivity data of the as-prepared Film ...... 96

5.2.2 Bis-sulfur silane...... 99

5.2.2.1 Nitrobenzene conditioning ...... 99

5.2.2.2 D2O conditioning...... 101

5.2.3 Mixed bis-sulfur/bis-amino silane film ...... 108

5.2.3.1 NB conditioning ...... 108

5.2.3.2 D2O conditioning...... 109

5.3. Conclusion ...... 114

CHAPTER 6. MORPHOLOGY AND WATER-BARRIER PROPERTIES OF SILANE

FILMS: THE EFFECT OF SUBSTRATE ...... 116

6.1. Introduction...... 116

6.2. Results and Discussion...... 118

6.2.1 Al layer on silicon wafer...... 119

6.2.2 Silane on Al-coated silicon wafer...... 122

6.2.2.1 Room temperature conditioning...... 122

6.2.2.2 80-°C conditioning ...... 131

6.3 Conclusion ...... 137

CHAPTER 7. MORPHOLOGY AND WATER BARRIER PROPERTIES OF SILANE

FILMS: THE EFFECT OF CURING TEMPERATURE...... 139

7.1 Introduction...... 139

4 7.2 Results and Discussion...... 142

7.2.1 Neutron and X-ray reflectivity...... 142

7.2.1.1 Bis-sulfur silane...... 142

7.2.1.2 Bis-amino silane...... 147

7.2.1.3 Mixed silane ...... 149

7.2.2 Ellipsometry...... 151

7.2.3 Contact Angle...... 151

7.3 Conclusions...... 152

CHAPTER 8. MORPHOLOGY AND WATER BARRIER PROPERTIES OF SILANE

FILMS: THE EFFECT OF THICKNESS...... 154

8.1 Introduction...... 154

8.2 Results and Discussion...... 155

8.2.1 Thickness test by ellipsometry...... 155

8.2.2 Water Conditioning ...... 159

8.2.2.1 Si Wafer Substrate and 80 °C Curing Temperature ...... 160

8.2.2.2 Al substrates ...... 165

8.2.2.3 180-°C Cure ...... 167

8.3 Conclusions...... 170

CHAPTER 9. GENERAL CONCLUSIONS AND SUGGESTED FUTURE WORK .. 172

9.1 General Conclusions ...... 172

5 9.2 Suggested Future Work...... 174

REFERENCES...... 176

6 List of Figures

Figure 2.1.3-1: Schematic comparison of the rates of hydrolysis and condensation in silicates.42 ...... 29

Figure 2.1.3-2: Evolution of the thickness (determined by SE and TEM) and sputter time (from AES depth profiles) as a function of the BTSE bath concentration.26...... 30

Figure 2.1.3-3. Proton NMR spectra of a 1% γ-GPS solution in deuterium oxide showing the peaks due to the methoxy protons of the silane and the methyl protons of methanol after hydrolysis times of: (A) 5 min.; (B) 14 min.;(C) 34 min. (from 22)...... 32

Figure 2.1.3-4. Si-29 NMR spectra of a 10% γ -GPS solution in water at hydrolysis times of: (A) 2 hours; (B) 9 hours; (C) 20 hours. (from 22)...... 32

Figure 2.1.3-5: SEM micrographs measured from a 2024-T3 alloy sample which had been coated with (a) BTSE (b) γ-APS immediately after polishing.6...... 33

Figure 2.1.4-1: High mass resolution ToF-SIMS spectra from the silane-treated Al substrate at the nominal mass m/z = 71. (a): silane-coated substrate before sputtering; (b): silane-coated substrate after 75 s. 16...... 37

Figure 2.1.4-2: Model of possible adsorption of silane molecules on metallic surfaces. (a) Surface bonding via the functional group Y; (b) surface bonding via the functional and the condensed silanol groups; (c) surface bonding via the interaction between the condensed silanol group and the metallic surface. 27 ...... 38

Figure 2.1.4-3: N (1s) high resolution spectra of polished 1100 aluminum coated with γ-APS from a 1% aqueous solution at pH 10.4 with take off angle 75°.25 ...... 39

Figure 2.1.4-4: The two-layered model used for quantifying water at the organic film/hydroxylated substrate interface by Tinh Nguyen et al.49...... 41

Figure 2.2.2-1: Geometry of scattering (and reflection) from surface...... 45

7 Figure 2.2.2-2: (a) Calculated Frenel neutron reflectivity profile from an infinitely sharp interface

between two media obtained from equation 2.2.2-16. Total reflection (R = 1) occurs for q < qc. 4 (b) Reflectivity multiplied by q , showing the asymptotic limit reached a large value of qz...... 50

Figure 2.2.2-3: Plane waves incident on a surface spatially separated by a distance ∆x impinge upon the surface of a specimen at a distance ∆x / sin θ apart...... 55

Figure 2.2.2-4: Effect of δλ and δα Comparison between a perfect instrument, and instrumental δλ, and δα for a measurement on a single 30 nm thick layer on a substrate. (From 68, p186)...... 56

Figure 2.2.2-5: Rough interface in which the radii of curvature are (a) larger than the coherence 62 length lc and (b) smaller than lc. (From , p. 256)...... 57

Figure 2.2.2-6: Reflectivity calculated for a deuterated polystyrene film of thickness 500 Å deposited on a Si substrate, to illustrate the effect of diffuse interface and small-scale roughness.

Thin solid curve, σ1 = σ2 = 0; thin broken curve, σ1 = 20 Å, σ2 = 0; thick broken curve, σ1 = 0, σ2 = 20 Å. (From 62, p. 255) ...... 59

Figure 2.2.4-1: (a) Neutron reflectivity data from a 0.3% GPS sample as-prepared ( ) and after swelling to equilibrium with d-NB (·). (b) Best-fit scattering length density profiles corresponding to the curves through the data in (a) for: as-prepared (-), after swelling (···). (from 44)...... 62

Figure 3.1-1: Molecular structure of (a) bis-sulfur silane; (b) bis-amino silane; (c) BTSE ...... 65

Figure 3.2.2-1: Experimental set up for “in situ” measurement of neutron reflectivity...... 68

Figure 3.2.2-2: Digital picture of the Al sample holder...... 69

Figure 3.2.2-3: Schematic diagram of SPEAR...... 70

Figure 3.2.2-4: NG 7 reflectometer at NIST...... 72

Figure 3.2.3-1: Geometry of an Ellipsometric Measurement.85...... 75

Figure 4-1: (a) Neutron reflectivity data from bis-amino silane film as-prepared, after swelling to equilibrium with d-NB vapor and after re-drying. The curves through the data points correspond

8 to the best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). “Rough” in the legend means roughness was included to fit the data. The film was found to be rough in the swollen states. The calculated volume fraction of d-NB in the

swollen film is φNB = 17%. After redrying, about 5% d-NB remained in the film. (The data were generated on SPEAR at LANL)...... 81

Figure 4-2: (a) Neutron reflectivity data from bis-amino silane film as-prepared, after exposure to

D2O vapor at room temperature for 14 h and after re-drying. The curves through the data points correspond to the best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). “Rough” in the legend means roughness was included to fit the data. The film was found to be rough in the swollen state. The calculated volume fraction of D2O in the swollen film is φD2O = 41%. (The data were generated on SPEAR at LANL)...... 88

Figure 4-3: Schematics of proposed structures of bis-amino silane film...... 90

Figure 4-4: (a) Neutron reflectivity data from bis-amino silane film as-prepared, after exposure to

D2O vapor at 80 °C for 14 h and after re-drying. The curves through the data points correspond to the best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (The data were generated on SPEAR at LANL)...... 92

Figure 5-1: Comparison of reflectivity curves for as-prepared bis-amino, bis-sulfur and mixed silane. For clarification, reflectivity curve of mixture silane is suppressed by 102 and reflectivity curve of bis-sulfur silane is suppressed by 104. (NR was performed on the SPEAR at LANL) .. 96

Figure 5-2: AFM images of (a) bis-amino silane film; (b) bis-sulfur silane film; (c) Mixed silane film...... 98

Figure 5-3: (a) Neutron reflectivity data from bis-sulfur silane film as-prepared and after swelling to equilibrium with d-NB. The curves through the data points correspond to the best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a) for the samples as-prepared and after swelling to equilibrium with d-NB. The

calculated volume fraction of d-NB in the swollen film is ϕNB = 5.9%. (NR was performed on the SPEAR at LANL)...... 99

9 Figure 5-4 (a): Neutron reflectivity data from bis-sulfur silane film as-prepared (○), after

exposure to D2O vapor at room temperature for 15 h (+) and after re-drying (∆). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a) The calculated volume fraction of

D2O in the swollen film is 7.8%. (NR was performed on the SPEAR at LANL)...... 103

Figure 5-5: (a) Neutron reflectivity data from bis-sulfur film as-prepared (●), after exposure to

D2O vapor at 80 °C for 14 h (+) and after re-drying (∆). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit scattering length density profiles corresponding to the curves through the data in (a). The volume fraction of free water in the swollen film is 7.2%. (NR was performed on the SPEAR at LANL)...... 105

Figure 5-6: SLD profiles of bis-sulfur silane as prepared (—), exposure to D2O vapor at room temperature for 15 hours (····), exposure to D2O at 80 °C for 14 hours (----), and exposure to D2O vapor at 80 °C for 2 days (−···−···). (NR was performed on the SPEAR at LANL)...... 106

Figure 5-7: (a) Neutron reflectivity data from mixed silane film as-prepared and after swelling to equilibrium with d-NB. The curves through the data points correspond to the best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the

data in (a). The calculated volume fraction of d-NB in the swollen film is φNB = 5.2%. (NR was performed on the SPEAR at LANL)...... 109

corresponding to the curves through the data in (a). The calculated volume fraction of D2O in the swollen film is 12.6%. (NR was performed on the SPEAR at LANL)...... 110

Figure 5-9: (a) Neutron reflectivity data from mixed silane film as-prepared (●), after exposure to D2O vapor at 80 °C for 11 h (+) and after re-drying (∆). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding

to the curves through the data in (a) for the samples as-prepared (—), after exposure to D2O vapor at 80 °C for 11 h (····) and after re-drying (−···−···). The volume fraction of free water in the swollen film is 13%. (NR was performed on the SPEAR at LANL)...... 112

10 Figure 6-1: The systems studied and the corresponding representative SLD profiles...... 118

Figure 6-2: The reflectivity curves of silane on silicon wafer, Al on silicon wafer and silane on Al-coated silicon wafer. Bis-amino silane films are shown as representatives. For clarification, reflectivity curve of silane on Al coated Si is suppressed by 102 and reflectivity curve of silane on Si is suppressed by 104. (NR data were generated on the NG7 reflectometer at NIST)...... 119

Figure 6-3 a: POSY NR data from Al layer on silicon wafer as-prepared and after exposure to

D2O vapor at room temperature for 24 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the POSYII reflectometer at ANL)...... 121

Figure 6-4: ESEM image of Al layer on silicon wafer. The bar is 200 nm...... 122

Figure 6-5(a): Neutron reflectivity data from bis-amino silane films on Al-coated silicon wafer

as-prepared and after exposure to D2O vapor at room temperature for 23 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST)...... 124

Figure 6-6(a): Neutron reflectivity data from mixed silane films on Al coated silicon wafer as-

prepared and after exposure to D2O vapor at room temperature for 23 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST)...... 127

Figure 6-7(a): Neutron reflectivity data from bis-sulfur silane films on Al coated silicon wafer

Figure 6-8: Neutron reflectivity data from (a) mixed silane film; (b) bis-amino silane film; (c)

bis-sulfur silane film as-prepared (○), and re-drying (∆) after exposure to D2O vapor at room temperature. (NR data were generated on the NG7 reflectometer at NIST)...... 131

11 Figure 6-9: Neutron reflectivity data from (a) mixed silane film; (b) bis-amino silane film; (c)

bis-sulfur silane film as-prepared (○), and redried (∆) after exposure to D2O vapor at 80 °C. (NR data were generated on the NG7 reflectometer at NIST)...... 133

Figure 6-10(a): Neutron reflectivity data from amino silane films on Al coated silicon wafer as- prepared and after exposure to D2O vapor at 80 °C for 11 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST)...... 135

Figure 6-11(a): Neutron reflectivity data from mixed silane films on Al coated silicon wafer as-

prepared and after exposure to D2O vapor at 80 °C for 11 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST)...... 136

Figure 6-12(a): Neutron reflectivity data from bis-sulfur silane films on Al coated silicon wafer as-prepared and after exposure to D2O vapor at 80 °C for 11 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST)...... 137

Figure 7-1 (a): Neutron reflectivity data from bis-sulfur silane film cured at 180 °C as-prepared

(○), after exposure to D2O vapor at room temperature (+). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (Data were taken on the SPEAR at LANL)...... 143

Figure 7-2: Possible reactions in the bis-sulfur silane films during 180-°C curing based on Tobolsky et al.’s work.99 Breakdown of tetrasulfide bonds into shorter linkage was achieved by bond interchange reactions. This process involves hemolytic scission of the tetrasulfide bond (reaction 1) followed by radical attack on neighboring polysulfide bond (reaction 2). Di-radical sulfur chain was released from the bond interchange reaction (reaction 3) and elemental sulfur was regenerated (reaction 4)...... 146

12 Figure 7-3: X-ray reflectivity of as-prepared bis-sulfur silane film cured at 80 °C and 180 °C. (X- ray data were taken on 1-BM beam-line at the Advanced Photon Source at ANL)...... 147

Figure 7-4 (a): Neutron reflectivity data from bis-amino silane film cured at 180 °C as-prepared

(○), after exposure to D2O vapor at room temperature (+). The curves through the data correspond to best fits using model SLD profiles in (b). (b): Best-fit SLD profiles corresponding

to the curves through the data in (a) for the samples as-prepared (—), after exposure to D2O vapor at room temperature (····). (Data were taken on the SPEAR at LANL)...... 149

Figure 7-5 (a): Neutron reflectivity data from mixed silane film cured at 180 °C as-prepared (○),

after exposure to D2O vapor at room temperature (+). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a) for the samples as-prepared (—), after exposure to D2O vapor at room temperature (····). (Data were taken on the POSYII at ANL)...... 150

Figure 8-1: The spectroscopic ellipsometry ψ (a) and ∆ (b) spectra taken at 60°, 65°, 70° and 75° for 1% bis-amino silane films on silicon wafer cured at 80 °C...... 156

Figure 8-2: The spectroscopic ellipsometry ψ (a) and ∆ (b) spectra taken at 60°, 65°, 70° and 75° for 5% bis-amino silane films on silicon wafer cured at 80 °C...... 157

Figure 8-3: Wavelength dependence of the refractive index of mixed silane at different thickness...... 159

Figure 8-4 (a): Neutron reflectivity data from 5% bis-sulfur silane film as-prepared (○), after

exposure to D2O vapor at room temperature. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). The calculated volume fraction of D2O in the swollen film is 11%. (Data were from SPEAR at LANL)...... 160

Figure 8-5 (a): Neutron reflectivity data from 5% bis-amino silane film as-prepared, after exposure to D2O vapor at room temperature with beam from air side. (Data were from POSYII at ANL)...... 161

13 Figure 8-6 (a): Neutron reflectivity data from 5% bis-amino silane film as-prepared (○), after

exposure to D2O vapor at room temperature with beam shooting from wafer side. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles

corresponding to the curves through the data in (a). The calculated volume fraction of D2O in the swollen film is 27%. (Data were from SPEAR at LANL)...... 163

Figure 8-7 (a): Neutron reflectivity data from 5% mixed silane film as-prepared (○), after exposure to D2O vapor at room temperature. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). The calculated volume fraction of D2O in the swollen film is 12%. (Data were from SPEAR at LANL)...... 164

Figure 8-8a: Neutron reflectivity data from 5% mixed silane on Al substrate as-prepared, after conditioned in D2O vapor. The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit scattering length density profiles corresponding to the curves through the data in (a). (Data were from SPEAR at LANL)...... 166

Figure 8-9 (a): Neutron reflectivity data from 5% bis-sulfur silane film cured at 180 °C as- prepared (○), after exposure to D2O vapor at room temperature. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (Data were from SPEAR at LANL)...... 168

Figure 8-10 (a): Neutron reflectivity data from 5% bis-amino silane film cured at 180 °C as-

prepared (○), after exposure to D2O vapor at room temperature. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (Data were from SPEAR at LANL)...... 170

List of Tables

Table 2.1.1-1: Silanes used for corrosion protection of metals...... 24

Table 3.2.2-1: Calculated neutron SLD for the materials used in this study...... 72

Table 4-1: Some possible chemical structures of hydrolyzed bis-amino silane and the calculated SLD...... 82

Table 7-1: Thickness of the as-prepared silanes cured at 80 °C and 180 °C measured by ellipsometry...... 151

Table 7-2: Contact angles of 3 silanes cured at different curing temperature...... 152

Table 8-1: Ellipsometry results of silanes prepared at different concentration and spin speed.. 158

List of Abbreviations

AFM Atomic Force Microscopy

AES Auger Electron Spectroscopy

ANL Argonne National Laboratory

γ-APS γ-aminopropyltriethoxysilane

Bis-amino Bis-[trimethoxysilylpropyl]amine

Bis-sulfur Bis-[triethoxysilylpropyl]-tetrasulfide

BTSE Bis-1,2-(triethoxysilyl)ethane

DI-water De-ionized water

D-NB Deuterated nitrobenzene

EIS Electrochemical Impedance Spectroscopy

FTIR Fourier transform infrared spectroscopy

γ-GPS γ-Glycidoxypropyltrimethoxysilane

HDG Hot-dip galvanized steel

IEP Isoelectric Point

IRSE Infrared Spectroscopic Ellipsometry

16 LANL Los Alamos National Laboratory

LV Linear Voltammeter

NIST National Institute of Standards and Technology

NR Neutron Reflectivity

RMS Root Mean Square

SERDP Strategic Environmental Research and Development Program

SEM Scanning Electron Microscopy

SPEAR Surface Profile Analysis Reflectometer

SLD Scattering Length Density

TEM Transmission Electron Microscopy

ToF-SIMS Time of Flight Secondary Iron Mass Spectra

VOC Volatile Organic Compound

VCA Video Contact Angle

WSA Weak Scattering Approximation

XPS X-ray Photoelectron Spectroscopy

17 Chapter 1. Introduction

1.1 Research Significance and Objectives

Organosilane films have been studied extensively as corrosion inhibitors in recent years.1-3 Of particular interest are bis-type silanes with the structure X3Si(CH2)3-R'-(CH2)3SiX3. In this

formula, R' is the “organic bridging group.” Bis-silanes with six hydrolyzable groups perform

effectively on a range of metals in various corrosion tests.1, 4-8 The work performed here supports a SERDP-funded project to develop an integrated organosilane metal-coating system that can replace the current chromate pretreatment and primer with no sacrifice of corrosion performance.

The proposed system is based on a University of Cincinnati (UC) invention termed

“superprimer.” This superprimer, consisting of silane, organic resin and nanoparticle filler, is totally water-borne and is amenable to dipping, spraying wiping or brushing onto any clean metal surface. No conversion coating, either phosphate or chromate, is required. The key innovation underlaying this technology is the hydrophobicity transition exhibited by organosilanes. Therefore, during the first stage of the project, the focus is the fundamental studies of the relationship between synthesis, processing, structure and properties of silane films.

The formation of protective silane films involves numerous chemical and physical processes.

Optimization of silane films evokes a number of seemingly contradictory requirements. The film precursors must be hydrophilic to permit water-borne deposition. The film itself, however, must be very hydrophobic to assure superior protection. To optimize film performance in the face of these competing requirements, a through understanding of the film formation is required.

Silane corrosion inhibitors are not as well understood as are silane adhesion promoters. Some

conclusions, however, can be drawn based on recent corrosion research.1, 6, 9-11 One important

18 conclusion is that silane films function as physical water barriers, exhibiting no electrochemical effect. To improve the corrosion performance, therefore, the foremost objective is to improve the water resistance. Understanding of water-silane interaction is essential to achieve this objective.

The major objective of this study is to elucidate the morphology of silane films in order to

understand the interaction of water with silane films and to quantify the effect of processing

variables on the water-barrier properties. Our goal is to clarify the relationship between silane

molecular structure, processing parameters, and film morphology and water-barrier properties while developing a database for optimizing the performance in anti-corrosion applications.

The silanes we studied are bis-[triethoxysilylpropyl]tetrasulfide (bis-sulfur) and bis-

[trimethoxysilylpropyl]amine (bis-amino) as well as mixtures of these silanes. These silanes

show contrasting behavior traceable to the bridging group. Performance tests show that bis-

amino silane does not offer good corrosion protection on either aluminum alloys or on hot-dip

galvanized steel (HDG). The more hydrophobic bis-sulfur silane, on the other hand, performs

well on aluminum alloys, but not steel. Interestingly, a bis-sulfur/bis-amino (3/1) mixture shows

greatly enhanced corrosion resistance compared to the two individual silanes, and provides

protection for many metals including Al alloys and HDG.12 In this work, we clarify the

mechanism underlying these performance results by investigating the structure of these silane

films and their response to water vapor.

We use neutron reflectivity (NR) to study the morphology of pristine and water-conditioned

silane films. NR can easily examine interfaces that are buried well within a sample as a

consequence of the weak interaction of neutrons with almost any material. NR is sensitive to

different isotopes allowing tracking of specific components by isotope substitution. The

19 technique is non-destructive, so prolonged exposure and repeated measurements are possible. To

track water penetration, heavy water (D2O) is used. NR data were obtained for the films as- prepared, after exposure of the film to saturated D2O vapor at both room temperature and 80 °C,

and again after re-drying. Measurements in the redried state determine if there are chemical

changes in the film.

Extensive research has been performed on silane films with various techniques, including

AFM,13 contact angle,14, 15 SIMS,16-20 FTIR,10, 21, 22 XPS,23-25 ellipsometry,26-28 EIS,2, 29-32 et al.

The silane film surface structure, interface bonding to metal substrate, hydrolysis and

condensation kinetics, film thickness as well as performance have been well reported. These

methods, however, do not address the key issues related to water penetration. Using neutron

reflectivity, we are able to detect the distribution of water in the film and the response of film to

water penetration in situ.

1.2 Dissertation outline

This dissertation summarizes the study on the morphology and structure of silane films as well

as the response to water vapor mainly by neutron reflectivity. This dissertation consists of three

sections: (1) Review of previous work on silane surface treatments and theory of neutron

reflectivity; (2) Results on baseline systems of the three silanes cured at 80 °C on Si wafer with

1% concentration; (3) Study of processing variables including Al substrates, 180 °C curing

temperature and thicker films.

Chapter 2: Previous work on silane surface treatment as well as the theory of neutron

reflectivity is reviewed.

20 Chapter 3: Sample preparation and characterization procedures as well as techniques are discussed.

Chapter 4: The structure and morphology of bis-amino silane film undergoing hydrothermal conditioning and solvent swelling are reported. The films were spin-coated on silicon wafer substrates from 1% solutions and cured at 80 °C. The results are compared with bis-

[triethoxysilyl]ethane (BTSE) and issues regarding to crosslink density and hydrophobicity are

discussed.

Chapters 5: Bis-sulfur silane films as well as the mixture of bis-amino and bis-sulfur were

prepared and investigated following the same procedure as bis-amino silane film in Chapter 4.

The extent of hydrolysis and condensation is inferred from the NR data and a mechanism of

corrosion protection is proposed

Chapter 6: An Al layer was e-beam evaporated on a silicon wafer. Silane films were then

applied to the Al surface using the same spin-coating technique. Neutron reflectivity was carried

out to study the effect of substrate on the water-barrier properties.

Chapter 7: Films were cured at 180 °C and characterized following the same procedures as

films cured at 80 °C. The purpose is to understand the effect of curing temperature on the

morphology and water-barrier properties of silane films.

Chapter 8: Films with thicknesses approaching practical levels are investigated in this chapter.

Both higher concentration and lower spin speed are used to obtain thicker films. Three silane

systems were investigated following the same procedure as was used for thin films. Each film

21 was tested as-prepared and again after exposure to water vapor at room temperature. Studies on thick films applied on Al substrate and cured at higher curing temperature are reported as well.

Chapter 9: General conclusions are summarized and future work is discussed.

22 Chapter 2. Literature Review

2.1. Corrosion Protection of Metals by Organosilane films

2.1.1 Chemical structure of organosilanes

Organofunctional silanes are hybrid organic-inorganic compounds, functionalized on the

organic group and bearing hydrolyzable groups on silicon. Commercial use of organosilanes has

developed steadily since the 1950’s.8 The major uses may be categorized as “coupling agents” and “crosslinkers.” However, recent work proposes the use of organosilanes as corrosion inhibitors for different substrates.1, 2, 5, 6, 27, 29, 33 This development is mostly a response to the

quest for an alternative to conventional chromating processes in metal-finishing industries.

Chromates are still the most efficient within the current repertoire of corrosion inhibitors.

However, during the formation of the chromate conversation layer, Cr (VI), which is highly toxic, is released. As a result, alternative surface pre-treatments have been developed to replace chromate treatment. In comparison to the other candidates, the performance of silanes is more promising, and is comparable to that of chromates in some cases.1

There are two categories of silanes studied for corrosion protection, mono-silylfunctional

silane (monosilane) and bis-silylfunctional silane (bis-silane). Mono silanes have the general

structure X3Si (CH2)n Y, where X is a group that can be hydrolyzed, typically ethoxy

(-OCH2CH3) or methoxy (-OCH3), and Y is an organofunctional group, such as chlorine, amine,

epoxy or mercapto. Mono silanes are common silane coupling agents used to promote adhesion

between mineral phases and organic phases. Bis silanes have been mainly used as crosslinkers.

Bis silanes have the general structure of (RO)3Si-R’-Si(OR)3, with an additional silylfunctional

23 group instead of the organic functional end group. Bis silanes with six hydrolyzable groups are believed to produce more substrate bonding and higher bulk crosslink density, resulting in better corrosion performance. Examples of both silanes commonly used are listed in Table 2.1.1-1.

Performance tests demonstrated that, although mono-silanes protect painted metals to a certain extent, bis-silanes usually provide a better performance on a broader range of metals. The metals and alloys that are protected by bis-silanes include: Al and Al alloys, Fe and steels, Zn and Zn- coated steels, Cu and Cu alloys, and Mg and Mg alloys. 12, 13, 15, 30, 34-36

Table 2.1.1-1: Silanes used for corrosion protection of metals. Name (Abbreviations) Chemical Structures Mono-silanes

(3-glycidoxypropyl) trimethoxysilane (GPS) CH2CHCH2O(CH2)3Si(OCH3)3

vinyltriethoxysilane (VS) CH2=CHSi(OC2H5)3

γ-aminopropyltriethoxysilane (γ-APS) H2N(CH2)3Si(OCH3)3 Bis-silanes

bis[3-(triethoxysilyl)propyl]tetrasulfide (bis- (H5C2O)Si-(CH2)3-S4-(CH2)3-Si(OC2H5) sulfur silane)

bis-[trimethoxysilylpropyl]amine (H3CO)Si-(CH2)3-NH-(CH2)3-Si(OCH3). (bis-amino silane)

bis-[triethoxysilyl]ethane (BTSE) (C2 H5O)3Si(CH2)2Si(OC2H5)3

2.1.2. Silane solution chemistry: hydrolysis and condensation

The idea of using silanes as coupling agents is based on the hydrolysis and condensation reaction of silane to form oxane bonds between the silane and metal substrate as proposed by

Plueddemann.8

In solution: the silanes hydrolyze to generate active silanols.

24 OR OH

R' (CH ) Si OR + 3 H O R' (CH2) Si OH + 3 ROH 2 n 2 n OR OH

When applied to the metal surface silanol groups can react with each other to form Si-O-Si bonds leading to self-condensation.

+ H2O

The kinetics and equilibrium of hydrolysis and condensation of silanes in solutions are influenced by factors such as the nature of the organofunctional groups, the solution pH, concentration of silanes and water, aging time and temperature.8, 37-40 The pH is often the most important factor in determining the behavior of a particular silane regarding hydrolysis and condensation.

The hydrolysis of alkoxy groups of organotrialkoxysilanes in their aqueous solution proceeds in a stepwise manner, as shown in equations (2.1.2-1)-(2.1.2-3):

k 1 + ROH 2.1.2-1 R'Si(OR)3 + H2O R'Si(OR)2OH k-1

k 2 + 2.1.2-2 R'Si(OR)2OH + H2O R'Si(OR)(OH)2 ROH k-2

k 3 2.1.2-3 R'Si(OR)(OH)2 + H2O R'Si(OH)3 + ROH k-3

Pohl et al.38 monitored each step of the hydrolysis reaction of vinyltrialkoxysilane solution under both basic and acidic conditions using an NMR technique. They found that the amount of

25 silanol and silanediol intermediates remains small and that the ratios of the rate constants k1/k2

and k1/k3 are small. The first step of the hydrolysis is, therefore, the slow step and thus becomes

the rate-determining step in the process of silane hydrolysis.

Osterholtz and Pohl reviewed the studies on kinetics of hydrolysis and condensation of neutral

organofunctionl alkoxysilanes.38 Two equations for the kinetics of hydrolysis and condensation

in aqueous and aqueous-organic silane solutions were developed. The kinetics of hydrolysis is

expressed in equation 2.1.2-4

− d[S] = k [H O]n [S] + k [H + ][H O]m [S] + k [HO − ]o [H O] p [S] + k [B][H O]q [S] dt spon 2 H 2 HO 2 B 2

Where the symbol S represents the silane ester, -d[S]/dt is hydrolysis rate and B represents any basic species other than hydroxide anion. The first term on the right is the spontaneous reaction rate without catalysis. This term is not of great interest because spontaneous hydrolysis occurs so slowly that the corresponding contribution to the observed rate of hydrolysis is negligibly small under most conditions.

Of interest here are the second, third and fourth terms on the right, which describe the contributions of acids and bases to silane hydrolysis. Both acids and bases accelerate silane hydrolysis but in different ways. It was stated38 that “acid-catalyzed hydrolysis involved attack

on an alkoxy oxygen by a hydronium ion followed by bimolecular SN2-type displacement of the leaving groups by water. Alkali-catalyzed hydrolysis involved attack on silicon by a hydroxyl ion (OH–) to form a pentacoordinate intermediate followed by a bimolecular displacement of

alkoxyl by hydroxyl. Rates of hydrolysis by both mechanisms were influenced by the nature of

the alkyl group on silicon as well as the leaving alkoxyl group.”

26 A kinetic expression describing the rate of silanol condensation to disiloxane is given as

equation 2.1.2-5,

− d[S'] = k [H + ][S']2 + k [HO − ][S']2 + k [B][S']2 2.1.2-5 dt H HO B where S’ = SiOH groups, –d[S’]/dt = rate of silanol condensation, and B = any basic species other

than hydroxide anion. Both acids and bases catalyze the condensation reaction in silane

solutions. The pH effect on the reaction rates will be explained in detail in part 2.1.3.

In addition to pH, steric effects influence the hydrolysis and condensation rates.37, 40 The rate of hydrolysis of alkoxy groups is generally associated with their steric bulk: CH3O > C2H5O > t-

C4H9O. That is, the smaller the size of the alkoxy groups, the faster the hydrolysis rate. A

methoxysilane, for instance, hydrolyzes at 6-10 times the rate of an ethoxysilane.

Arkles et al. 41 found that the hydrolysis rate increases with the increase of organic substitution: Me3SiOMe > Me2Si(OMe)2 > MeSi(OMe)3. It was also reported that increasing size

of alkyl group on silicon increases the stability of intermediate silanols and the tendency to form

cyclic siloxanes upon condensation.8

It is seen in equation 2.1.2-5 that the rate of condensation is second order with respect to the silanol concentration. Thus, low silane concentration (or silanol concentration) should minimize condensation, thereby improving stability. The overall stability is a combined effect of the hydrolysis and condensation. Silanols are stable in a very dilute solution, where the silanols are isolated and stabilized by hydrogen bonding to water.

27 Hydrophilic organofunctional groups (amines) influence the stability of the silane solutions.

Plueddemann8 found that aminosilanes with three carbons between nitrogen and silicon gave

very good stability. A relatively concentrated γ-APS aqueous solution shows good stability, for

example. The phenomenon has been explained by the fact that bonding of silanol hydrogen to

amino groups inhibits the crosslinking between silanols.

2.1.3 Processing Variables of Silanes in Solution

The effect of process variables on the structure and properties of silane films has been studied

extensively, including pH, hydrolysis time, solution concentration, metal substrate, curing

temperature and curing time,.

2.1.3.1 pH effect:

The pH effects of the silane solution can be explained by considering the hydrolysis and condensation kinetics in the previous section. Osterholtz and Pohl38 concluded that the slowest rate of hydrolysis was obtained at neutral pH (around pH = 7), with a ten-fold acceleration in hydrolysis rate for unit change in pH in either the acid or the basic direction, assuming an excess of available water. So both acids and bases may be used to catalyze hydrolysis. Condensation is also pH-dependent. The minimum rate of condensation was obtained at around pH = 4. Recently, the kinetics of hydrolysis and condensation of bis-triethoxysilylethane (BTSE) has also been published by Pu et al.,39 which show a similar trend. The kinetics of hydrolysis and condensation

silanes are shown schematically in Figure 2.1.3-1 based on the results of tetraethoxy silanes

(TEOS) 42 under different pH conditions. Therefore, if fast hydrolysis followed by slow condensation is desired, acidic catalysis is preferred. If base catalysis is used, condensation to siloxane may occur even before all the alkoxy groups are hydrolyzed.

Figure 2.1.3-1: Schematic comparison of the rates of hydrolysis and condensation in silicates.42

2.1.3.2 Silane Concentration

The silane concentration is one of the major factors governing the film. It is found that in

almost all combinations of metals and silanes, a straight line relationship exists between film

thickness and concentration in a range that covers all practical concentrations.26, 27, 43 Franquet et al.26, 27 performed a systematic study of film thickness of BTSE coatings applied on aluminum

using spectroscopic ellipsometry (SE). To confirm the results, auger electron spectroscopy

(AES) depth profiles and transmission electron microscopy (TEM) micrographs of the uncured

silane-coated Al samples were obtained. From AES, the sputtering time needed to remove the

silane film was determined, which is related to the film thickness. All results are shown in Figure

2.1.3-2. Franquet et al. also found that the film becomes more porous and brittle as the thickness

increases. But performance tests show that the minimum thickness of the silane film for

corrosion protection of bare metal is 2500 Å.1 These results imply contradictory requirements for optimizing silane film. Silanes films must be thin enough to be pore free yet thick enough to provide an adequate barrier to water penetration. Therefore, concentration management is

29 critical. And when silane is used alone, the challenge is to build thick film without pores and

brittleness.

Figure 2.1.3-2: Evolution of the thickness (determined by SE and TEM) and sputter time (from AES depth profiles) as a function of the BTSE bath concentration.26

2.1.3.3 Hydrolysis Time

Regarding hydrolysis time (the time after initial mixing with water and before the silane

solution is applied to substrate), it is widely accepted that there is an operational “window.” If

the hydrolysis time is too long, condensation of the silane in solution occurs before application to

substrate, decreasing the effectiveness of the film. If the hydrolysis time is too short, there are not enough silanol groups to condense after application to substrate, also decreasing the effectiveness of the film.8 The time to get a workable solution depends on the structure of silanes

as well as pH, solvent, etc. For example, the methoxy group hydrolyzes faster than ethoxy group,

and hydrophobic silane is less reactive than hydrophilic silane. Some researchers, however, also

believe that the starting point of the operational “window” is not that critical based on the

30 evidence of the continued hydrolysis of residual ester groups in deposited films exposed to the

atmosphere. 12, 26, 29, 44

Bertelsen et al.22 studied the effects of hydrolysis and condensation on the molecular structure

of γ-glycidoxypropyltrimethoxysilane (γ-GPS) in aqueous solution. As shown in Figure 2.1.3-3,

the hydrolysis was characterized by monitoring the production of the methanol and the decrease

in concentration of SiOCH3 groups in 1% solutions of deuterium oxide (D2O) using proton

NMR. The sodium salt of 3-(trimethylsilyl) propionic acid (TSP) was used as a standard. NMR shows that the hydrolysis is rapid, and is complete in a 1% GPS solution in deuterium oxide after

34 minutes. Condensation, on the other hand, takes a relatively long time. These authors used

29Si NMR to measure oligomer growth as shown in Figure 2.1.3-4. TSP was again used as a

standard. Correlating the data with the mechanical results, they conclude that the presence of

dimers and trimers/network in the “over hydrolyzed” solution decreases the performance of

silane films. This result demonstrates the critical endpoint of the “operation window,” also

known as the stability limit of the silane solutions. Based on their studies, for γ-GPS under the

processing condition they used, the operation window is from 35 minutes to 9 hours.

Figure 2.1.3-3. Proton NMR spectra of a 1% γ-GPS Figure 2.1.3-4. Si-29 NMR spectra of a 10% γ -GPS solution in deuterium oxide showing the peaks due to solution in water at hydrolysis times of: (A) 2 hours; the methoxy protons of the silane and the methyl (B) 9 hours; (C) 20 hours. (from 22) protons of methanol after hydrolysis times of: (A) 5

min.; (B) 14 min.;(C) 34 min. (from 22)

In summary, pH, concentration and time-after-initial-mixing are all important variables that

determine a “workable” silane solution–sufficient silanol groups without dimer/trimers.

2.1.3.4 The effect of metal substrate

The nature of metal substrate affects the molecular structure of the deposited silanes. Metal oxide surfaces may differ in isoelectric point (IEP), solubility of the metal hydroxide in water and the extent of contamination. 6, 8, 16, 17, 24, 45

32 Horner and Boerio25 found that the acidity or basicity (IEP) of metal oxide substrate has a

strong effect on the orientation of amino-silane. More “upside down” orientation with the amino

group on the substrate was found on oxide surface with lower IEP, such as silicon, titanium

substrate.

Susac et al.6 used scanning electron microscopy (SEM) to study the absorption of organosilanes on different micro structural regions of aluminum alloy. Figure 2.1.3-5(a) shows that SEM micrograph measured from a 2024-T3 aluminum alloy surface coated with BTSE immediately after polishing. The adsorption of BTSE on a freshly polished aluminum alloy is non-uniform. They postulate that the second-phase particles represent different surface chemistry. The IEP and solubility, for example, are different from surface of the alloy matrix.

Both the density and acidity of the hydroxyl groups are different. As a consequence, when the alloy is exposed to BTSE solution, local variation in the silane coverage is obtained.

Figure 2.1.3-5(b) shows that the adsorption of γ-APS yields more uniform film compared to

(a) (b)

Figure 2.1.3-5: SEM micrographs measured from a 2024-T3 alloy sample which had been coated with (a) BTSE (b) γ-APS immediately after polishing.6

33 BTSE. These authors postulated that the hydrogen bonding through the amino groups promotes

uniformity. This result shows the trade-off in the application of silanes: BTSE, without the

hydrophilic organofunctionality, forms more hydrophobic film, but does not deposit uniformly

on some substrates. γ-APS, on the other hand, forms a uniform film, but the film is not

hydrophobic enough because of the presence of amine group. The way to get the optimum

condition, then, is to blend hydrophilic silanes with hydrophobic silanes.8, 10, 12

2.1.3.5 The Effect of Curing Step

The curing step, performed after the silane deposition, induces an important modification of the structure of the film: an increase of the crosslink density leading to a denser thinner film.18, 26

Abel et al.18 used XPS and TOF-SIMS to determine the thickness of various GPS coatings.

Increasing the curing temperature leads to thickness variation. The increase of curing temperature, however, has various consequences, such as degradation of film as well as elimination of water and methanol (ethanol) formed during hydrolysis.

Various authors have shown that elevating the temperature of drying increases the crosslink

density of silane films. Van Schaftinghen et al.2 tested BTSE films using infrared spectroscopic ellipsometry (IRSE) at different curing times. The results show that the intensity of the absorption band corresponding to the silanols (Si-OH) decreases while the intensity of the absorption band corresponding to the Si-O-Si bonds increases with increasing curing time. No change is observed after 40 min at 200°C.

Based on EIS, spectroscopic ellipsometry (SE) and IRSE, Franquet et al.46 evaluated the curing effect on the BTSE films. They found a decrease of thickness with curing time. The variation of thickness is attributed to the formation of denser layer due to reticulation of the BTSE.

34 Based on EIS, Franquet et al.26 find a new dielectric time constant at longer curing time in

BTSE films. Van Ooij et al.10proposed that this new feature is due to a silicate conversion layer between the silane and the metal. These authors attribute the excellent water-barrier performance of bis type silane to this layer. There might also be a higher crosslink density zone near the

substrate, as suggested by Kent et al.9

2.1.4 Structure Property Relationships in Corrosion-Inhibiting Films

The absorption and film chemistry of silanes films is quite complicated. A variety of surface

analytical techniques are used to characterize thin silane films, including FTIR and multinuclear

(29Si, 1H) NMR, ellipsometry, time of flight-secondary ion mass spectrometry (ToF SIMS), X- ray photoelectron spectroscopy (XPS), AFM, SEM/EDX, X-ray and Neutron reflectivity, etc.

2.1.4.1 Interaction of silane with substrate

The widely accepted chemical bonding theory is used to describe the “coupling effect” of silanes: Silanol group can react with the metal oxide surface to form the oxane bond Si-O-metal;

the functional group of silane molecules can then react with polymer to form better adhesion

between polymer and metal:

35 Si (OH) 2R' 3 R' R' + OH Si OH OH Si OH H H O O H H H H O O O O M M M M

Adhesive Adhesive R' R' R' R' + Adhesive Si O Si Si O Si O O O O M M M M

This model presumes a covalent bonding via condensation between hydroxyls present at the surface of the metal and silanols of the hydrolyzed silane. The organofunctional groups of silane then would react with the polymer if a polymer adhesive is applied over the silane.8 The presence of oxane bond (M-O-Si) between metal and silane has been proven for aluminum and other metals for various silanes, mainly by use of SIMS.18-20 Abel et al.20 studied the interaction between oxidized aluminum and hydrolyzed γ-GPS. By examination of fragments at the nominal mass m/z = +71, they show the presence of the AlOSi+ units at high mass resolution (m/∆m =

+ 3800 at m/z = +41(C3H5 )).

In another ToF-SIMS study performed by Ulf Bexell et al.,16, 45 the interface bonding between

BTSE and three different metal substrates (aluminum, zinc and an aluminum-zinc alloy) was analyzed. Ion etching using Ga+ was used to expose the interfacial region. As shown in Figure

2.1.4-1, ion fragments from the samples were examined where supposed metal-oxygen-silicon ion fragments should appear in the mass spectra. It was concluded that AlOSi+ ion fragments

36 exist at nominal mass m/z = 71 on the aluminum and aluminum-zinc alloy substrate and ZnOSi+ ion fragment exists at nominal mass m/z = 108 on the zinc and zinc alloy substrate.

(a) (b)

Adsorption of silanes on metals through the silanol groups is widely assumed. In some cases,

especially with aminosilanes, however, the coupling agent can adsorb on metal surface with

inverted orientation. Three kinds of orientation can occur,27 as shown in Figure 2.1.4-2.

Orientations b and c have the amino group bound to the substrate through hydrogen bonding.

This “upside-down” orientation of amino silanes has been widely studied and depends on the

solution pH and substrate acidity.

Horner et al.25used XPS to study the orientation of γ-APS on the oxide surface with a range of isoelectric points. The amine group was protonated when γ-APS is orientated upside down with the amine group interact with the metal. The change of N (1s) high-resolution spectra shows the extent of protonation of amine groups. Figure 2.1.4-3 is one example. Two components were observed in the N (1s) spectra near 399.4 eV and 401.3 eV. The component at lower binding energy was assigned to free amino groups but the component at higher binding energy was assigned to protonated amino groups. By monitoring the intensity of two components, they showed that the extent of protonation varies with acidity of the surface oxide, with the greatest value on acidic surfaces such as silicon and titanium and least on basic surface such as magnesium.

Figure 2.1.4-3: N (1s) high resolution spectra of polished 1100 aluminum coated with γ-APS from a 1% aqueous solution at pH 10.4 with take off angle 75°.25

The absorption of silane through amino groups is considered detrimental for corrosion

protection. Subramanian et al.43 compared the corrosion behavior of BTSE and γ-APS films deposited on Fe. They found that γ-APS showed no protection on Fe, while the BTSE-treated Fe exhibited no visible signs of corrosion. They explained this difference by considering the effect of the amino group in γ-APS. The absorption through amino group results in a hydrophilic interface between γ-APS film and the substrate thus promoting the ingress of corrosive chloride

ions (Cl-) into the film interface. A two-step process was proposed15, 43, 47 where the metal is first

coated by a non-functional silane (BTSE), and later by a functional silane (γ-APS). Thereby, a double layer film with strong anchoring to the substrate and a higher degree of organofunctionality in the surface layer is formed. The conclusion that the effect of the amino group in

γ-APS is the only reason for poor anticorrosion performance is not fully justified at this stage.

The hydrophobicity of the organic group is an important issue for anticorrosion, but to explain the diminished effectiveness of γ-APS compared to BTSE, the difference in oxane bond density also should be considered.

39 2.1.4.2 Interaction of water with silane

It is known that water is essential for corrosion to occur and spread on the metal surface. In

order to understand the mechanism of water interaction with silane film, direct measurement of

the amount and distribution of absorbed water in-situ is needed. Some advanced surface

analytical techniques have been used to characterize the water in the silane films.

One in-situ technique is dielectric relaxation spectroscopy, which provides information about

absorbed interfacial moisture.48 With this technique, water within the interphase region can be distinguished from water in the bulk resin, and bound water can be distinguished from mobile water. However, the detailed concentration profile of water cannot be obtained. In particular, water located on the surface of a glass fiber cannot be distinguished from water in the interphase.

Tinh Nguyen et al.49 developed a method based on Fourier transform infrared-multiple internal

reflection (FTIR-MIR) spectroscopy to determine the amount and thickness of water at an

organic/hydroxylated substrate in situ. They used a two-layer model (Figure 2.1.4-4) to qualify

the water. The first layer consists of a water layer having thickness of i and the second layer

contains the water uptake in the organic film within the penetration depth of the evanescent

wave, dp. This technique allows one to distinguish between physically absorbed water and

chemically-incorporated water. But, the problem of this technique is that, due to the fairly large

penetration depth of the evanescent wave, the depth resolution is limited. A detailed profile of

water also cannot be obtained. Rather, like dielectric relaxation spectroscopy, only the integrated

total amount of water is obtained.

40 Water sorbed in polymer

Probing depth CW Water sorbed in polymer within the of the Evanescent penetration depth wave i Water Layer

Hydroxylated substrate

Figure 2.1.4-4: The two-layered model used for quantifying water at the organic film/hydroxylated substrate interface by Tinh Nguyen et al.49

In order to get the detailed distribution of water within the silane film, length-scale resolution in the nanometer range is required. Neutron reflection (NR) can meet this requirement. NR has been used to investigate the interaction of moisture with the resin/silane/substrate systems. A detailed review of research on silane films by neutron reflectivity is giving in 2.2.4 after an introduction to the NR technique.

2.2. Neutron reflectivity

2.2.1 History of Neutron (X-ray) reflectivity

The reflection of X-rays and neutrons from surfaces has existed as an experimental technique for almost sixty years. Nevertheless, it is only during the 1990’s that these methods have become popular as probes of surface and interfaces. Apart from the scientific and technological demand for more and better surface characterization, at least two factors explain this blooming, especially in the case of neutron reflectivity. One factor is the development of neutron sources and instrumentation. The second factor is the improvement in the sample

41 preparation to make films smooth enough to utilize the high spatial resolution of the

techniques.50

The starting point for X-ray (neutron) reflectivity is Compton’s51 1922 statement that “if the

refractive index of a substance for x-rays is less than unity, it ought to be possible, according to

the laws of optics, to obtain total external reflection from a smooth surface of it.” In 1931

Kiessig52 observed reflection oscillations in the study of nickel films evaporated on glass. These so called” Kiessig fringes” allow the measurement of thin film thickness and are the basis of one of the most important applications of X-ray and neutron reflectivity. It was, however, not until

1954 that Parratt53 suggested inverting the reflectivity data to achieve a film profile.

Parratt noted in his 1954 paper that “it is at first surprising that any experimental surface appears smooth to x-rays.” But the effects of surface roughness were soon investigated, the most dramatic of them being the asymmetric surface reflection known as Yoneda wings.54 These

Yoneda wings were subsequently interpreted as diffuse scattering. The theoretical basis for the analysis of diffuse scattering was established in particular through the pioneering work by Croce et al.55

The neutron was discovered in 1932 by Chadwich. The possibility of using the scattering of

neutrons as a probe of materials, however, developed only after the availability of copious quantities of slow neutrons from reactors in 1945. The reflection principles and the interference effect exhibited by X-rays and neutrons are exactly the same. Most theoretical results developed for the interpretation of the scattering are equally applicable to both types of radiation. In recent years, neutron reflectometry has been extensively used for solving soft matter problems like polymer mixing50, 56-59 or the structure of liquids at the surface.60 The advantage of neutrons for

42 polymer studies is their small absorption compared to x-rays and the large contrast between H

and D, which allows the selective labeling by deuterium.

2.2.2 Theory of Neutron (X-ray) reflectivity

2.2.2.1 Refractive Index

The propagation of radiation is generally presented according to an optical formalism in which

the properties of a medium are described by a refractive index. The variation in refractive index

provides contrast between the species of interest and the surrounding medium. Knowledge of

refractive index is sufficient to predict what will happen at an interface, in effect to establish the

Snell-Descaters’s laws and to calculate the Fresnel coefficient for reflection and transmission.

For both X-rays and neutrons, the refractive index n of a material is in general slightly less than 1

and is given to a good approximation by61

n = 1− δ + βi (2.2.2-1)

λ2 ρ Where δ is the real part of the refractive index,δ = , with ρ being the scattering length 2π density (SLD) of the material. β is the imaginary component, which accounts for absorption.

λρ abs β = where ρabs is the absorption cross-section density. 4π

In the case of X-rays,

2 δ x = λ ρ el r0 / 2π (2.2.2-2)

β x = µλ / 4π (2.2.2-3)

43 where λ is the X-ray wavelength, ρel the electron density (proportional to atomic number), r0 the

classical electron radius and µ the linear absorption coefficient. For most materials δ is of the

order of 10-6 for both X–rays and neutrons. With CuKα X-rays β for most organic materials is

about 10-2 to 10-3 times δ and therefore can usually be ignored without introducing significant errors.

In case of neutrons, the absorption cross-section density is sufficiently small that β can be neglected in most cases, so βN = 0. The parameters ρ and δ follow from the properties of the

material.

ρ m × N A ρ = ∑bα × n = ∑bα × (2.2.2-4) α α ∑ M α α

2 ∑bα ρ m N A λ α δ N = (2.2.2-5) 2π ∑ M α α

The scattering length density, ρ, is determined by the number density, n, of the molecular species

and the sum of the scattering lengths of all the atoms in that molecule∑bα . The number density α n is obtained from the mass density, ρm, divided by the sum of atomic masses, M, times

Avogadro’s number, NA.

With neutrons, the scattering length, b, does not vary in a systematic manner with atomic number as in the case of X-rays. The neutron scattering lengths have to be obtained experimentally. One of the largest differences in scattering length is that between the proton and

the deuteron. The scattering length for hydrogen and deuterium are -0.374×10-12 cm and

44 0.667×10-12 cm. This large difference in the scattering length provides a unique means of labeling polymer molecules with minimal perturbation to the thermodynamics. For X-rays, the contrast, or difference in the refractive index, is provided by the presence of higher atomic number elements where the number of electrons per unit volume can be large.62

2.2.2.2 Snell’s law and Fresnel’s law

When radiation is incident on an interface between two materials, part of the energy is reflected at the interface and the rest is transmitted. Irrespective of whether the radiation involved is a beam of light, x-rays, or neutrons, the geometry and the relative intensities of the reflected and refracted rays can be described by the principles of optics.61

Z Reflected beam Incident beam

Medium 0 θ0 θ

θ1

Medium 1 Refracted beam

Figure 2.2.2-1: Geometry of scattering (and reflection) from surface.

Figure 2.2.2-1 depicts the geometry of scattering at a surface, where the incident, reflected and refracted rays are represented by wave vectors k0, k and k1. Most studies of X-ray and neutron reflectivity are concerned with the measurement of specular reflectivity, where θ is equal to θ0.

The scattering vector q (= k - k0) for specular reflection is normal to the surface, and its x and y

45 components are both equal to zero. In vacuum, the component of the scattering vector normal to the surface (z direction) is given by

4π q = sinθ (2.2.2-6) λ

The reflectivity, R, is the ratio of the reflected beam energy to the incident beam energy. R is measured as a function of the magnitude of q while its direction is kept normal to the surface.

The change in q can be accomplished either by changing the angles θ0 and θ of incidence and reflection at the same time or by changing the wavelength.

When the surface is not perfectly flat or when material near the surface contains some inhomogeneities in scattering length density in the direction parallel the surface, scattering in non-specular direction may become pronounced. This scattering is referred to as off-specular or diffuse scattering. The scattering vector q in such a diffuse scattering measurement contains a finite qx or qy components. The result is then analyzed to derive information about scattering length density inhomogeneities in the x or y direction. Diffuse scattering has not been developed as much as specular reflectivity due to its greater complexity and the delayed development of the accompanying theory. However, considerable progress has been made in this area in the last few years and a variety of experiments have been performed recently. Russell57 and Foster63 review this topic. Due to the uniform nature of silane films, in our study, we concentrated on the specular reflectivity.

In the specular reflection the angle of reflection is related to the angle of incidence by the law of reflection:

46 cosθ0 = cosθ (2.2.2-7)

Snell’s law gives the relationship between angle of incidence and angle of refraction:

n0 cosθ 0 = n1 cosθ1 (2.2.2-8)

In most cases, we consider the first medium is a vacuum (or air), therefore n0 is equal to 1. For

X-rays and neutrons, the refractive index of most materials is less than one, implying that there is a critical angle, θc, below which there is total external reflection of the incident beam. The critical angle is thus given by

cosθ c = n1 (2.2.2-9)

and at small θ by

θc = 2δ = λ ρ /π (2.2.2-10)

-6 For both CuKα x-rays and neutrons, δ is ~10 , so θc is on the order of milliradians.

In a medium of refractive index n, the wavelength is equal to λ/n and the magnitude of the

incident wave vector k is equal to 2πn/λ, where λ is the wavelength in a vacuum. Based on the

Snell’s law:

2 2 1 2 k z1 = (k z0 − k zc ) (2.2.2-11)

where kz0 and kz1 are the z component of the incident beam in medium 0 and refracted beam in medium 1. kzc is the value of kz0 when θ0 is equal to the critical angle θc.

1/ 2 k zc = (2π / λ)sinθc = (2π / λ)sin(λ ρ /π ) = 2(πρ) (2.2.2-12)

47 so in a medium i with scattering length density ρi,:

2 2 k zi = (k z0 − 4πρi ) (2.2.2-13)

During the analysis of reflectivity, which will be discussed below, this equation was applied in

the calculation formalism to connect reflectivity with the scattering length density of each layer.

2.2.2.3 Reflectivity from a system with one interface

The reflectivity R is defined as the fraction of the incident energy that is reflected. Consider a

general case with an infinitely sharp interface between medium 0 and medium 1 as shown in

Figure 2.2.2-1. The fraction of the amplitude that is reflected back to medium 0, r0,1 is called reflection coefficient (or the Fresnel coefficient for reflection). The fraction that is transmitted to the medium 1, t0,1 is called transmission coefficient (or the Fresnel coefficient for transmission).

k z,0 − k z,1 r0,1 = (2.2.2-14) k z,0 + k z,1

t0,1 = 1− r0,1 (2.2.2-15)

The reflectivity R is then given by:

2 k − k R = rr * = z,0 z,1 (2.2.2-16) k z,0 + k z,1

R is also known as the Fresnel Reflectivity and is often denotes as RF. After substitution of

(2.2.2-11) for kz,1 in (2.2.2-16), the Fresnel reflectivity is then given as

48 2 4 k − k 2 − k 2 ⎛ k ⎞ z,0 z0 zc 1 ⎜ z,c ⎟ −4 RF = ≅ ∝ q (2.2.2-17) 2 2 16 ⎜ k ⎟ k z,0 + k z0 − k zc ⎝ z,0 ⎠

This equation shows that for the tail of the reflectivity curves decays as q-4 for large q, just as

4 does the small angle scattering intensity in the Porod region. Conversely, RF×q will be a

constant at high q. For θ < θc, RF is nearly unity, falling short by a very small amount attributable to absorption. A calculated Fresnel reflectivity is shown in Figure 2.2.2-2. The region of total

reflection below qc is clearly visible. The absorption effect is important only for the incident angles in the vicinity of the critical angle qc. In the reflectivity curve, the absorption effect is only

53 a slight rounding off of the curve around qc.

1 10

0 qc 10

-1 10

-2 10 ctivity -3

fle 10

Re -4 10

-5 10

-6 10 7 8 9 2 3 4 5 6 7 8 9 2 0.01 0.1 -1 q (Å )

(a)

49 -7 10

4 -8 q 10 c

-9 10 Reflectivity*q

-10 10

0.00 0.05 0.10 0.15 0.20 -1 q (Å )

(b)

Figure 2.2.2-2: (a) Calculated Frenel neutron reflectivity profile from an infinitely sharp interface between two media obtained from equation 2.2.2-16. Total reflection (R = 1) occurs for q < qc. (b) 4 Reflectivity multiplied by q , showing the asymptotic limit reached a large value of qz.

2.2.2.4 Reflectivity from a system with two parallel interfaces

When there is a film with thickness d between substrate and air, R is similar to the one calculated for the substrate only, but interference of the beams reflected at the two interfaces results in a pronounced oscillation superimposed on the Fresnel curve. These oscillations are called “Kiessig fringes,” named after H. Kiessig, who first observed such fringes with metal films. The thickness of the film may be readily estimated from the spacing in q (∆q) between successive maxima or minima, as

d = 2π / ∆q (2.2.2-18)

50 The spacing of the fringes, ∆q, is independent of the refractive index of the film, but the amplitude increases with the magnitude of scattering length density difference at the interface.

Diffuse interfaces will generally reduce the oscillation amplitude particularly at large q.

Overall film thickness and characteristic spacing of periodic structures may all be derived from reflectivity data without detailed analysis. Further information, particularly on interfacial profiles in the sample, can be gained only from a more involved analysis using the techniques outlined briefly below. Due to the loss of phase information, it is not possible to directly invert a reflectivity curve to determine the SLD profile. Generally one assumes a model SLD profile to calculate the reflectivity profile.

2.2.2.5 Analysis of Reflectivity

Two primary classes of techniques are available for the analysis of specular reflectivity data:

“exact” and “approximate” formalisms.64 The exact method involves simulating the reflectivity from a candidate model structure and varying the model parameters by means of nonlinear

regression to obtain the best possible agreement between the simulated and measured

reflectivity. These methods are exact in that the reflectivity is exactly calculated based on the

real-space model. More than one real-space model, however, may fit the data. The approximate

methods use the weak scattering approximation (WSA) or born approximation so the reflectivity

is calculated as a modulus squared of Fourier transform of the scattering length density profile.

65, 66

2 2 2 16π 2 16π dρ(z) R = F{ρ(z)} = F{ } (2.2.2-19) q 2 q 4 dz

51 The advantage of this approximation is that the contributions to the reflectivity by various features in the structure can more easily be understood and identified individually. However, it is valid only for weak interactions, that is, when the incident angle, θ, is much larger than the critical angle, θc.

The “exact” method is commonly used in computer modeling program to get the scattering length density profiles. In the process of simulating the reflectivity model, several calculational schemes are used. The calculational schemes are described below in detail first using the simpler case of ideal interfaces. The consideration of microscopic roughness and instrumental resolution, which is always necessary, is discussed after the basic methods are introduced.

In the first scheme the reflectance and transmittance at each interface in the discrete structure are calculated recursively starting with the layer/substrate interface. The Parratt formalism, which is used for data analysis of the work presented in this dissertation, is based on this scheme.

A sample with arbitrary number of layers i is considered with the (i+1) th layer being the substrate layer and 0 designating the medium surrounding the sample. First, the reflectance between the substrate (i+1) and layer i, ri, i+1, is given by equation 2.2.2-14. The reflectance between the (i-1)th layer and the ith layer is then given by

' ' ri−1,i + ri,i+1 exp(2idi k z,i ) ri−1,i = ' ' (2.2.2-20) 1+ ri−1,i ri,i+1 exp(2idi k z,i )

Where the prime denotes the reflectance at the interfaces given by eq. (2.2.2-14) while ri-1, i includes the internal reflections at the i-1, i and i, i+1 interfaces of the single layer film. The reflectance, ri-1, i , is then used to calculate the reflectance for the next layer by

52 ' ' ri−2,i−1 + ri−1,i exp(2idi−1k z,i−1 ) ri−2,i−1 = ' ' (2.2.2- 21) 1+ ri−2,i−1ri−1,i exp(2idi−1k z,i−1 )

This recursion is then continued until the reflection coefficient at the 0, 1 or air/sample interface is obtained. Since the reflection coefficient of each layer is dependent upon the reflection coefficient of the underlying layer, the reflection coefficient at the 0, 1 interface is then that of the entire sample. Deviation between the calculated and measured reflectivity is minimized by varying the scattering length density and widths of the layers in the histogram.

Optimized parameters are obtained by least-squares regression. The best-fit parameters were

calc meas 2 m ⎛ Rq − Rq ⎞ determined by the minimization of χ 2 = ⎜ z ,i z ,i ⎟ , where m is the number of data ∑⎜ ⎟ i=1 ⎝ weighting ⎠

points, Rcalc is the calculated reflectivity at data point i, R meas is the measured reflectivity at data qz,i qz,i

point i, and weighting is either 1( no weighting) or R meas (statistical weighting) or δ R meas ( error qz,i qz ,i weighting).

A computationally more efficient method is offered by the use of an optical transfer matrix

formalism, as described by Lekner.67 The optical characteristics of any single layer i are

summarized concisely in a matrix Mi:

⎡ cos(k z,i d i ) sin(k z,i di ) / k z,i ⎤ Mi = ⎢ ⎥ (2.2.2-22) ⎣− k z,i sin(k z,i di ) cos(k z,i di ) ⎦ which contains the coefficients of two simultaneous linear differential equations linking the amplitude of the electromagnetic field in the layer and its derivative. The reflectance of a stack

53 of uniform layers may be found simply by first multiplying the transfer matrices of all N layers to form the transfer matrix of the total film, M

⎡m11 m12 ⎤ M = ⎢ ⎥ = M N −1M N −2 ...M 2 M 1 (2.2.2-23) ⎣m21 m22 ⎦

The reflection coefficient r from the surface of the 0-1 interface is then obtained, in terms of the matrix element of M, by

(k z0 k zN m12 + m21 ) − i(k zN m11 − k z0 m22 ) r = (2.2.2-24) (k z0 k zN m12 − m21 ) + i(k zN m11 + k z0 m22 )

2.2.2.6 Non perfect layers and Practical problems:

The measured reflectivity is a result of the variation in the scattering length density ρ(z) normal to the surface. If there is a moderate variation in the scattering length density in the x and y directions, ρ(z) is to be understood to be the average over an x-y area at a given z. However, the question arises over what distance the scattering length is averaged. The answer is given in reference to the concept of the coherence area and coherence length. The coherence length is the distance between two points on the sample from which scattered rays will interfere coherently at the detector. Consider two parallel rays separated by ∆x, impinging on a surface at an angle θ as shown in Figure 2.2.2-3. The two rays at the sample surface are separated by a distance of

∆x/sinθ. The rays are then specularly reflected onto a detector. Under the condition of specular reflectivity there is no path length difference of either ray and the coherence is determined by the effective width of the wave front. For the beam with angular divergence of δα and wavelength of

λ, it is known that the effective width (∆x) on the wave front of two rays that can interfere with

54 each other is of the order of λ / 2δα. With λ = 1 Å and δα = 0.005°, λ/2δα is of the order of 0.5

µm.61 At θ = 1°, then the coherence length (λ / 2δα) / sin θ is 25 µm, which is a macroscopic

distance. The distance over which the densities are being averaged is large.57 If the scattering length density over the coherence area still varies somewhat as the area is moved parallel to the interface, then an additional nonspecular diffuse scattering will become superimposed on the specular reflectivity.50

∆x

θ θ ∆x sinθ

Figure 2.2.2-3: Plane waves incident on a surface spatially separated by a distance ∆x impinge upon the surface of a specimen at a distance ∆x / sin θ apart.

The different expressions given above are valid for a perfect incident beam. For the fit of experimental data, it is important to have a good knowledge of the beam divergence and homogeneity. The beam angular divergence and wavelength dispersion must be taken into account in the simulations.68 The divergence of the beam, δα, is usually determined by two slits if the beam is smaller than the effective width of the sample seen by the neutron beam, or by the first slit and the sample itself if the sample is small enough to be totally illuminated by the neutron beam. Usually, δα is fixed during the experiment. δα has two effects: a decrease of the amplitude of the oscillations and a rounding of the discontinuity at the critical angle. Figure

2.2.2-4 gives an example of this effect. Wavelength dispersion δλ, is strongly dependent on the

55 monochromator or on the time resolution in the case of time of flight spectrometers. The effect of δλ is that the oscillations disappear at high angles (see Figure 2.2.2-4).

Figure 2.2.2-4: Effect of δλ and δα Comparison between a perfect instrument, and instrumental δλ, and δα for a measurement on a single 30 nm thick layer on a substrate. (From 68, p186)

Up to this point the analysis of the reflectivity has dealt specifically with the reflectivity from specimens with infinitely sharp interfaces. However, no interface is infinitely sharp, but exhibits a gradient in density not only between the specimen and the surrounding media, i.e. the substrate or air, but also between consecutive layers. This gradient may represent a diffuse interface between the layers or interfacial roughness or waviness. Extensive treatment of interfacial roughness and waviness can be found in the literature. 57, 63, 69-71

There are two size scales of roughness. Considering the two different cases shown in Figure

2.2.2-5 where there is radiation incident on a surface with curvature, the only difference between

56 these two is that in case b the curvature of the interface is so small in comparison to the coherence length lc whereas in case a the curvature is much greater. It is clear that the density profile is quite different in these two cases.

Figure 2.2.2-5: Rough interface in which the radii of curvature are (a) larger than the coherence length lc 62 and (b) smaller than lc. (From , p. 256)

In Figure 2.2.2-5a, the interface can be regarded as planar and smooth. The scattering length density variation at the surface is sharp despite the curvature. However, the angle of incidence,

θ1, θ2, etc., varies from place to place. Thus, the total effect of long-range roughness or waviness will be quite similar to the one produced when a beam containing a degree of divergence is incident on a planar, smooth interface.

In Figure 2.2.2-5b, on the other hand, the waviness on the surface occurs over distance scale much smaller than lc. In this case, the scattering length density averages over a coherence area no longer change abruptly as the interfacial boundary is crossed. It will undergo a gradual transition, as a function of z, from one medium to the next. The effect on the reflectivity would then be the same as that produced by a diffuse interface. In general, this leads to a reduction in the reflectivity arising from that interface and a blurring of the coherent oscillations for large q.

57 Névot and Croce69 showed that the nature of the diffuseness of the interface can be characterized using a smearing function g(z). Usually g(z) can be approximately by a Gaussian function:

1 z 2 g(z) = exp(− 2 ) (2.2.2-25) 2πσ 2 2σ

with σ characterizing the “width” of the diffuse interface. For Gaussian roughness, then, the

Fresnel reflectivity is modified by an exponential factor, the so called Névot-Crocé factor.

2 2 R = R f exp(−q σ ) (2.2.2-26)

Equation (2.2.2-26) shows that with a diffuse interface the reflectivity falls off more rapidly

than it would be with a sharp interface.

In case of a stack of multilayer each having a specific roughness, the Névot-Crocé factor can

be incorporated into the reflectivity coefficient of each interface.

k z,i − k z,i+1 2 ri,i+1 = ( )exp(−2k z,i k z,i+1σ i+1 ) (2.2.2-27) k z,i + k z,i+1

Figure 2.2.2-6: Reflectivity calculated for a deuterated polystyrene film of thickness 500 Å deposited on a Si substrate, to illustrate the effect of diffuse interface and small-scale roughness.

Thin solid curve, σ1 = σ2 = 0; thin broken curve, σ1 = 20 Å, σ2 = 0; thick broken curve, σ1 = 0, σ2 = 20 Å. (From 62, p. 255)

Figure 2.2.2-6 gives numerical results that illustrate the effect of diffuse interface, as predicted by equation 2.2.2-26. Here, neutron reflectivity curves have been calculated for a system consisting of a 500-Å deuterated polystyrene film laid over a Si substrate. The solid curve shows the result obtained when the interfaces between air and the polymer and between the polymer and substrate are both sharp (σ1 = σ2 = 0), the thin broken curve is obtained when the air-polymer

interface is diffuse (σ1 = 20 Å, σ2 = 0), and the thick broken curve is obtained when the polymer-

Si interface is diffuse (σ1 = 0, σ2 = 20 Å).

2.2.3 Instrumentation for reflectivity

The reflectometers can be divided in two different groups: time of flight reflectometers like

POSYII at Argonne National Laboratory, SPEAR at Los Alamos National Laboratory, and

59 monochromatic reflectometers like NG7 at National Institute Standard and Technology (NIST).

Corresponding to the two designs of reflectometers, there are two types of neutron sources:

fission (reactor) and spallation.

Time of flight reflectometers:

The time of flight technique consists of sending a pulsed white beam onto the sample. Since

the speed of the neutron varies as the inverse of the wavelength, the latter is directly related to

the time taken by the neutron to travel from the pulsed source to the detector.

For a reflectivity measurement, the angle is fixed and the reflectivity curve is obtained by

measuring the reflectivity signal for each wavelength of the available spectrum, each wavelength

corresponding to a different scattering wave-vector magnitude q. Sometimes, it is necessary to

use several angles because the q range is not large enough.

Monochromatic reflectometers:

Monochromatic reflectometers are basically two axes spectrometers. The wavelength is fixed

(0.47 nm for NIST) and the reflectivity curve is obtained by changing the incident angle θ.

Typically, neutron reflectivity measurements are made at grazing angles of incidence in the range of 0.25 to 1.8°, using neutrons in the wavelength range 1 to 7 Å to cover a q range 0.008 to

0.5 Å-1.72 The incident beam is highly collimated in the z direction to give values of ∆q/q

typically from 2 to 5%. Vertical and horizontal slits, in the ranges 0.2 to 3.0 mm and 10.0 to 40

mm, provide sample illumination area in the range of 1.0 to 100 cm2. Measurements can be made

from total reflection at low q to reflectivity ≤ 10-7 at high q.

60 2.2.4 Application of neutron reflectivity in the study of silane films

Wu et al. 73 first used neutron reflection to examine water absorption at the interface between a polyamide and the native surface of a silicon wafer. They reported that a D2O level of ~17% at

the interface and ~3% in the bulk of the polyamide film after exposure to a saturated D2O

atmosphere at room temperature. When the silane coupling agent, aminopropylethoxysilane was

first coated onto the surface of the wafer, the D2O content at the interface reduced to 12%.

Kent et al. has done a lot of work using neutron reflection (NR) to probe the structure of

organosilanes. 9, 44, 74-77 In their early work, they used the NR to study the profile of absorbed water at a molybdenum/polyurethane interface.77 They reported that the amount of interfacial water was greatly reduced when a silane coupling agent was mixed into the resin. They concluded that silane only promotes adhesion of the adiprene to the molybdenum oxide layer, but does not provide a barrier to moisture penetration. Their films, however, were very thin.

In one of Kent’s studies, the structure of GPS and BTSE films, especially the crosslink density distribution was studied using solvent swelling.9 Deuterated nitrobenzene (d-NB) was used to swell both films. The as-prepared measurement was carried out with desiccant placed in the chamber, while, the swollen state, was studied with the reservoir filled with d-NB. Figure 2.2.4-1

(a) shows the result of NR from 0.3% GPS samples as prepared and after swelling to equilibrium. Figure 2.2.4-1(b) is the scattering length density distribution calculated from the reflectivity data in (a), which shows the SLD profile normal to the surface. A large increase in neutron reflectivity was observed upon swelling with d-NB. Also the d-NB concentration profile has a maximum in the central region of the film. Since NB is a good solvent for both GPS and

BTSE monomers, the variance in concentration of solvent is attributed to the crosslink density

61 variations. It was concluded that the GPS films have a higher crosslink density near silicon

surface than in the bulk of film. In sharp contrast to GPS film, BTSE films do not swell with d-

NB at all, indicating a highly crosslinked structure throughout the film.9

(a) (b)

In another study, Kent et al.44 concentrated on the structure of silane films after hydrothermal

conditioning. They examined the NR from BTSE films that were conditioned for 2 days in D2O or H2O saturated air at 80 °C. A large increase in reflectivity was observed upon conditioning

with D2O, but no change in reflectivity was observed upon conditioning with H2O. The D2O result can be explained by either adsorption of free D2O, or by deuterium incorporated

chemically into the film (the hydrolysis of residual ethoxysilyl groups). But, considering the H2O conditioning result, free water absorption can be ruled out, because the increase in reflectivity is

62 also expected if absorbing free water occurs. Their interpretation is summarized in the following

reactions where only the Si-O-X portions of the BTSE molecules are shown:

As-prepared film:

n(-Si-O-Si-) + m(-Si-OCH2CH3)

D2O conditioning:

n(-Si-O-Si-) + m(-Si-OCH2CH3) D2O n(-Si-O-Si-) + m(-Si-OD) + mDOCH2CH3

A large increase in reflectivity results; Because the SLD of m(-Si-OD) is much larger than that of m(-Si-OCH2CH3)

H2O conditioning:

n(-Si-O-Si-) + m(-Si-OCH2CH3) H2O n(-Si-O-Si-) + m(-Si-OH) + mHOCH2CH3

A negligible change in reflectivity results; Because the SLD of m(-Si-OCH2CH3) and m(-Si-

OH) are similar.

Neutron reflectivity is a powerful technique with an excellent spatial resolution. Unfortunately,

no information of a chemical nature is obtained. For example, no distinction can be made

between the presence of D2O molecule and the presence of D and OD species chemically attached to the silane film. So, reflectivity methods must be used in conjunction with other techniques like X-ray photoelectron spectroscopy, or dynamic secondary ion mass spectrometry to get the detailed proper profile.

63 Chapter 3. Experimental

3.1. Sample Preparation

3.1.1 Materials

Bis[3-(triethoxysilyl)propyl]tetrasulfide (bis-sulfur) (H5C2O)3Si-(CH2)3-S4-(CH2)3-

Si(OC2H5)3 and bis-[trimethoxysilylpropyl]amine (bis-amino silane) (H3CO)3Si-(CH2)3-NH-

(CH2)3-Si(OCH3)3 were provided by OSi Specialties (Tarrytown, NY). The molecular structures are shown in Figure 3.1-1. Since the properties of bis-amino silane will be compared to BTSE, the molecular structure of BTSE is also shown in Figure 3.1-1. The silanes were used without further purification. It should be noted that bis-sulfur silane doesn't contain a "pure" tetrasulfide linkage but also contain small amounts of di- and trisulfide linkge. Bis-amino silane is also a mixture of bis-[trimethoxysilylpropyl]amine, gamma-aminopropyltrimethoxysilane, tris(trimethoxysilylpropyl)amine and methanol. The amount of bis-[trimethoxysilylpropyl]amine is more than 90% in the mixture. The silicon wafers used as substrates were polished 2-inch and

3-inch diameter single crystals (111) wafers obtained from Wafer World, Inc. (West Palm Beach,

FL, USA). D2O (99.9 atom% D) and Nitrobenzene-d5 (d-NB, 99.5 atom% D) were obtained from Aldrich and used as received.

OC2H5 OC2H5 Si C2H5O S S OC2H5 S S Si C2H5O OC2H5

(a) Bis[3-(triethoxysilyl)propyl]tetrasulfide (bis-sulfur silane)

64 H3CO OCH3 Si NH Si H3CO OCH3 H3CO OCH3

(b) Bis-[trimethoxysilylpropyl]amine (bis-amino silane)

OC H C2H5O 2 5 Si Si C2H5O OC2H5 C2H5O OC2H5

(c): Bis(triethoxysilyl)ethane (BTSE)

Figure 3.1-1: Molecular structure of (a) bis-sulfur silane; (b) bis-amino silane; (c) BTSE

3.1.2 Procedure:

3.1.2.1 Silane solution preparation

A 1-wt% bis-sulfur silane solution was prepared by adding the silane to a mixture of DI water and ethanol. The ratio of silane/water/ethanol was (1/9/90) (w/w/w). It is known that after mixing with water, the solution has to be hydrolyzed for some time to reach the “workable” condition. To find the best hydrolysis time, the first step is checking the appearance of the films spun at different hydrolysis times. If the silane is under hydrolyzed, an oily and non-uniform film forms; if the silane is over hydrolyzed, a non-uniform film forms and white spots may also appear. A uniform, transparent film can only be obtained from a solution that contains enough silanols generated from hydrolysis for the subsequent condensation. After the hydrolysis time window is found by checking the appearance of the film, films coated during this time window were examed by X-ray reflectivity to find the best hydrolysis time. Both the hydrolysis and

65 condensation rates are pH-dependent, so the optimal hydrolysis time is actually controlled by

pH. Under acidic conditions, fast hydrolysis is followed by slow condensation. An acidic

solution, therefore, is more stable making it easier to obtain uniform films. Since the natural pH

of the bis-sulfur solution is around 6.5. Acetic acid was used to lower the pH to 4 to accelerate

hydrolysis. At this condition, the best hydrolysis time was found to be 42 ± 1 hours for bis-sulfur

silane based on the procedure described above. A 1% bis-amino silane solution was prepared

similarly, but the pH was adjusted to 7.5 and the solution was hydrolyzed for 18 ± 1 hours. The

natural pH of bis-amino silane is above 10. Therefore, the solution is more stable at 7.5. The

hydrolysis of bis-amino silane is much faster than bis-sulfur silane due to the catalytic effect of

the secondary amine group. The organic solvent used for the bis-amino solution was methanol

rather than ethanol because the hydrolysis product for bis-amino silane is methanol instead of

ethanol. The mixture of bis-sulfur and bis-amino silane was made by mixing the above workable

individual silane solutions at a bis-sulfur solution/bis-amino solution weight ratio of 3/1. The

final solution contains 0.75-wt% bis-sulfur silane and 0.25-wt% bis-amino silane.

A 5-wt% silane solution was prepared in a similar way by adding the silane into a mixture of

DI water and ethanol. The ratio of silane/DI water/ethanol was (5/5/90) (w/w/w). The mixture of bis-sulfur and bis-amino silane was made by mixing the “aged” individual silane solutions at a bis-sulfur solution/bis-amino solution weight ratio of 3/1. In the final solution, there will be 3.75- wt% bis-sulfur silane and 1.25-wt% bis-amino silane.

3.1.2.2 Silane film formation

After the solution is properly aged, the film is deposited using a Larrel single-wafer spin processor. During the first stage of study, silicon wafer substrates were used. The wafers are

66 cleaned by immersion in a freshly prepared “piranha” solutions (conc.H2SO4/H2O2 30% = 7/3 v/v) at room temperature for at least 30 minutes. After immersion, the substrates are rinsed repeatedly with de-ionized (DI) water. The silane solution is pipetted onto the wafers covering the whole surface. After one minute, the wafer is accelerated to 2000 rpm and held for 1 minute to spin off the excess solution and dry the film. The samples are dried in an oven for an hour at

80 °C for the first-stage study. In the second stage, similar samples are cured at 180 °C.

During the second stage of study, NR on Al substrates instead of silicon wafers is carried out to study the effect of substrate. Because a thick silicon wafer is required to achieve the smoothness required for neutron reflectivity, an Al layer was e-beam evaporated on a silicon wafer. A cathode is heated by driving 25 amp of current through it. The resulting electrons are accelerated toward the target (Al) by a 12-kV dc power supply. The beam is bent by electromagnets to center the beam in the target. The power is increased until the metal's evaporation temperature is reached. The thickness is monitored by a crystal mass monitor. The

Al film thickness is around 20 nm and roughness is about 1 nm. Silane films are then applied to

Al surface using the same spin-coating technique as on silicon wafer substrate.

3.2 Characterization of Silane films

3.2.1 X-ray reflectivity

X-ray reflectivity (XR) measurement was performed at 1-BM beamline78 at the Advanced

Photon Source (APS), Argonne National Laboratory, Argonne, IL, USA, using standard 4 circle

Huber diffractometer. An Oxford Cyberstar scintillation detector with YAP head was mounted on the arm of the diffractometer with two sets of slits separated by an evacuated beam path. The distances from the sample to slits were 350 mm (guard slits) and 825 mm (detector slits). The

67 beam was focused both horizontally and vertically onto the plane behind the scintillation detector

using the beamline optics. The slits in front of the detector were set to about 0.8 mm in width and

0.2 mm in height. The width of the beam in the sample position was even larger (about 1.5 mm).

The wavelength was 1.024 Å. Another two sets of slits were placed at the entrance to the hutch

(about 2040 mm before the sample) and right before the sample (270 mm) and were separated by

vacuum path. The instrument was operated using SPEC (Certified Scientific Software,

www.certif.com) and data were reduced using set of custom Igor macros.

3.2.2 Neutron reflectivity

3.2.2.1 Test procedure

Neutron beam

Silane Open Film Reservoir D2O Silicon Substrate

Figure 3.2.2-1: Experimental set up for “in situ” measurement of neutron reflectivity.

The interaction of water with silane film was studied by moisture conditioning followed by

desiccating the sample. Neutron reflectivity data were obtained from the as-prepared film, the

film after exposure to saturated D2O vapor, and again after re-drying. A sealed aluminum can

68 with a reservoir to hold solvent or water is used. This set up (as shown in Figure 3.2.2-1) makes

the “in situ” measurement possible. Figure 3.2.2-2 is a picture of the Al sample holder. The

neutron beam travels through the thin Al wall and is incident on the film from the air side. There

is almost no flux loss while going through the Al wall. The reflectivity of the as-prepared and

redried film is measured with desiccant inside the Al can. For the swelling measurement the

desiccant is removed and several drops of D2O or deuterated nitrobenzene are introduced. All

NR measurements were done at room temperature. Equilibrium saturation within the films is

confirmed by measuring the reflectivity with time until no further change is observed.

Figure 3.2.2-2: Digital picture of the Al sample holder.

D2O conditioning was performed by placing the samples into sealed Teflon cans in the presence of D2O-saturated air. The can was then either maintained at room temperature or placed

69 into an oven at 80 °C. NR measurement is performed at room temperature on both 25-°C

conditioned and 80-°C conditioned samples. After finishing the NR measurement in the “wet”

condition, the samples were again dried in a desiccator for about 8 hours. The 80-°C

conditioning was performed on the samples after 25-°C conditioning. Kent and coworkers 44, 75 proposed that the chemical change in the film could be measured by comparing the “redried” state to the as-prepared film. Physically bound water would eventually be removed during drying, but water (or individual deuterium atoms) bound by chemical bonds would remain in the film.

3.2.2.2 Data acquisition and analysis

NR was performed on the Surface Profile Analysis Reflectometer (SPEAR) at Los Alamos

National Laboratory (LANL), on the NG 7 reflectometer at National Institute of Standards and

Technology (NIST) and also on the POSY II reflectometer at Argonne National Laboratory

(ANL). The theory of NR was reviewed in chapter 2.

target coupled liquid H2 moderator mercury shutter sample slits T0 chopper detector slits goniometer frame overlap chopper detector after sample slits T0 slits frame overlap mirrors beam stop

4.45 m 8.73 m 12.4 m

Figure 3.2.2-3: Schematic diagram of SPEAR.

70 Neutron reflectivity, R(q), defined as the intensity ratio between reflected and incident

neutron beam, is measured as a function of the scattering vector, q = (4π/λ) sinθ, where θ is the angle of incidence and λ is the neutron wavelength. SPEAR and POSY II are both time of flight reflectometers. In both cases, q was varied by collecting intensity for a range of wavelengths at a fixed angle of incidence. The wavelength range is 1.4 to 16 Å. To obtain a sufficient range of q, the reflectivity curves were obtained by merging data from two angles of incidence. Figure 3.2.2-

3 is the schematic diagram of the SPEAR instrument.

NG 7 is a monochromatic reflectometer. The wavelength is fixed at 0.47 nm and the reflectivity curve is obtained by changing the incident angle θ. Figure 3.2.2-4 is a picture of NG

7. Beam travels from right to left. In the picture, the Al sample holder is placed on the sample table.

Figure 3.2.2-4: NG 7 reflectometer at NIST.

The q-dependence of the reflectivity depends on the neutron scattering length density (SLD) profile normal to the substrate surface. The neutron SLD is a function of the density and atomic

ρ N composition, expressed as SLD = b × A , where ρ is the mass density and M is the molecular M

weight of a molecular unit having a scattering length b. NA is Avogadro’s number. The calculated neutron SLD values for the materials used in this study are listed in Table 3.2.2-1.

Table 3.2.2-1: Calculated neutron SLD for the materials used in this study. (http://www.ncnr.nist.gov/resources/sldcalc.html)

Material Mass density (g/cm3) 106 x SLD(Å-2)

72 bis-amino (Si2O6C12NH31) 1.04 0.299

bis-sulfur (Si2O6C18S4H31) 1.1 0.21

D2O 1.1 6.33

H2O 1.0 -0.56

ethanol (C2O1H6) 0.8 -0.35

methanol (COH4) 0.8 -0.38

SiO2 2 3.40

Si 2.3 2.05

Al 2.7 2.08

Al2O3 3.9 5.60

d-NB 1.2 5.55

We used the simplest multi-layer structure capable of fitting the data. For the as-prepared film,

the model was an infinite Si substrate, a SiO2 layer, a silane layer and an infinite air layer.

Instrumental resolution, which is a function of q, was included to fit the data. In the case of some

swollen films and the redried films after 80-°C hydrothermal aging, no reasonable fit could be

obtained using only the instrumental resolution. We could fit the data by decreasing the effective

resolution at large q below the instrumental value. Following Roe,62 we took the apparent loss of

resolution as evidence for roughness. Since the reduced resolution was required only for q > 0.05

Å-1, the correlation length parallel to the surface was roughly π/0.05 ≅ 60 Å. This length scale is

similar to the correlation length range observed for bulk bis–silane xerogels and is likely caused by topological fluctuations (fluctuations in cross-link density) in the films.79

73 3.2.3 Variable angle spectroscopic ellipsometry

Ellipsometry measurements were done using a J. A. Woollam Co. VASE (variable-angle

spectroscopic ellipsometer) working in the 300-1000 nm range with angles of incidence at 60˚,

65˚, 70˚, 75˚. For the analysis of the spectra, the Wollam-supplied WVASE32 software was

used. The thickness and optical constants were obtained after fitting the data using the Cauchy

model. Ellipsometry measurements were performed before NR to determine the thickness and

optical constants of the silane films.

Variable angle spectroscopic ellipsometry is an optical technique used to characterize film

thickness, optical constants and material properties.80-82 Compared to other surface analysis technique (AES, XPS, SIMS), spectroscopic ellipsometry requires no vacuum and is non- destructive. Visual spectroscopic ellipsometry uses visible light (0.3-0.8 µm).

In principle spectroscopic ellipsometry is based on the optics of layered structures. The properties under investigation are not measured themselves, but are determined by means of a simulation and regression procedure that is based on layered optical model. The perfectly flat and parallel layers of the optical model are described by their thickness (d) and their complex refractive index (N).

The complex index of refraction for each medium can be expressed as

N = n + jk 3.2.3.1

Where n is usually called the index of refraction and k is called the extinction coefficient.

These two quantities are used to describe how light interacts with the material. These quantities

are referred to as optical constants, although in reality, they do vary with wavelength,

74 temperature, and so on. When the medium surrounding the sample under study is air or vacuum,

N is equal to 1.0. 83, 84

Figure 3.2.3-1 illustrates the basic principle behind ellipsometry. First, the polarization state of incoming light is known. This incident light interacts with the sample and reflects from it. The interaction of the light with the sample causes a polarization change in the light, from linear to elliptical polarization. The polarization change is then measured by analyzing the light reflected from the sample.

Figure 3.2.3-1: Geometry of an Ellipsometric Measurement.85

Ellipsometry measures two values, Psi (ψ) and Delta (∆) that describe this polarization

change. These values are related to the ratio of Fresnel reflection coefficients, Rp and Rs for p- and s- polarized light, respectively. As shown in equation 3.2.3.2 below:

i∆ R p ρ = tan(Ψ)e = 3.2.3.2 Rs

75 Because ellipsometry measures two values, it can be highly accurate and very reproducible.

The ratio, ρ, is a complex number; thus it contains “phase” information, delta, which makes the

measurement very sensitive. These two values by themselves aren’t very useful in characterizing

a sample. What we really want to know are properties like film thickness, optical constants,

refractive index and other useful data. Because of the “phase” problem, the refractive index and

thickness can not be directly obtained. They are found by using the measured values, Psi (ψ) and

Delta (∆), in various equations and algorithms to produce a model that describes the interaction

of light with the sample. The detailed process of data analysis can be found in the literature. 80-82

The optical properties are often dependent on other microstructural properties of the materials—degree of crystallinity, alloy concentration, surface roughness, interfacial roughness, and any other property that affects the optical constant. Hence, an understanding of the relationship between other material properties and optical properties can facilitate the measurement of those other properties. 83, 84

3.2.4 SEM

An SEM image was taken with a FEI XL-30 ESEM FEG. A Shottky hot-field emission tip is

employed as the electron source. The instrument has an ultimate resolution of 12-15 Å. The

working distance is 8–10 mm. The chamber pressures: 0.9-1.3 Torr.

3.2.5 Contact angle

Contact angle measurements were done using a VCA 2002 contact angle analyzer from

Advanced Surface Technology, Inc. Billerica, Massachusetts. DI water was used as the probing

liquid. The contact angles were averaged over 10 drops.

76 3.2.6 Atomic force Microscopy

Surface roughness of the silane films was determined using a Dimension 3100 scanning probe microscope (Digital Instruments, Santa Barbara, CA). Etched silicon tips (125 µm in length, resonant frequency in the 200-400 kHz range, single beam cantilever configuration) were applied in tapping-mode experiments.

77 Chapter 4. Neutron Reflectivity Investigation of Bis-amino

Silane Films1

Summary: Bis-silyl functional silanes studied here have six hydrolyzable groups and are

believed to be more crosslinked than tri- and tetra-functional analogs. The enhanced crosslinking

leads to better barrier properties in anti-corrosion applications. In this study, solvent swelling is

used to assess the degree of condensation and the crosslink density in bis-[trimethoxysilylpropyl]

amine (bis-amino) films. Nitrobenzene swells the films but does not react chemically with the

films. The results show that bis-amino silane films are highly condensed, with a nitrobenzene-

depleted layer near the silicon substrate. D2O both swells (at 25 °C and 80 °C) and chemically

alters the films (at 80 °C). The reflectivity data upon exposure to D2O vapor at 80 °C are

consistent with exchange of the amino proton (1H) with deuterium (D). A hydrophilic layer is postulated at both the air and substrate interfaces.

4.1 Introduction

In this chapter, we study bis-[trimethoxysilylpropyl]amine (bis-amino) and make comparison

to bis -triethoxysilylethane (BTSE), which was investigated previously by Kent and coworkers.9,

44 The chemical structures are shown in Figure 3.1-1.

There is some evidence for gradients in crosslink density within silane films.9, 86 A more highly crosslinked silane near the substrate is more resistant to water. Finally, to explain the

1 This chapter has been published on J. Adhesion Sci. Technol., 2003, 17, 2175

78 excellent water barrier ability of some bis-type silane films, a dense layer between the silane and

the metal substrate has been postulated.11, 26, 87, 88

Bis(triethoxysilyl)ethane (BTSE) is known to impart excellent corrosion resistance to many

metal surfaces under appropriate conditions. BTSE’s response to solvent swelling and moisture

conditioning has been investigated by Kent and coworkers.9, 44 BTSE is highly crosslinked and

does not swell when exposed to deuterated-nitrobenzene (d-NB), even though NB is a good

9 solvent for BTSE monomer. When exposed to D2O at room temperature, little absorption occurs

for BTSE film.44 Bis-amino silanes show contrasting behavior traceable to the bridging groups.

That is, barrier properties depend strongly on the type of bridging group, R'. The corrosion performance of bis-amino silane, for example, is not as good as that of the BTSE.11

We have used the solvent-swelling method developed by Kent and coworkers 9, 74 to probe the

crosslink density profile within the film. This method is based on the fact that the equilibrium

volume fraction of a swelling solvent depends on the local cross-link density, as well as the

polymer-solvent interaction parameter. The solvent profile in the direction normal to the film

surface was measured by neutron reflectivity (NR). NR offers nanometer resolution and it can

easily examine interfaces that are buried well within a sample. Strong neutron reflectivity

contrast between the silane film and the swelling solvent was obtained using deuterium-rich

50 solvents such as deuterated nitrobenzene (d-NB) and heavy water (D2O). Nitrobenzene is a

high-surface-tension liquid (40 mJ/m2) and is a good solvent for most polymers. The high-

surface-tension precludes the formation of a wetting film and the favorable thermodynamics

assures swelling of silanes. Water is also a high-surface-tension liquid (73 mJ/m2), a good solvent for hydrophilic polymers and a potential participant in film chemistry as well.

79 NR data shown in this chapter were performed on the Surface Profile Analysis

Reflectometer (SPEAR) at Los Alamos National Laboratory.

4.2: Results and Discussion

4.2.1 Bis-amino silane under d-NB swelling

Reflectivity data for films of bis-amino silane on silicon oxide as-prepared, swollen to

equilibrium in d-NB vapors, and then redried are shown in Figure 4-1a. The SLD profiles

corresponding to the fits in Figure 4-1a are shown in Figure 4-1b. Upon swelling to equilibrium

with d-NB, the reflectivity increased substantially compared to the as-prepared state and ∆q

between the fringes decreased. The increase in reflectivity is due to physical absorption of d-NB

into the film. The decrease in the fringe spacing indicates an increase in the film thickness.

After allowing the solvent to evaporate in dry air, the reflectivity decreased and the fringes

returned to their original spacing. But there was a noticeable difference between the curves for

the as-prepared state and the redried state. In particular, the first minimum at q ≅ 0.01 Å-1 was more distinct than that of the original state. This change indicates that the SLD of the redried film remained higher than that of the as-prepared film.

80 0 10 d-NB 25 °C -1 10 Redried -2 10 As-prepared

-3 10

-4 10

-5

Reflectivity 10

-6 10

-7 10 0.00 0.05 0.10 0.15 0.20 -1 q (Å )

(a)

4 d-NB 25 °C (rough) Redried As-prepared ) 3 -2

ϕΝΒ = 17% 2

ϕΝΒ = 5% x SLD (Å 6 1 10

0 0 200 400 600 Distance from Si (Å)

(b)

Figure 4-1: (a) Neutron reflectivity data from bis-amino silane film as-prepared, after swelling to equilibrium with d-NB vapor and after re-drying. The curves through the data points correspond to the best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data

81 in (a). “Rough” in the legend means roughness was included to fit the data. The film was found to be

rough in the swollen states. The calculated volume fraction of d-NB in the swollen film is φNB = 17%. After redrying, about 5% d-NB remained in the film. (The data were generated on SPEAR at LANL).

In solution, silanes undergo hydrolysis followed by condensation of the silanol groups to form

siloxane bonds. Condensation can occur in solution as well as during film formation.

Information about the extent of hydrolysis and condensation as well as the volume fraction of d-

NB within the films at equilibrium can be obtained from the SLD values.

In order to interpret the measured SLD in terms of the extent of hydrolysis and condensation

and the volume fraction of d-NB at equilibrium, some possible chemical structures of bis-amino

silanes are listed in Table 4-1. The structures correspond to various levels of hydrolysis and

condensation. The table shows that the hydrolysis of a single methoxy group with H2O results in

a very small increase in SLD (#1 to #2), assuming no change in density. Hydrolysis has a small

effect because the SLD value of the CH2 moiety is almost zero due to the negative scattering length of atom H.62 Condensation to form –Si-O-Si– also results in a small increase in SLD (#2 to #3) again because the H2O molecule that is eliminated has a small SLD. For a fully

hydrolyzed bis-amino silane film (all six methoxy groups form –Si-OH bonds), the SLD

calculated (#4) is 0.522 x 10-6 Å-2. For a fully condensed bis-amino silane film (all six methoxy

groups form –Si-O-Si– bonds), assuming the mass density is the same as that of the monomer,

the SLD calculated (#5) is 0.816 x 10-6 Å-2. Contrary to the above differences in SLD, the

presence of D in the silane leads to a substantial increase in SLD, either in the form of Si-OD

(#6) or –ND (#8).

Table 4-1: Some possible chemical structures of hydrolyzed bis-amino silane and the calculated SLD.

82 Mass 106 x SLD density (Å-2) # Structure Formula (g/cm3)

OCH OCH3 3 Si NH Si OCH3 1 H3CO OCH Si2N1C12O6H31 1.04 0.299 OCH3 3 OCH OCH3 3 NH Si Si OH 2 H3CO OCH Si2N1C11O6H29 1.04 0.328 OCH3 3 OCH OCH3 3 Si NH Si 3 O Si N C O H 1.04 0.354 H3CO OCH 2 1 11 5.5 28 OCH3 3 OH OH Si NH Si 4 HO OH Si2N1C6O6H19 1.04 0.522 OH OH O O NH Si Si O 5 O Si2N1C6O3H13 1.04 0.816 O O O O Si NH Si 6 DO OD Si2N1C6O4H13D2 1.18 1.45 O O 7 O O Si NH Si DO OD Si2N1C6O4.5H13D3 1.18 1.68 O OD O O ND Si Si O 8 O Si2N1C6O3H12D1 1.18 1.28 O O

9 O O Si ND Si DO OD Si2N1C6O4H12D3 1.18 1.77 O O

The best-fit SLD profiles corresponding to the curves through the data points in Figure 4-1a are shown in Figure 4-1b. For the as-prepared state, a single-layer silane film SLD profile model with instrumental resolution gives a good fit. The film thickness is around 450 Å and the SLD of the film is 0.93 x 10-6 Å-2. This SLD value is substantially larger than that of the bis-amino

83 silane monomer 0.299 x 10-6 Å-2; so based on values in Table 4-1, the silanes must be highly

condensed. Assuming the silane is fully condensed, the density of the film is calculated to be

1.18 g/cm3. If the film were not fully condensed, a density higher than 1.18 g/cm3 would be

required to fit the data. Therefore, we conclude that the minimum film density is 1.18 g/cm3. At

any rate, the film is highly condensed and most, if not all, of the methoxy groups have been

transformed to –Si-O-Si– bonds.

From the best-fit SLD profile for the d-NB swollen bis-amino silane sample (Figure 4-1b) both

the thickness and SLD of the swollen film have increased relative to the as-prepared film.

Additionally a substantial increase in the roughness had to be included to fit the data. We can

calculate the volume fraction of d-NB, ϕNB, in the swollen film since the SLD of the swollen film is just that of the two components (dry film and d-NB) weighted by their volume fraction

( SLD = ϕ NB SLDNB + ϕ silane SLDsilane ).

We find ϕNB = 17%. This rather low number is consistent with a high crosslink density, which limits swelling. For comparison, films of 3-glycidoxypropyl trimethoxysilane (GPS) swollen with d-NB gave an equilibrium volume fraction of 50% – 60%, whereas films of BTSE showed no swelling in d-NB.

The amount of swelling is a function of both the crosslink density and the solvent quality. In

89 the theory of Flory and Rehner the crosslink density (ve) is related to the polymer volume fraction (φP) and the polymer-solvent interaction parameter (χ) as follows:

2 NA []ln()1− φ p + φ p + χ1φ p ν e =− 1/3 (4-1) V1()φ p − φ p /2

84 where NA is Avogadro’s number and V1 is the molar volume of the solvent. Flory-Rehner relationship assumes ideal statistical behavior (Gaussian chains) between crosslinks, so it cannot be used in any quantitative sense for silanes. The relationship does, however, illustrate that the swelling (1-φP) is controlled by two factors, ve and χ. Assuming χ is comparable for these systems, one can infer that the crosslink density of bis-amino silane films is much higher than for

GPS films but not as high as for BTSE films. The greater crosslink density for bis-amino silane relative to GPS is consistent with the fact that bis-amino silane has 6-fold functionality whereas

GPS has only 3-fold functionality. The greater crosslink density for BTSE than for bis-amino silane might be due to the shorter bridging group of BTSE. i.e., at the same degree of condensation, BTSE would have more crosslinks per unit volume than bis-amino silane.

More importantly, the SLD of the swollen sample in Figure 4-1b is not uniform. A single-layer model cannot fit the reflectivity data, even with roughness. The best-fit SLD profile indicates little or no d-NB near the substrate. From equation (4-1), this result could indicate either a higher crosslink density or else an altered chemical composition near the substrate leading to a larger χ parameter (less compatibility). In Section 4.2.2 below we show that no such solvent exclusion effect is present for D2O. Since a high crosslink density at the interface should have excluded D2O as well, we attribute the solvent exclusion effect in Figure 4-1b to a gradient in the

chemical composition. This observation is consistent with the postulate of a distinguishable

layer between the bulk silane film and the substrate.86-88 Zhu and Van Ooij10 proposed that this

region was likely to contain bicarbonates due to the reaction of the bridging secondary amine

with CO2. Electrochemical impedance spectroscopy showed that the formation and dissolution of

these moieties did not actually influence the resistance of the film although it did influence the

85 frequency dependence of the impedance. Therefore, these authors concluded that these moieties

might not influence the film water-barrier performance.

The observation of low d-NB concentration near the substrate surface is similar to Yim et al.’s

9 finding for GPS films. However, in that case the effect was apparent for both d-NB and D2O and so it was attributed to a crosslink density gradient.

Interestingly, the SLD of the redried film is higher than the SLD of the as-prepared film, but the film is again smooth. It seems that the film might have restructured during swelling and densified on drying. It is more likely, however, that the film retains residual d-NB. In a similar experiment with GPS films no d-NB could be detected within the films after exposure to ambient dry air. In the case of bis-amino silane there may be an interaction between d-NB and the amine groups that is sufficient to cause retention of d-NB under these conditions. The presence of residual swelling leads us to believe that NB is a better solvent (smaller χ) for condensed bis- amino silane than for condensed BTSE, consistent with the complete exclusion of d-NB from

BTSE films.9 Further work involving evacuation of the samples could resolve this issue.

86 4.2.2 Bis-amino silane under hydrothermal conditioning

4.2.2.1 Exposure to D2O at room temperature for 14h

0 10

-1 10 D2O 25 °C 14 h Redried -2 10 As-prepared

-3 10

-4 10 Reflectivity

-5 10

-6 10 0.00 0.04 0.08 0.12 0.16

-1 q (Å )

(a)

4 ϕ D2O = 41% )

-2 3

Å D O 25 °C 14 h (rough) ( 2 Redried 2 As-prepared SLD x 6

10 1

0 0 200 400 600 Distance from Si (Å)

87 (b)

Figure 4-2: (a) Neutron reflectivity data from bis-amino silane film as-prepared, after exposure to D2O vapor at room temperature for 14 h and after re-drying. The curves through the data points correspond to the best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). “Rough” in the legend means roughness was included to fit the data. The film was found to be

rough in the swollen state. The calculated volume fraction of D2O in the swollen film is φD2O = 41%. (The data were generated on SPEAR at LANL).

Figure 4-2a shows the reflectivity from a bis-amino film as-prepared, after exposure to D2O at room temperature for 14 h, and again after re-drying. Figure 4-2b shows the SLD profiles for the bis-amino-D2O system. After conditioning, the reflectivity increased relative to the as-prepared film due to the absorption of D2O and the film roughness had to be assumed to fit the data. This

result shows that bis-amino silane is hydrophilic. The calculated volume fraction of D2O in the

swollen film is ϕ = 41%, implying substantially more swelling than found for d-NB (ϕ = D2O NB

17%).

After re-drying, the reflectivity almost returned to that of the as-prepared state and the film was smooth again. Also, on experimental time scales, there are no reactions that incorporate D into the film. At room temperature, D2O is physically absorbed into the silane film because all D2O is removed upon re-drying.

An interesting observation, revealed in the fitting, is an SLD peak at the air side surface for the swollen system. This result implies there is a hydrophilic layer near the air-side surface attracting the D2O. Near the substrate-side interface and within the native oxide layer, there is also a D2O-

rich layer, although not nearly as pronounced as at the air-side surface. Considering the

molecular structure of bis-amino silane, we propose that clustering of the secondary amino

groups might form the hydrophilic region at both the air and substrate side surface. The amino

88 groups have a higher surface energy than hydrocarbon chains, and thus they would not be

expected to be in excess at the surface in dry air. An excess of amino groups, however, may result upon exposure to water (liquid or vapor) through surface reconstruction. Regarding the slight excess of D at the substrate surface, this excess may simply reflect the presence of hydrophilic sites on the substrate surface not occupied by silane molecules. It is also possible that there is a small excess of amino groups in the substrate surface due to the upside-down orientation of silane molecules. This idea is based on the study of the tri-functional analog of bis- amino silane: γ-aminopropyltriethoxysilane (γ-APS). In most cases, silanes are adsorbed onto the

metal surface through silanol groups by forming oxane bonds with the hydroxyl groups on the

metal substrate.8 For γ-APS, however, some of the molecules are adsorbed onto the metal surface through amino groups.25, 90 In this case, the amino groups are protonated by the formation of

hydrogen bonds with the surface hydroxyl groups.25 Another interpretation is to attribute protonation of the amino groups to reaction of amino group with carbon dioxide and water to form amine bicarbonate salt, as mentioned earlier.10

H N H N O SI O SI O SI O O SI O Air surface O O O O O

Bulk film

H N O O O O O O Si O Si O Si OSi SI O O O O N Si Substrate surface O O O H OH N + - H2 HCO3

Substrate

89 Figure 4-3: Schematics of proposed structures of bis-amino silane film.

In γ-APS, both orientations exist and some XPS studies have been carried out to determine the

relative amounts of the two orientations by monitoring the amount of free and protonated amino

groups. Using angle-resolved XPS, Fowkes et al.90 showed that the protonated amino groups were located adjacent to the silicon oxide surface, whereas free amino groups were found at the air-side surface. For bis-amino silane, a similar orientation might occur, as illustrated in Figure

4-3. A fraction of the bis-amino silane might adsorb onto the substrate through the reaction of silanol groups with surface hydroxyl groups, leaving the secondary amino group free. However, bis-amino might also adsorb onto the substrate through the hydrogen bonding between an amino group and a surface hydroxyl group, leaving all the silanol groups available to form siloxane bonds. We conclude that the free amino groups close to the air surface attract the D2O molecules, resulting in a D2O-rich layer. Presumably, the layer formed at the substrate surface is less compatible with d-NB than that at the air side surface accounting for the depletion observed in the d-NB studies (Figure 4-1).

4.2.2.2 Exposure to D2O at 80 °C for 14 h

Figure 4-4a shows the reflectivity data from bis-amino silane films as-prepared, after exposure to

o o D2O at 80 C for 14 h and again after re-drying. The NR measurement on the 80- C conditioned sample was performed at room temperature. Figure 4-4b shows the best-fit SLD profiles. Two important observations should be noticed from these data. First, the bis-amino silane films conditioned with D2O at 80 °C show a large increase in reflectivity, as well as a D2O-rich layer near both the air-side surface and the substrate-side interface. Surprisingly, however, the films

swelled less after 80-°C exposure (ϕD2O = 30%) than after exposure at room temperature (ϕD2O =

90 41%). Second, after drying, the reflectivity remained elevated relative to that of the as-prepared sample over most of the q range. In addition, the film was found to be rough in both the swollen and redried states. Both of these observations indicate that a chemical change occurred upon heating in saturated D2O at 80 °C. Several reactions appear possible. The observed SLD of the redried film is 1.78 x 10-6 Å-2. Based on Table 4-1, some deuterium must be chemically incorporated into the redried film. We know from the SLD of the as-prepared film that bis-amino silane is highly condensed with very few residual methoxysilyl groups. Therefore, the irreversible component of D incorporation is either in the form of –ND from amine proton exchange (H → D) or hydrolysis of siloxane bonds. Even though hydrolysis of siloxane bonds was not observed with BTSE under these conditions,94 it may nevertheless occur at a measurable rate for bis-amino due to a higher local concentration of water in this more hydrophilic film.

However, we actually observe less swelling at 80 °C than at 25 °C, which is inconsistent with hydrolysis of Si–O–Si at 80 °C. The irreversible D incorporation is most likely exchange of the amine proton with a deuteron from D2O.

0 10 D O 80 ºC 14 h -1 2 10 Redried -2 As-prepared ty 10 vi -3 10 ecti

-4 10 Refl

-5 10

-6 10 0.00 0.04 0.08 0.12 0.16 -1 q (Å )

(a)

91 D O 80 °C 14 h (rough) 4 2 Redried (rough)

) As-prepared -2 3 D (Å

L ϕ 2 D2O = 30% x S 6

10 1

0 0 200 400 600 Distance from Si (Å)

(b)

Figure 4-4: (a) Neutron reflectivity data from bis-amino silane film as-prepared, after exposure to D2O vapor at 80 °C for 14 h and after re-drying. The curves through the data points correspond to the best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (The data were generated on SPEAR at LANL).

For BTSE exposure to D2O at 80 °C, a large increase in reflectivity is also observed. Kent and

Yim44 concluded that the increase in reflectivity was due to the hydrolysis of residual ethoxysilyl

groups, instead of physically absorbed water. This conclusion was based on the fact that no

change in reflectivity was observed for conditioning in H2O, whereas a strong effect would have resulted if the films simply swelled with H2O. In our case, however, the SLD of the as-prepared film indicates that the film is highly condensed. Considering the differences between BTSE and bis-amino silane, we conclude that BTSE is hydrophobic and not highly condensed whereas bis- amino silane is hydrophilic and highly condensed.

92 4.3 Conclusions

The SLD of the as-prepared bis-amino silane film shows that it is highly condensed, but it is

not uniformly crosslinked based on the result of d-NB swelling. The film shows reduced

swelling near the substrate consistent with an NB-incompatible layer that is attributed to

protonated secondary amine groups at the film-substrate interface. After re-drying, the films

show an increase in reflectivity over the as-prepared state. This increase is probably due to

retention of residual d-NB in the film.

The D2O-conditioned films have a D-rich layer near both air-side surface and metal substrate.

The D-rich layer is attributed to the hydrophilic character of the layer formed by the presence of free amino groups near the air-side surface and the protonated amino groups near the substrate.

Compared to BTSE, therefore, bis-amino silane is less effective as a water barrier even though it is more condensed.

Comparing the difference between D2O conditioning at room temperature and at 80 °C, there

is more swelling at room temperature than at 80 °C. Moreover, the reflectivity is not reversible

after exposure to D2O vapor at 80 °C whereas the reflectivity is reversible when swelled with

D2O vapor at room temperature. The increased SLD of the redried film after 80-°C conditioning

is attributed to the presence of –ND moieties in the film.

Chapter 5. Water Barrier Properties of Bis-sulfur and

Mixed Silane Films

Summary: Here we use neutron reflectivity to address the effect of bridging group on the

hydrothermal response of bis silane films prepared using bis-[3-(triethoxysilyl)

propyl]tetrasulfide (bis-sulfur) and bis-[trimethoxysilylpropyl]amine (bis-amino) as well their

mixture. The reflectivity is reversible at room temperature but not at 80 ºC, indicating chemical

alternation of the film at 80 ºC. Based on the scattering length density profiles the degree of

condensation as well as other structural information is obtained. Of particular interest is the fact

that a bis-sulfur/bis-amino mixture shows enhanced protective character in moist environment

compared to either neat silane. The superior properties of the mixed film are attributed to bis-

amino-catalyzed hydrolysis of residual ethoxy group of bis-sulfur silane, leading to better

coverage and superior performance. Variation in the structure normal to the substrate is also

examined by swelling the film with d-nitrobenzene (d-NB), a non-reacting swelling agent.

5.1. Introduction

In this chapter, we elucidate the water barrier properties of bis-[triethoxysilylpropyl]- tetrasulfide (bis-sulfur) and bis-[trimethoxysilylpropyl]amine (bis-amino) as well as mixtures of

these silanes using neutron reflectivity. Bis-sulfur silane, with an average of four sulfur atoms in

the bridge, is more hydrophobic but hydrolyzes slowly. In order to obtain a sufficient amount of

silanols for good film formation, at least 50-h hydrolysis time is required in water/alcohol

solutions at natural pH. Bis-amino silane, on the other hand, is more hydrophilic due to the

secondary amine in the bridge, and needs just 4 h to reach complete hydrolysis at pH 7.5.10 The

94 bare anti-corrosion performance (without top coating) of bis-amino, however, is not as good as

that of the bis-sulfur silane, presumably due to its more hydrophilic nature.

Van Ooij et al.34concluded that the silane film functions as a physical water barrier, no electrochemical effect. The direct evidence is DC polarization data, where silane treatment uniformly shifts the polarization curves to lower current density, but does not change their shapes (i.e., curve slopes). This observation shows that the silane film acts as a barrier, whose performance is related to its efficiency in keeping corrosive species (water, chloride, oxygen) away from the metal. Therefore, understanding the silane and water interaction is essential.

Zhu et al.12 proposed that enhanced performance of the mixed silane results from

overcoming the major drawbacks of the two individual silanes. Bis-amino silane makes the

mixture sufficiently hydrophilic to wet Zn oxide on HDG, whereas, bis-sulfur silane enhances

the hydrophobicity, which is the basis for good performance as a water barrier. We attempt to

further clarify the mechanism underlying these performance results by investigating the structure

of films and their response to water vapor.

This study builds on previous results91, 92 for neat bis-amino silane films swelled with

vapors of deuterated nitrobenzene (D-NB) and D2O. In this chapter, we report NR results for bis- sulfur and mixed silane films and compare the water swelling behavior of these films to that for bis-amino films. The goal is to investigate the corrosion protection mechanism of bis-sulfur and mixed silane, and to determine whether the mixed silanes have some unique morphology or response to water that might explain their outstanding ability to protect metals.

NR data shown in this chapter were performed on the Surface Profile Analysis

Reflectometer (SPEAR) at Los Alamos National Laboratory.

95 5.2. Results and Discussion

5.2.1 Reflectivity data of the as-prepared Film

0 10 -1 10 Bis-amino silane -2 10 Mixed silane -3 Bis-sulfur silane 10 ty -4 10 -5 10 ectivi

fl -6 10 e

R -7 10 -8 10 -9 10 -10 10 0.00 0.05 0.10 0.15 0.20 -1 q (Å )

Some structural features such as thickness and roughness can be inferred from the reflectivity data without modeling. Reflectivity curves from the as-prepared bis-amino silane, bis-sulfur silane and their mixture are displayed in Figure 5-1. For a clear view of the data, the reflectivity curve of mixed silane is shifted by 102 and the bis-sulfur silane is shifted by 104. The smaller fringe spacing of bis-amino silane compared to bis-sulfur silane indicates that bis-amino film is thicker. The thickness of mixed silane lies between the two individual silanes. Thickness obtained from ellipsometry measurements is in good agreement with these observations. From ellipsometry, the thickness of silane film deposited at the concentration of 1% is 226 Å for bis-

96 sulfur silane film, 300 Å for bis-amino silane film and 240 Å for the mixed film. The amplitude of the fringes is strongly affected by the SLD value, as well as the film uniformity. So, the sharper fringes of bis-amino silane can indicate that bis-amino silane has higher SLD than bis- sulfur silane, but it can also indicate that bis-amino silane is more uniform than bis-sulfur silane.

AFM test (Figure 5-2) shows the root mean square (RMS) roughness is 19.2 Å for bis-amino silane film and 26.8 Å for bis-sulfur silane. The roughness of mixed silane (25 Å) again, is in between. The SLD profiles, which will be presented later, show that the SLD of as-prepared bis- amino silane film is larger than bis-sulfur silane film. Therefore, the sharper fringes of bis-amino silane are due to both higher SLD and less roughness. The internal structure of the films is revealed in more detail after modeling as presented below.

(a)

(b)

98 5.2.2 Bis-sulfur silane

5.2.2.1 Nitrobenzene conditioning

0 10

-1 10 Bis-sulfur silane NB 25 °C As-prepared -2 d-NB 25 °C y 10 it iv t -3 c 10 e l f

e -4 R 10

-5 10

-6 10 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 -1 q (Å ) (a)

Bis-sulfur silane NB 25 ºC 3 As-prepared ) d-NB 25 ºC -2 D (Å

L 2 S

x ϕ

6 DNB = 5.9%

10 1

0 0 100 200 300 Distance from Si (Å)

(b)

99 profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a) for the samples as-prepared and after swelling to equilibrium with d-NB. The calculated volume fraction of d-NB

in the swollen film is ϕNB = 5.9%. (NR was performed on the SPEAR at LANL).

Reflectivity data for films of bis-sulfur silane as-prepared and after swelling to equilibrium in

d-NB vapors are shown in Figure 5-3a. The change in reflectivity is small, indicating the

marginal inclusion of solvent in the film. However the increase in thickness is noticeable from

the shift in the Kiessig fringes to lower q. The best-fit SLD profiles corresponding to the lines

through the data in Figure 5-3a are shown in Figure 5-3b. For the as-prepared state, the film

thickness is 222 Å, which is in consistent with the result obtained from ellipsometry (226 Å).

The SLD of the dry film is 0.61 x 10-6 Å-2, which is substantially higher than the SLD of the bis-

sulfur monomer (0.21 x 10-6 Å-2). The difference is due to the fact that Si-O-Si is formed in the hydrolysis and condensation process, replacing the ethoxy groups that have lower SLD. A fully condensed bis-sulfur silane film has an SLD of 0.67 x 10-6 Å-2, which is calculated based the

density of the monomer (1.1 g/cm3). Since the bulk density of the film should increase after condensation, the SLD of a fully condensed bis-sulfur silane should actually be larger than 0.67 x 10-6 Å-2. At any rate, the measured SLD of the film (0.61 x 10-6 Å-2) is marginally less than the calculated SLD of fully condensed bis-sulfur silane film. We conclude that bis-sulfur silane film

is not fully condensed, at least not as condensed as bis-amino silane, where the SLD is greater

than that of the fully condensed silane assuming no change in density upon crosslinking.91 For bis-sulfur silane, apparently there are unhydrolyzed ethoxy groups or uncondensed silanol groups present. Later, in the discussion of the D2O conditioning study, we provide further support for this conclusion and estimate the degree of condensation.

100 From the best-fit SLD profile for the d-NB swollen bis-sulfur silane sample (Figure 5-3b) both

the thickness and SLD of the swollen film have increased relative to the as-prepared film. The

thickness of the film increases from 231 Å to 252 Å. We can calculate the volume fraction of d-

NB, ϕNB, assuming that the SLD of the swollen film is that of the two components (dry film and

d-NB) weighted by their volume fractions ( SLD = ϕ NB SLDNB + ϕ dry− film SLDdry− film ). We find ϕNB

= 5.9%. This value is much lower than the φNB of bis-amino silane, which is 17%. Because the

amount of swelling is a function of both the crosslink density and the solvent quality, this result

indicates that bis-sulfur silane either has a higher crosslinking density or a larger polymer solvent

interaction parameter (χ) with d-NB than bis-amino silane.

Based on the study of D2O conditioning of bis-sulfur silane, which will be explained below, we conclude that bis-sulfur silane is less condensed than bis-amino silane. Since the bridging group of bis-sulfur silane is longer than that of bis-amino silane, the possibility that bis-sulfur silane has higher crosslink density can be ruled out. We conclude, therefore, that NB is a better solvent (smaller χ, the polymer-solvent interaction parameter) for condensed bis-amino silane film than bis-sulfur silane film.

In contrast with the result for bis-amino silane,91 the SLD of the bulk layer of the film

increases uniformly when exposed to NB. That is, there is no evidence for altered chemical

composition near the substrate or the air surface as is for the case of bis-amino silane.14

5.2.2.2 D2O conditioning

Figure 5-4a shows the reflectivity of bis-sulfur film as-prepared, after exposure to D2O at room temperature for 15 h and after re-drying. In contrast to the result of NB swelling, after conditioning, the change in thickness is negligible. An increase in R(q), however, is noticeable

101 from the fact that the first minimum is lower and the tail of the curve is slightly higher relative to

the as-prepared state. The SLD profiles for the data in Figure 5-4a are shown in Figure 5-4b.

The elevated SLD after exposure to water indicates the inclusion of deuterium, which could

result from either absorption of free D2O, or from deuterium incorporated chemically into the

film. After redrying, the reflectivity almost returns to that of the as-prepared state, indicating that

almost all water is physically absorbed. The calculated volume fraction of D2O in the swollen film is 7.8%. The thickness of the swollen film is 227 Å, which is very close to the thickness of

the as-prepared state 224 Å.

0 10

Bis-sulfur silane D2O 25 ºC -1 10 As-prepared D O 25 ºC 15 h -2 2 10 ty Redried vi -3 10 ecti

-4 Refl 10

-5 10

-6 10 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 -1 q (Å )

(a)

102 Bis-sulfur silane D2O 25 ºC As-prepared 3 ) D O 25 ºC 15 h -2 2 Redried

x SLD (Å ϕ D2O = 7.8% 6

10 1

0 -50 0 50 100 150 200 250 300 Distance from Si (Å)

(b)

Figure 5-4 (a): Neutron reflectivity data from bis-sulfur silane film as-prepared (○), after exposure to D2O vapor at room temperature for 15 h (+) and after re-drying (∆). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a) The calculated volume fraction of D2O in the swollen film is 7.8%. (NR was performed on the SPEAR at LANL).

Recall that after exposure to d-NB, the volume fraction of NB in the film is 5.9% and the thickness increases from 231 Å to 252 Å. Thus, for bis-sulfur silane the change in film thickness is greater for d-NB than for water even though the volume fraction of the water is greater than that of d-NB. These results show that d-NB swells the polymer chain whereas water resides in the free volume. NB is therefore a better solvent than water or water enhances condensation of the residual ethoxy group, which would counteract chain expansion.

For comparison, the calculated water volume fraction within bis-amino silane films at room temperature is 41%.91 Bis-sulfur silane films absorb far less water even though they are less

103 condensed than bis-amino silane films. The –Sx– bridging group enhances the hydrophobicity of the film.

For 80-°C conditioning, surprisingly, the thickness of the film decreases after conditioning as indicated by the shift of Kiessig fringes to higher q (Figure 5-5a). The data are displayed as reflectivity × q4 for clarity. From the SLD profiles shown in Figure 5-5b, we know that the thickness of the film decreases from 225 Å to 215 Å upon conditioning. In addition, the reflectivity does not return to the original state after redrying. The thickness of the redried film is close to that of the wet film, 216 Å, indicating that a chemical change has occurred in the film.

-8 10 8 6 4

2 4 -9 10 8 6 4

-10 Reflectivity*q Bis-sulfur silane D2O 80 ºC 10 8 6 As-prepared

4 D2O 80 ºC 14 h Redried 2

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16

-1 q (Å )

(a)

104 Bis-sulfur silane D2O 80 ºC 3 As-prepared ) D O 80 ºC 14 h -2 2 Redried D (Å

L 2

x S ϕ D2O = 7.2% 6

10 1

0 -50 0 50 100 150 200 250 300 Distance from Si (Å)

(b)

Figure 5-5: (a) Neutron reflectivity data from bis-sulfur film as-prepared (●), after exposure to D2O vapor at 80 °C for 14 h (+) and after re-drying (∆). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit scattering length density profiles corresponding to the curves through the data in (a). The volume fraction of free water in the swollen film is 7.2%. (NR was performed on the SPEAR at LANL).

After continued exposure to D2O at 80 °C for 2 days, the film shrinks even more to 205 Å.

The SLD profiles are plotted in Figure 5-6. From this plot, we see that at room temperature the

thickness of the film did not increase despite the 7.2 vol % absorbed D2O. Apparently the D2O fills in the free volume of the silane film rather than swelling the polymer chains. After 80-°C conditioning for 14 h, the film shrinks even though the SLD is almost the same as when swelled at room temperature. The film continues to shrink when the film was conditioned at 80 °C for another 2 days. These results all indicate that chemical alternation must have occurred during 80-

°C conditioning. Considering the structure of the bis-sulfur silane, it is likely that ethoxy silyl groups (Si–OCH2CH3) hydrolyze and condense leading to a higher crosslinking density.

105 Bis-sulfur silane D2O 3 As-prepared )

-2 D2O 25 ºC 15 h

D2O 80 ºC 14 h D (Å

L 2 D2O 80 ºC 2 days x S 6

10 1

0 -50 0 50 100 150 200 250 300 Distance from Si (Å)

Information relating to the nature of the chemical reactions occurring during 80-°C

conditioning can be obtained from the SLD changes. The SLD profiles do not return to the as-

prepared state after redrying. So, not only is free water absorbed, but also the chemical structure

of the film changes. The amount of free water can be calculated from the SLD change upon

drying, which is 4.6 vol %. If all the ethoxysilyl groups condense to Si-O-Si, the SLD would

only reach 0.67 x 10-6 Å-2, assuming the density is unchanged. The SLD of the as-prepared film

is 0.61 x 10-6 Å-2. The SLD of the redried film after 80 °C, however, is 0.79 x 10-6 Å-2.

Densification of the film due to further condensation could account for the increase in SLD following 80-˚C conditioning. There also might be some deuterium included in the film. For the

latter case, there are three possibilities: hydrolysis of –Si-O-Si– to form –Si-OD, hydrolysis of

residual ethoxysilyl groups –Si-O-C2H5 to form –Si-OD, and H-D exchange of the uncondensed silanol groups (–Si-OH to –Si-OD). Previous work15 shows that the bridging siloxane bond (–Si-

106 O-Si–) is very robust. Siloxane bonds in bistriethoxysilyethane films are not hydrolyzed at 80

°C for up to 42 days. Therefore, hydrolysis of siloxane bonds can be reasonably ruled out. Since

the SLD values of –OC2H5 and –OH are very similar, we can not distinguish the existence of –

Si-OC2H5 and –Si-OH. But, since the SLD increases in the redried film, we can conclude that the bis-sulfur silane is not fully condensed in the as prepared state.

The combined fraction of unhydrolyzed ethoxy and uncondensed silanol groups in the as- prepared film can be estimated based on the increase in SLD of the redried film after 80-°C conditioning for 14 h. Assuming constant density and complete transformation of residual Si-

OC2H5 and –Si-OH to –Si-OD and Si-O-Si in the redried film, the fraction of –Si-OD moieties

per bis-sulfur molecule in the redried film is 11%. Based on the shrinkage of film, some

siloxane linkages are formed in the redried film by further hydrolysis and condensation of

ethoxysilyl groups. The amount of unhydrolyzed ethoxy or uncondensed silanol in the as- prepared film, therefore, must be more than 11%.

107 5.2.3 Mixed bis-sulfur/bis-amino silane film

5.2.3.1 NB conditioning

-8 10 8 6 Mixed silane NB 25 °C 4 As-prepared d-NB 25 °C

4 2 q -9 x 10 8

ty 6

vi 4 ti

ec 2 fl -10 Re 10 8 6 4

2 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16

-1 q (Å )

(a)

3.0 Mixed Silane NB 25 ºC 2.5 As-prepared ) d-NB 25 ºC -2 2.0 D (Å

L 1.5

ϕDNB = 5.2% x S 6 1.0 10

0.5

0.0 0 100 200 300 Distance from Si (Å)

(b)

108 Figure 5-7: (a) Neutron reflectivity data from mixed silane film as-prepared and after swelling to equilibrium with d-NB. The curves through the data points correspond to the best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a). The

calculated volume fraction of d-NB in the swollen film is φNB = 5.2%. (NR was performed on the SPEAR at LANL). The result for the mixed silane film under NB conditioning is shown in Figure 5-7. The

reflectivity data are displayed as reflectivity × q4 for clarity. Again, the increase in thickness is noticeable from the shift in the Kiessig fringes to lower q, but the increase in intensity is negligible. From the SLD profiles shown in Figure 5-7b, the thickness of the as-prepared mixed silane is exactly the same as obtained from ellipsometry measurement (240 Å), which lies between the thickness of bis-sulfur silane film (222 Å) and bis-amino silane film (410 Å). The

SLD, 0.77 x 10-6 Å-2, is also in between the SLD of bis-sulfur silane (0.61 x 10-6 Å-2) and bis-

amino silane (1.11x 10-6 Å-2). The calculated volume fraction of NB is 5.9%, close to the value

of bis-sulfur silane. The SLD of the bulk film also increases uniformly when exposed to NB,

indicating a uniform structure of the film normal to the surface. Based on the information above,

the structure of the mixed silane lies between bis-sulfur silane and bis-amino silane. The film is

closer to structure of bis-sulfur silane because of the higher fraction of bis-sulfur silane in the

mixture.

5.2.3.2 D2O conditioning

The reflectivity curves as well as the corresponding SLD profiles for mixed silane under D2O

conditioning are shown in Figure 5-8 for 25 °C and Figure 5-9 for 80 °C. The increase in

reflectivity of the swollen films is greater than that of bis-sulfur film, indicating more water is

absorbed.

109 2 Mixed silane D2O 25 °C -8 10 As-prepared 8 6 D2O 25 °C 12 h 4 4 Redried

2 ty x q

vi -9 10 8 6 ecti 4 Refl 2

-10 10 8 6 0.02 0.04 0.06 0.08 0.10 0.12 0.14

-1 q (Å )

(a)

4 Mixed Silane D2O 25 ºC As-Prepared 3 D2O 25 ºC 12 h )

-2 Redried D (Å

L 2 ϕ = 12.6% D2O x S 6

10 1

0 0 100 200 300 Distance from Si (Å)

(b)

Figure 5-8 (a): Neutron reflectivity data from mixed silane film as-prepared (○), after exposure to D2O vapor at room temperature for 12 h (+) and after re-drying (∆). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a). The calculated volume fraction of D2O in the swollen film is 12.6%. (NR was performed on the SPEAR at LANL).

110 For 25-°C conditioning as shown in Figure 5-8, the calculated volume fraction of free water for mixed silane is 12.6%. It is known that free water absorption at 25 °C for bis-sulfur silane is

7.8% and 41% for bis-amino silane. Based on these values, we can conclude that, at least for silicon substrates and the thickness of films we used, the water barrier ability of mixed silane is not enhanced. The water volume fraction of the mixed silane film is only slightly less than that of both components weighted by their volume fraction. Surprisingly, the thickness of the “swollen” film decreased compared to the as-prepared state, even though the SLD indicates 13% water within the film. The decrease in film thickness is clearly seen in the shift of the Keissig fringes to higher q. At 80-°C conditioning, as shown in Figure 5-9, the shrinkage is more obvious.

Mixed silane D2O 80 °C -8 10 8 As-prepared

4 6 D2O 80 °C 4 Redried y x q t

i 2 v

-9

ecti 10 8 6

Refl 4

-10 10 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 -1 q (Å )

(a)

111 Mixed silane D O 80 ºC 4 2 As-prepared

) D2O 80 ºC 11 h -2 3 Redried D (Å L 2 ϕ = 13% D2O x S 6 10 1

0 0 100 200 300 Distance from Si (Å)

(b)

(a) for the samples as-prepared (—), after exposure to D2O vapor at 80 °C for 11 h (····) and after re- drying (−···−···). The volume fraction of free water in the swollen film is 13%. (NR was performed on the SPEAR at LANL).

In the case of neat bis-sulfur silane, shrinkage only occurs at 80 °C, not at 25 °C.

Undoubtedly the shrinkage is due to the condensation of residual ethoxy groups upon

conditioning. We already concluded that bis-amino silane film is highly condensed and most, if

not all, of the methoxy groups have been transformed to –Si-O-Si– bonds. Therefore, the

shrinkage of the mixed silane film is mainly due to the transformation of the ethoxy groups of

bis-sulfur component to –Si-O-Si- bonds. Apparently, for the mixed silane film, the condensation

of residual ethoxysilyl groups happens even at 25 °C.

The fact that more film shrinkage occurs upon conditioning for the mixed film than for the

pure bis-sulfur film suggests that condensation is accelerated in the mixed silane film. Bis-amino

112 silane, acts as a catalyst in the hydrolysis of bis-sulfur silane leading to more silanols groups in

the solution, which both improve the wettability of the precursor solution and leads to more

rebust films when exposed to water vapor. This effect accounts for the enhanced performance of

mixed silane compared to individual silanes. Temperature also accelerates the hydrolysis of

residual ethoxy group. Therefore, high temperature curing might be another factor that can

improve the performance of mixed and bis-sulfur silane.

After redrying following the 80-°C conditioning, the SLD of the mixed films does not return to

the as-prepared state. Based on the discussion above, the increase of the SLD of the redried film

relative to the as-prepared film can arise from the following sources: the condensation of

unhydrolyzed ethoxysilyl groups to siloxane bonds; the formation of Si-OD from hydrolysis of

ethoxysilyl groups for bis-sulfur silane; and the exchange of the amine proton with a deuteron for bis-amino silane.

For the mixed silane film, after exposure to D2O both at 25 °C and at 80 °C, the reflectivity

data are consistent with an excess of D2O at the air surface. Since no such effect is observed for

bis-sulfur silane film, this D-rich layer must be due to the bis-amino silane in the mixture. As

discussed in our previous paper, for bis-amino silane, the D-rich layer is attributed to the

hydrophilic character of the layer formed by the presence of free amino groups near the air-side

surface.

As for the interface with the substrate, it should be noticed that for all the conditioned films,

there is a thin, elevated-SLD layer adjacent to the silicon oxide surface. A much lower level of

water is present in the bulk of the silane films than at either interface. In the case of bis-amino

silane, the elevated SLD adjacent to the silicon oxide surface is distinct enough to appear as an

113 extra layer. But in most other cases, it appears as the enlarged oxide layer thickness or increased

SLD for the oxide.

5.3. Conclusion

1: Bridging group is the key factor that controls the morphology and water-barrier properties of silane films. Bis-sulfur silane is not as condensed as bis-amino silane. There are more than

11% uncondensed groups per bis-sulfur molecule in the as-prepared film. Bis-sulfur film swells less in water because of the hydrophobic nature of bridging group. By contrast, bis-amino film is more hydrophilic since the secondary amine group hydrogen bonds with water. Bis-amino silane films are thicker and smoother than bis-sulfur silane films prepared at the same concentration.

2: At elevated temperature, water-conditioning leads to further chemical reactions for all films studied. After re-drying following room-temperature conditioning, the reflectivity curves of bis- amino, bis-sulfur as well as the mixed film all return to the as-prepared profile indicating no chemical reaction occurred at room temperature. With 80-°C water-vapor conditioning, however, the reflectivity of the redried film remains elevated relative to the as-prepared film due to formation of Si-O-Si- and Si-OD in bis-sulfur silane and the exchange of the amine proton with a deuteron in bis-amino silane. After 80-°C vapor conditioning, the thicknesses of the bis-sulfur film and mixed silane film decrease, which is consistent with the condensation of ethoxysilyl groups to Si-O-Si

3: The improved anti-corrosion performance of the mixed film is traced to modification of the chemistry in both the film and the precursor solution. Based on the enhanced shrinkage that occurs following water-vapor conditioning of the mixed film, condensation is accelerated in the mixed silane. Regarding to the precursor solution, bis-amino silane may act as a catalyst in the

114 hydrolysis of bis-sulfur silane leading to more silanols groups in the solution, which in turn improves the wettability of the solution. More uniform coverage of the substrate, therefore, can be obtained. Therefore, accelerating the hydrolysis of residual unhydrolysed group of hydrophobic silane is the key to improving wettability and anti-corrosion performance.

Although the redried films show enhanced shrinkage and improved coverage of the substrate is expected for the mixed film, the vapor-swollen state does not show any unusual character.

The water volume fraction of the mixed silane film swells is roughly that of both components weighted by their volume fraction.

4. Some excess water was detected at the interface between the silane films and the silicon wafers. This absorption might due to un-reacted OH groups on the silicon wafer surface forming a hydrophilic layer that attracts water through hydrogen bonding.

115

Chapter 6. Morphology and Water-Barrier Properties of

Silane Films: The Effect of Substrate2

Summary: The goal of this study is to understand the effect of the substrate on the morphology

and water-barrier properties of silane films. Silane films are deposited on both Si and Al.

Neutron reflectivity is used to assess the effect of hydrothermal conditioning on the films.

Aluminum on silicon (no silane) was characterized first to understand better the more

complicated silane on Al-coated Si. A 200-Å Al layer with 55-Å oxide covers the surface of the

silicon wafer. The reflectivity data show that water penetrates into the oxide. Films deposited on

either Al or Si substrates have similar bulk and top-surface morphology. Studies of silanes on Si

wafers, therefore, can be generalized to include Al. The substrate-silane interface, however, does depend on both the substrate and the silane. Because pH of the bis-sulfur silane solution is outside of the stability range for Al2O3, dissolution of the thin oxide film occurs during solution

deposition. A water-depletion area is formed at the interface region due to this reaction.

6.1. Introduction

Using neutron reflectivity (NR) we have completed a comprehensive study on the morphology

of silanes films deposited on silicon wafers, and the response of these films to water conditioning

in chapter 4 & 5. The systems we studied are bis-[triethoxysilylpropyl]tetrasulfide (bis-sulfur)

2 This chapter has been published on Thin Solid Films, 2006, 503, 259.

116 and bis-[trimethoxysilylpropyl]amine (bis-amino) as well as mixtures of these silanes. One

important feature revealed in the SLD profile of films deposited on Si wafer is that there is a 10-

Å hydrophilic layer at the oxide-silane interface in all three films. This layer might due to

unreacted OH groups on the silicon wafer surface forming a hydrophilic layer that attracts water

through hydrogen bonding.

Studies have shown that metal oxides strongly affect the molecular structure as well as other

properties of silane films.6, 25 Metal oxide surfaces may differ due to isoelectric point (IEP),

solubility of the metal hydroxide in water as well as the density and acidity of hydroxyl groups.8

Therefore, we expected that the amount of unreacted –OH depends on the substrate. Al alloy substrates might have different density of unreacted –OH that could influence the deposited silane film.

NR on Al substrate was carried out to study the effect of substrate on the water barrier properties. Because a thick silicon wafer is required to achieve the smoothness required for neutron reflectivity, an Al layer was e-beam evaporated on a silicon wafer. The film thickness is around 200 Å and roughness is about 10 Å. Silane films were then applied to the Al surface using the same spin-coating technique. NR data were obtained on virgin films and on the same films following the hydrothermal conditioning.

NR data shown in this chapter were performed on NG 7 reflectometer at National Institute of

Standards and Technology (NIST) and also on the POSY II reflectometer at Argonne National

Laboratory (ANL).

117 6.2. Results and Discussion

4 Silane on Si SiO2

) 3 Silane films -2 Å D ( L 2

x S Si Silane layer 6 Silicon wafer 10 1

0 0 100 200 300 Distance from Si (Å)

5 Al on Si

4 SiO2 )

Al layer -2 Al2O3 3 D (Å 2

x SL Al Silicon wafer 6 Si 10 1

0 0 100 200 300 Distance from Si (Å)

5 Silane on Al coated Si

Silane films 4 SiO2 ) Al2O3 -2 Å Al layer 3

2 Al x SLD ( 6 Si Silane Silicon wafer 10 1

0 0 200 400 600 Distance from Si (Å)

Figure 6-1: The systems studied and the corresponding representative SLD profiles.

It is of interest to study the uncoated substrates as well as the silane-coated systems. The three

systems, as well as the representative SLD profiles, are shown in Figure 6-1: silane on silicon, Al

on silicon and silane on Al-coated silicon. The SLD profile of silane on Al is a combination of

the SLD of Al on silicon wafer with the SLD profiles of silane on silicon wafer. The reflectivity

curves from silane on Al-coated silicon do not have equally spaced fringes. Instead, some double peaks are obtained because of the contribution of both Al layer and silane film, as shown in

Figure 6-2. For a clear view of the data, the reflectivity curve of silane on Al-coated silicon is

118 shifted by 102 and the silane film on silicon is shifted by 104. In the following section, we will first discuss the results for Al on silicon wafer. Then the results of the complicated system of silane on Al-coated silicon wafer will be discussed.

1 10 Al on Si -1 Silane on Al-coated Si 10 Silane on Si

-3 10 From Al

-5 10

-7 10 Reflectivity

-9 10 From Silane -11 10 0.00 0.02 0.04 0.06 0.08 0.10 0.12 -1 q (Å )

6.2.1 Al layer on silicon wafer.

Figure 6-3a shows the reflectivity data for films of Al on silicon wafer as-prepared and after

exposure to D2O vapor at room temperature for 24 hours. The best-fit SLD profiles corresponding to the lines through the data in Figure 6-3a are shown in Figure 6-3b. A layer of

Al2O3 forms rapidly on an aluminum surface even in high vacuum. Therefore, a 4-layer model

(Si, SiO2, Al, Al2O3) is used to fit the Al-on-silicon-wafer data. The Al layer is 215 Å thick with

55-Å Al2O3 surface layer. Polarimetric study of evaporated films exposed to the atmosphere

119 indicates that the surface oxide layer is from 20 to 55-Å thick but thicker layers readily form in

moist environments.93 Thickness obtained from ellipsometry and X-ray reflectivity (not shown) is consistent with the result obtained with NR. The SLD of Al obtained is 2.1 x 10-6 Å-2, which is

-6 -2 -6 -2 close to the theoretical value 2.5 x 10 Å . But the SLD of Al2O3 3.2 x 10 Å is much lower than the theoretical value 5.6 x 10-6 Å-2.

0 10 AL on Si

-1 As-prepared 10 D2O 25 ºC ty -2 vi 10 ecti -3 fl 10 e R

-4 10

-3 10 20 30 40 50 60 70 80x10 -1 q (Å )

(a)

120 4 )

-2 3 Å (

2 SLD x

6 Al on Silicon As-prepared 10 1 D2O 25 °C

0 0 100 200 300

Distance from Si (Å)

(b)

Figure 6-3 a: POSY NR data from Al layer on silicon wafer as-prepared and after exposure to D2O vapor at room temperature for 24 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the POSYII reflectometer at ANL).

Enhanced reflectivity in the presence of D2O vapor (Figure 6-3a) implies absorption of D2O.

The SLD profile (Figure 6-3b) also indicates penetration of D2O into the oxide. This penetration could be due to oxide porosity or due to surface roughness caused by an island-like morphology.

Neutron reflectivity measures the concentration gradients normal to the surface that is averaged over the coherence length of neutrons in the lateral direction. The coherence length is on the order of microns, so the effect of island morphology will be the same as porous structure in SLD profiles.

Based on the distance of water penetration, the difference between the highest point and lowest point on the surface is about 50 Å, equivalent to the thickness of the oxide implying an

121 exceedingly rough layer. Such a layer would be expected to show strong off-specular scattering,

which is not observed based on data taken on the SPEAR reflectometer at Los Alamos National

Laboratory (not shown). The ESEM image, however, shows some surface irregularity (Figure 6-

4). At this point we cannot definitively rule out island-morphology, but the porous oxide is more

consistent with the collection of observations.

Figure 6-4: ESEM image of Al layer on silicon wafer. The bar is 200 nm.

6.2.2 Silane on Al-coated silicon wafer

6.2.2.1 Room temperature conditioning

Figure 6-5a shows the reflectivity of bis-amino film as-prepared, after exposure to D2O at

room temperature for 23 hours. After conditioning the reflectivity increases relative to the as-

prepared film due to the absorption of D2O. The reflectivity from silane films is so strong that the peaks from Al layer are masked. Therefore, there are no obvious double peaks observed in the reflectivity curves of conditioned film. The SLD profiles corresponding to the fits in Figure 6-5a

122 are shown in Figure 6-5b. The silane layer thickness is 276 Å and SLD is 1.2 x 10-6 Å-2. These

values are close to the results on silicon wafer substrate. The elevated SLD after exposure to

water indicates the inclusion of deuterium in the silane film. Based on the argument above, the

Al2O3 is very porous; therefore, the SLD of Al2O3 is allowed to vary during fitting. To obtain the best fit, a water penetration distance into the oxide layer of 46 Å is required. This result confirms

the conclusion obtained from the study of Al on silicon.

0 10

-1 Bis-amino on Al 10 Dry -2 10 D2O 25 °C ty

vi -3 10

ecti -4 10

-5 Refl 10

-6 10

-7 10 0.02 0.04 0.06 0.08 0.10 0.12 0.14 -1 q (Å )

(a)

123 Bis-amino on Al 5 Dry D O 25°C

) 2

-2 4

D (Å 3 L ϕ D2O = 43%

x S 2 6 10 1

0 0 200 400 600

Distance from Si (Å)

(b)

Figure 6-5(a): Neutron reflectivity data from bis-amino silane films on Al-coated silicon wafer as-

We can calculate the volume fraction of D2O, ϕD2O, assuming that the SLD of the swollen film

is that of the two components (dry film and D2O) weighted by their volume fractions

( SLD = ϕ D2O SLDD2O + ϕ dry− film SLDdry− film ). The calculated volume fraction of D2O in the

swollen film is 43%. This value is close to that for bis-amino silane film on silicon wafer

(41%).91 At this point, we conclude that silane films deposited on silicon wafer substrate and Al substrate have similar bulk properties. In other words, at least for bis-amino silane films, the substrate does not affect the silane film deposited on it. This conclusion is modified below for bis-sulfur silane film.

124 In contrast to the results for comparable films on Si substrates,91 we find no D-rich layer on

the air side for D2O-conditioned films on Al. For Si substrates we proposed that the D-rich layer is caused by orientation of the free amino group to the air-side forming a hydrophilic region when exposed to water. The amino groups have a higher surface energy than hydrocarbon chains, and thus they would not be expected to be in excess at the surface in dry air. An excess of amino groups, however, may result, upon exposure to water (liquid or vapor) through surface reconstruction. The fact that a D-rich layer is shown at the air side on silicon substrate not on Al substrate is troublesome and may be due to intrinsic limitations of inversion of scattering data.

These films are thick enough that substrate should not make a difference on the morphology of the top surface.

Bexell et al.24 investigated the influence of different metal substrates on the structure and

composition of the silane film with XPS and AES. In their study, the hydrolyzed γ-

mercaptopropyltrimethoxysilane (γ–MPS) was deposited on Al, Zn and Al-Zn-alloy-coated steel.

The results show that the silane film covers the entire substrate surface, although the thickness is

non-uniform. The majority of the silane molecules are oriented randomly within the silane films

on all of the investigated substrates. Another study94 on γ–MPS shows that thiol groups were

randomly oriented at the topmost surface. Only for monolayer films does the substrate affect the

orientation of silane molecules close to the substrate interface.25, 90

The water-conditioning result of mixed silane is shown in Figure 6-6. The SLD of the mixed silane is lower than that of bis-amino silane, so the reflectivity from the silane is weak.

Therefore, the reflectivity curve from mixed silane does not show obvious double peaks. After conditioning, the reflectivity increases due to absorption of D2O. From the SLD profiles shown in Figure 6-6b, the thickness of the as-prepared mixed silane is 245 Å and the SLD is 0.79 x 10-6

125 -2 Å . The calculated volume fraction of D2O in mixed silane is 10%. All these results are almost the same as the result on silicon wafer. For mixed silane film deposited on silicon wafer substrate, the thickness is 240 Å and the SLD is 0.77 x 10-6 Å-2. The calculated volume fraction

is 12%.

0 10

-1 Mixed silane on Al 10 Dry -2 10 D2O 25 °C y t i

v -3 10

ecti -4 10

-5 Refl 10

-6 10

-7 10 0.04 0.08 0.12 -1 q (Å )

(a)

126 5 Mixed silane on Al Dry D O 25 °C ) 4 2 -2

D (Å 3 L ϕ D2O = 10%

x S 2 6 10 1

0 0 200 400 600 Distance from Si (Å)

(b)

Figure 6-6(a): Neutron reflectivity data from mixed silane films on Al coated silicon wafer as-prepared

and after exposure to D2O vapor at room temperature for 23 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST).

Based on the results of these two silanes, the hydrophilic layer between substrate and silane is more pronounced compared to silicon wafer substrate due to the large amount of water in the

“pores” of the oxide layer. The effect of the oxide layer porosity overwhelms any effect of difference in area density of surface hydroxyls.

The reflectivity from bis-sulfur silane is unusual, as shown in Figure 6-7. The reflectivity data are noisy and the films appeared to have rough surfaces. The reflectivity fits are not that good and the resulting Al layer profile differs from that of the bare Al layer on silicon wafer substrate.

Instead, the SLD profile of the Al layer is severely deformed. Some modification of the Al layer must have occurred. This phenomenon is not seen for films deposited on silicon wafer substrate.

127 According to Pourbaix diagrams of Al,95 Al is passive in the pH range 4-9. The pH of the bis- sulfur solution (3.9) is just outside the stability range for Al2O3. Therefore, some etching is expected to accompany the film-formation process. Since SiO2 is stable in this pH range no such effect is observed on Si wafers. It seems that this reaction leads to a water-barrier layer at the interface. According to the SLD profile of the film after exposure to water vapor, a water- depletion area is formed in the interface region. Due to the low quality of fitting, further experiments are needed to confirm this speculation.

0 10 -1 10 Bis-sulfur on Al -2 Dry 10 D2O 25 ºC ty -3

vi 10 -4 10 ecti -5 10 Refl -6 10 -7 10 -8 10 0.02 0.04 0.06 0.08 0.10 0.12 0.14 -1 q (Å )

(a)

128 4 Bis-sulfur on Al Dry

) 3 D2O 25 °C -2 D (Å 2 x SL 6

10 1

0 0 200 400 600

Distance from Si (Å)

(b)

Figure 6-7(a): Neutron reflectivity data from bis-sulfur silane films on Al coated silicon wafer as-prepared and after exposure to D2O vapor at room temperature for 23 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST).

The contact angles of the three silanes on Al substrate are 60° for bis-amino silane, 76° for bis- sulfur silane and 81° for mixed silane. These values are consistent with the results obtained on silicon wafer substrate. Again, this result confirms that when the silane film is thick enough the film morphology and top interface is unaffected.

Figure 6-8 shows the reflectivity data of bis-amino, bis-sulfur as well as mixed silane on Al substrate as-prepared and redried after exposure to D2O at room temperature. After re-drying, the reflectivity almost returns to that of the as-prepared state indicating that on experimental time scales, there are no reactions that incorporate D into the film. At room temperature, D2O is

129 physically absorbed into the silane film because almost all D2O is removed upon drying. These results correlate well with the results on silicon wafer substrates.

0 10

-1 Mixed silane on Al 10 Dry -2 10 Redried ty i -3 10 tiv

-4

flec 10

Re -5 10

-6 10

-7 10 0.04 0.08 0.12 -1 q (Å )

(a)

0 10

-1 Bis-amino on Al 10 Dry -2 Redried 10

-3 ivity 10

-4 10

-5 Reflect 10

-6 10

-7 10 0.02 0.04 0.06 0.08 0.10 0.12 0.14 -1 q (Å )

(b)

130 0 10 -1 10 Bis-sulfur on Al Dry -2 10 Redried y

t -3 i 10 v -4 10 ecti -5 10

Refl -6 10 -7 10 -8 10 0.02 0.04 0.06 0.08 0.10 0.12 0.14 -1 q (Å )

(c)

Figure 6-8: Neutron reflectivity data from (a) mixed silane film; (b) bis-amino silane film; (c) bis-sulfur silane film as-prepared (○), and re-drying (∆) after exposure to D2O vapor at room temperature. (NR data were generated on the NG7 reflectometer at NIST).

6.2.2.2 80-°C conditioning

Compared to the result of room temperature conditioning the major difference for all three silane systems is that the redried film after 80-°C conditioning does not return to the original state. Figure 6-9 shows the reflectivity data of bis-amino, bis-sulfur as well as mixed silane on Al substrate as-prepared and redried after exposure to D2O at 80 °C. Chemical modification must have occurred.

131 0 10 Mixed on Al D O 80 ºC -1 2 10 Dry -2 Redried 10

-3 10

-4 10

Reflectivity -5 10

-6 10

-7 10 0.04 0.08 0.12 0.16 -1 q (Å )

(a)

0 10 Bis-amino on Al D O 80 ºC -1 2 10 Dry -2 10 Redried

-3 10

-4 10

Reflectivity -5 10

-6 10

-7 10 0.04 0.08 0.12 0.16 -1 q (Å )

(b)

132 0 10

-1 Bis-sulfur on Al D2O 80 ºC 10 Dry -2 10 Redried

-3 10

-4 10

Reflectivity -5 10

-6 10

-7 10 0.04 0.08 0.12 0.16 -1 q (Å )

(c)

Figure 6-9: Neutron reflectivity data from (a) mixed silane film; (b) bis-amino silane film; (c) bis-sulfur

silane film as-prepared (○), and redried (∆) after exposure to D2O vapor at 80 °C. (NR data were generated on the NG7 reflectometer at NIST).

In our previous study of silane films on silicon wafer, the chemical reactions of the silane films

during 80-°C conditioning were proposed based on the SLD of the dry films and the thickness

change after water conditioning. For bis-amino silane the film is highly condensed and most, if

not all, of the methoxy groups have been transformed to –Si-O-Si– bonds. No further

condensation leading to the decrease in thickness occurs when conditioned at 80 °C. The

increased SLD of the redried film after 80-°C conditioning is attributed to the presence of –ND

moieties in the film.

By contrast, bis-sulfur silane is not fully condensed based on the SLD of the as-prepared film.

Therefore, further hydrolysis and condensation of unhydrolysed ethoxy groups to siloxane bonds

133 occurs when conditioned at 80 °C, as evidenced by the shrinkage of the film instead of swelling after exposure to water at 80 °C. Mixed silane films also shrunk after 80-°C conditioning due to further hydrolysis and condensation of bis-sulfur silane in the mixture. The 80-°C conditioning results for bis-amino silane, mixed silane and sulfur silane are shown in Figures 6-10, 6-11 and

6-12. The thickness change of the “swollen” films follows the same trend as discussed above.

These results also confirm that 80 °C is sufficient for bis-amino silane to reach fully condensed state, but for bis-sulfur silane, higher curing temperature is needed to fully hydrolyze and condense.

0 10 Bis-amino Al D O 80 ºC -1 2 10 Dry -2 10 D2O 80 ºC

-3 10

-4 10

Reflectivity -5 10

-6 10

-7 10 0.04 0.08 0.12 0.16 -1 q (Å )

(a)

134 6 Bis-amino on Al 5 Dry D O 80 °C

) 2 -2

Å 4

3 ϕ D2O = 43% x SLD (

6 2 10 1

0 0 200 400 600 Distance from Si (Å)

(b)

Figure 6-10(a): Neutron reflectivity data from amino silane films on Al coated silicon wafer as-prepared and after exposure to D2O vapor at 80 °C for 11 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST).

0 10

-1 Mixed silane on Al 10 Dry -2 10 D2O 80 ºC ty

vi -3 10

ecti -4 10

-5 Refl 10

-6 10

-7 10 0.04 0.08 0.12 -1 q (Å )

(a)

135 5 Mixed silane on Al Dry ) D2O 80 °C -2 4 (Å

D 3 ϕ D2O = 12% x SL

6 2 10 1

0 0 200 400 600

Distance from Si (Å)

(b)

Figure 6-11(a): Neutron reflectivity data from mixed silane films on Al coated silicon wafer as-prepared and after exposure to D2O vapor at 80 °C for 11 hours. The curves through the data points correspond to the best fits using model SLD profiles. (b): Best-fit SLD profiles corresponding to the curves through the data in (a). (NR data were generated on the NG7 reflectometer at NIST).

0 10

-1 Bis-sulfur on Al 10 Dry -2 10 D2O 80 ºC ty

vi -3 10

ecti -4 10

-5 Refl 10

-6 10

-7 10 0.04 0.08 0.12 0.16 -1 q (Å ) (a)

136 4 Bis-sulfur on Al Dry

) 3 D2O 80 °C -2 Å ( 2 SLD x 6 1 10

0 0 200 400 600

Distance from Si (Å)

(b)

Figure 6-12(a): Neutron reflectivity data from bis-sulfur silane films on Al coated silicon wafer as-

6.3 Conclusion

The films deposited on Al substrate and silicon wafer have similar bulk properties and top

surface morphology. We conclude that 200-Å silane films are thick enough that the substrate

does not affect the top surface or the bulk structure. After re-drying, the SLD curves of bis-

amino, bis-sulfur as well as mixture film all return to the as-prepared profile indicating no chemical reaction occurred at room temperature. At 80 °C, however, the reflectivity of the redried film remains elevated relative to the as-prepared film due to formation of Si-O-Si- as well the formation of Si-OD in bis-sulfur silane and the exchange of the amine proton with a deuteron in bis-amino silane. After 80-°C conditioning, the thicknesses of the bis-sulfur film and

137 mixed silane film decrease due to the condensation of ethoxysilyl groups to Si-O-Si crosslinks.

For all films studied, water penetrates up to the substrate. Apparently the substrate is not playing the key role for the water barrier property of silane.

Because pH of the bis-sulfur silane solution is outside of the stability range for Al2O3, dissolution of the thin oxide film occurred in the interface when this silane is deposited. A water- depletion area is formed in the interface region due to this reaction.

138 Chapter 7. Morphology and Water Barrier Properties of

Silane Films: The Effect of Curing Temperature

Summary: The morphology of silane films and the response of these films to water vapor were

studied mainly by neutron reflectivity. Contact angle, ellipsometry as well as X-ray reflectivity

were also used to determine the surface energy as well as thickness. The systems include bis-[3-

(triethoxysilyl) propyl]tetrasulfide (bis-sulfur) and bis-[trimethoxysilylpropyl]amine (bis-amino)

as well as their mixture. In previous chapter, a complete study has been done on these systems

with 80-°C curing temperature. In this paper, the curing temperature of 180 °C is applied on the

same systems. The effect of curing temperature on the morphology and water barrier property of

organosilane films is investigated. Higher curing temperature leads to an increase of the

crosslink density and degradation of bis-sulfur silane film. While for bis-amino silane film, no

further condensation reaction but only dehydration reaction is observed. The effect on the water

barrier ability of bis-amino silane film is negligible.

7.1 Introduction

The formation of protective silane films involves numerous chemical and physical processes.

Optimization of a variety of processing parameters is the foremost objective to obtain anti-

corrosion silane films.8 The pH of the solution and hydrolysis time have been manipulated in

any silane system to improve film quality.3, 10, 38, 39 Silane concentration has been used to control film thickness.19, 26, 43 One of the variables, which to date has received little attention in the optimization of silane films, is the curing temperature.

139 The idea of using silane as corrosion inhibitor is based on the hydrolysis and condensation of

silane to form siloxane bond between each other and oxane bonds between silane and metal

substrate.8 Evidence indicates, however, that silanes are initially hydrogen bonded to the

substrate. Covalent bonds form during curing.13 Two silanol groups (Si-OH) of different silane molecules condense and dehydrate and contribute to the formation of a Si-O-Si network.13

Using both linear voltammetry and electrochemical impedance spectroscopy (EIS), Van

Schaftinghen et al.2 showed that there is no measurable difference between the uncoated substrate and the fresh films, where fresh film means the measurement is made right after applying the solution—a curing time of 0 minute. Therefore, the curing of silane films generally improves the barrier properties and leads to better corrosion protection of the metallic substrate.

Various authors have shown that elevating the drying temperature increases the crosslink density of silane films. Van Schaftinghen et al.2 tested the bis-1,2-(triethoxysilyl)ethane (BTSE) films using infrared spectroscopic ellipsometry (IRSE) at different curing times. The results showed that the intensity of the absorption band corresponding to the silanols (Si-OH) decrease while the intensity of the absorption band corresponding to the Si-O-Si bonds increases with curing time. No change is observed after 40 min of curing at 200 °C. Based on EIS, spectroscopic ellipsometry (SE) and IRSE, Franquet et al.26 also concluded that the curing of the

BTSE layer improves the barrier properties of the film by forming a denser and more crosslinked thin layer. The barrier properties reach a maximum after 30 min of curing at 200 °C.

Complete studies on silanes films with the curing temperature of 80 °C have been conducted previously. Based on these studies of thin films cured at 80 °C, we concluded that in contrast to bis-amino silane, bis-sulfur silane does not reach the fully condensed state. These unhydrolyzed groups lead to poor wetting and non-uniform films. Although not as condensed as

140 bis-amino silane, bis-sulfur silane swells less in water than bis-amino silane because of the

hydrophobic nature of bridging group. By contrast, the bis-amino film is more hydrophilic since

the secondary amine group hydrogen bonds with water.

After conditioning in D2O at 80 °C for 14 h, the thickness of the bis-sulfur film decreases. The

film continues to shrink when the film is conditioned at 80 °C for another 2 days. We conclude

that this contraction is due to the condensation of ethoxysilyl groups to Si-O-Si cross-links. The

shrinkage is not seen in bis-amino silane, consistent with the conclusion that the film is highly condensed and most, if not all, of the methoxy groups have been transformed to Si-O-Si bonds.

For mixed silane, a decrease in film thickness is clearly seen from the shift of the Keissig fringes to higher q when water-vapor conditioned room temperature. With 80-°C conditioning, the shrinkage is more obvious. The fact that more film shrinkage occurs upon conditioning for the mixed film than for the pure bis-sulfur film suggests that condensation is accelerated in the mixed silane film. Bis-amino silane presumably acts as a catalyst in the hydrolysis of bis-sulfur silane leading to more silanols groups in the solution, which in turn improves the wettability of the solution. This effect might account for the enhanced performance of mixed silane compared to individual silanes. Temperature also accelerates the hydrolysis of residual ethoxy group.

Therefore, high temperature curing might be another factor that can improve the performance of

mixed and bis-sulfur silane.

To elucidate the above issues, films were cured at 180 °C and characterized following the

same procedures as films cured at 80 °C. The purpose is to understand the effect of curing

temperature on the morphology and water barrier properties of neat and mixed silane films.

141 Neutron reflection data shown in this chapter was obtained on the Surface Profile Analysis

Reflectometer (SPEAR) at Los Alamos National Laboratory (LANL) and also on the POSY II reflectometer at Argonne National Laboratory (ANL).

7.2 Results and Discussion

7.2.1 Neutron and X-ray reflectivity

7.2.1.1 Bis-sulfur silane

0 10

-1 Bis-sulfur 180 ºC Cure 10 As-prepared -2 10 D2O 25 ºC ty

vi -3 10

ecti -4 10

-5 Refl 10

-6 10

-7 10 0.05 0.10 0.15 0.20 -1 q (Å )

(a)

142 5 Bis-sulfur 180 ºC Cure As-prepared )

-2 D O 25 ºC 4 2 (Å D

L 3 x S

6 2 ϕ D2O = 2.0% 10 1

0 0 50 100 150 200 250

Distance from Si (Å)

(b)

Figure 7-1 (a): Neutron reflectivity data from bis-sulfur silane film cured at 180 °C as-prepared (○), after

exposure to D2O vapor at room temperature (+). The curves through the data correspond to best fits using model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (Data were taken on the SPEAR at LANL).

Reflectivity data for films of bis-sulfur silane as-prepared and after exposure to D2O vapor at room temperature are shown in Figure 7-1a. After conditioning, the increase in reflectivity is very small based on the fact that only the tail of the swollen curve is slightly higher relative to the as- prepared state. The best-fit SLD profiles corresponding to the lines through the data in Figure 7-1a are shown in Figure 7-1b. For the as-prepared state, the film thickness is 175 Å, which is substantially lower than the thickness (222 Å) of the bis-sulfur silane film cured at 80 °C. The

SLD of the dry film is 0.73 x 10-6 Å-2, higher than the film cured at 80 °C (0.61 x 10-6 Å-2). Both observations indicate that a denser film is formed when cured at 180 °C compared to 80 °C. The calculated volume fraction of D2O in the swollen film is 2% based on the equation of

143 SLDswollen− film = ϕ D2O SLDD2O + ϕ dry− film SLDdry− film . Considering that the volume fraction of D2O is

7.8% for films cured at 80 °C, this observation is consistent with the conclusion based on the dry film: when cured at 180 °C, a denser film is formed and less water is absorbed.

The denser films obtained at 180-°C curing might due to several reasons. First, it is expected

the elevated temperature will eliminate more water and ethanol trapped in the film during

deposition or formed during the hydrolysis. Based on the study of curing temperature effect on γ-

glycidoxypropyltrimethoxysilane (γ-GPS), Abel et al.18 also proposed that elevated temperature removes the solvents (water/methanol) and results in film shrinkage. They observed thickness varying from above monolayer to sub-monolayer with increasing cure temperature.

Based on previous studies, bis-sulfur silane film is not fully condensed at the curing

temperature of 80 °C. There are more than 11% uncondensed groups per bis-sulfur molecule in the as-prepared film, including unhydrolyzed ethoxysilyl groups –Si-O-C2H5 and uncondensed silanol group –Si-OH. Therefore, further condensation of –Si-OH to form –Si-O-Si- is expected at higher curing temperature leading to higher crosslink density. This process leads to the elevated SLD and shrinkage of the film. Abel et al.18 also reported an increase of crosslink density of γ-GPS film with increased cure temperature, based on the decrease in intensity of the various fragments in the TOF-SIMS spectrum. But they also stated that degradation of the samples or decrease of the film thickness may lead to the same phenomenon. Bertelsen96 showed that an increase in temperature produces an increase of Si-O-Si formation in a γ-GPS film formed on aluminum, indicative of an increase in crosslinking density. He also showed that drying at 180 °C results in formation of carbonyl due to the oxidation and/or degradation of the epoxide ring.

144 At the curing temperature as high as 180 °C, bis-sulfur silane degrades due to cleavage of the –

S-S- bond. Bis-sulfur silane, generally called Si-69 or TESPT in tire industry, is commonly used

as coupling agents in tire industry to improve the reinforcement properties of silica in rubber.97, 98

The mechanism of bis-sulfur silane reinforcement is bonding to silica through ethoxy group and

reacting with rubber through sulfur moiety. Therefore, the breakage of the S-S bond has been

studied extensively in literature.

Eight-membered sulfur rings, long chain polymeric sulfur, and low molecular weight alkyl

tetrasulfide all undergo thermal hemolytic scission with an activation energy of approximately 35

kcal mole-1.99, 100 Polysulfide can undergo bond interchange reactions. Bond interchange involves hemolytic scission of the nominal tetrasulfide bonds followed by radical attack on neighboring polysulfide bonds. Based on the investigation of tetrasulfide linkages, bond interchange becomes rapid above 100 °C. Normally the silane vulcanization reaction is carried out in situ at between

150 °C and 160 °C in an internal mixer. This high temperature is required because of the steric hindrance around the silylpropyl group in bis-sulfur silane.97

Therefore, for bis-sulfur silane film cured at 180 °C, scission and recombination of sulfur-

sulfur bond occurs constantly. Since alkyl disulfides are more stable thermally than alkyl

polysulfide,99 the most likely reaction is breakdown of the tetrasulfide bridging groups to disulfide bridging groups or even monosulfide groups. Following Tobolsky et al.,99 this process is shown by reaction 1 and 2 in Figure 7-2. The sulfur atoms released from the bis-sulfur network may link to form long chain-polymeric sulfur, or even eight-membered sulfur rings (S8).

Reaction 3 and 4 in Figure 7-2 demonstrate this process.

145 (1) S4 S2

S + (2) S2 + S4 2 S4

+ S (3) S4 + S2 S2 4

(4) S (ring) 2 S4 (diradical chain) 8

The evidence for such reactions is that thick bis-sulfur silane films applied on Al panels turn yellow after a 180-°C cure. But for our thin film on silicon wafer, the color change can not be observed either because the film is too thin or because there is no sufficient contrast between silane film and silicon wafer.

Apparently, these reactions lead to network with higher crosslinking density and therefore better performance is expected.

Compared to other silanes, the significant result for bis-sulfur silane cured at 180 °C compared to 80 °C is the substantial reduction of film thickness seen by X-ray reflectivity. As shown in

Figure 7-3, the film cured at 180 °C has a much wider fringe space. The peak width ∆q of film cured at 180 °C is almost 3 times larger than film cured at 80 °C. Based on equation d = 2π / ∆q , the thickness of film cured at 180 °C is much smaller. The decrease in thickness is confirmed by

146 ellipsometry as shown later. This shrinkage reflects the chemical reactions that occur during

the180-°C cure.

0 10 Bis-sulfur as-prepared -1 10 180 ºC Cure

-2 80 ºC Cure 10

-3 10

-4 10 Reflectivity -5 10

-6 10

2 3 4 5 6 7 8 9 2 3 4 5 6 7 0.1 -1 q (Å )

Figure 7-3: X-ray reflectivity of as-prepared bis-sulfur silane film cured at 80 °C and 180 °C. (X-ray data were taken on 1-BM beam-line at the Advanced Photon Source at ANL).

In summary, the increase of cure temperature has various effects on bis-sulfur silane films:

elimination of water and ethanol enclosed in the film, further condensation of residual silanol

group and breakdown of tetrasulfide bonds into shorter linkage. All those effects result in a

higher crosslinking density and lower thickness.

7.2.1.2 Bis-amino silane

Figure 7-4 shows the reflectivity data as well as the corresponding SLD profiles of bis-amino

silane film cured at 180 °C observed at room temperature water vapor conditioning. In contrast

to bis-sulfur silane film, substantial absorption of D2O is observed based on the increase of

147 reflectivity relative to the as-prepared film. Based on SLD profiles, the thickness of the dry film is 280 Å and the SLD is 1.03 x 10-6 Å-2. Compared to films cured at 80 °C, there is no substantial increase in the SLD as discussed above in the case of bis-sulfur silane film. The observed SLD of bis-amino film cured at 80 °C was 1.05 x 10-6 Å-2 and the thickness was 350 Å. Therefore, the

SLD of film cured at 180 °C is very close to film cured at 80 °C. The calculated volume fraction of D O in the swollen film is ϕ = 37%, which is also very close to the film cured at 80 °C 2 D2O

(ϕ = 41%). We can confirm the conclusion from the previous study that bis-amino silane is D2O fully hydrolyzed and condensed at the curing temperature of 80 °C. Further increasing in temperature does not affect the bulk structure of the film or the amount of water absorbed. The dehydration reaction as discussed above, which removes the water/methanol enclosed in the films, is also expected in the bis-amino silane film at the curing temperature of 180 °C.

Therefore, decease in thickness is also observed in bis-amino silane films, although not as significant as for the bis-sulfur silane film.

0 10

-1 Bis-amino 180 °C Cure 10 As-prepared -2 10 D2O 25 °C

-3 10

-4 10

-5

Reflectiviy 10

-6 10

-7 10

0.05 0.10 0.15 0.20 -1 q (Å )

(a)

148 6 Bis-amino 180 °C Cure 5 As-Prepared )

-2 D2O 25 °C 4 D (Å

L 3 x S

6 ϕ 2 D2O = 37% 10 1

0 0 100 200 300 400

Distance from Si (Å)

(b)

Figure 7-4 (a): Neutron reflectivity data from bis-amino silane film cured at 180 °C as-prepared (○), after exposure to D2O vapor at room temperature (+). The curves through the data correspond to best fits using model SLD profiles in (b). (b): Best-fit SLD profiles corresponding to the curves through the data in (a) for the samples as-prepared (—), after exposure to D2O vapor at room temperature (····). (Data were taken on the SPEAR at LANL).

7.2.1.3 Mixed silane

The reflectivity data as well as the corresponding SLD profiles of mixed silane under water

conditioning is shown in Figure 7-5. For the as-prepared state, the film thickness is around 200 Å

with a diffuse interface, lower than the thickness (240 Å) of mixed silane film cured at 80 °C.

The SLD of the dry film is 0.99 x 10-6 Å-2, much higher than the film cured at 80 °C (0.66 x 10-6

-2 Å ). The calculated volume fraction of D2O in the swollen film is 6.6 %, significant decrease compared to the amount of water absorbed for films cured at 80 °C (13%). These values lead to the same conclusion as obtained from bis-sulfur silane film that a denser film is formed when cured at 180 °C. Thus less water is absorbed. There is a large amount of bis-sulfur silane in the

149 mixed silane film. The condensation, dehydration as well as degradation reactions discussed above for bis-sulfur are also present in the mixed silane film. Therefore, similar property changes are observed in mixed silane.

0 10 Mixed silane 180 °C cure -1 As-prepared 10 D2O 25 ºC

-2 vity

i 10

-3 10 Reflect

-4 10

-5 10 -3 20 40 60 80x10 -1 q (Å )

(a)

Mixed silane 180 ºC Cure 4 As-prepared ) D O 25 ºC -2 2 3 (Å D L ϕ 2 D2O = 6.6% x S 6 10 1

0 0 100 200 300 Distance from Si (Å)

(b)

Figure 7-5 (a): Neutron reflectivity data from mixed silane film cured at 180 °C as-prepared (○), after exposure to D2O vapor at room temperature (+). The curves through the data correspond to best fits using

150 model SLD profiles in (b). (b) Best-fit SLD profiles corresponding to the curves through the data in (a) for the samples as-prepared (—), after exposure to D2O vapor at room temperature (····). (Data were taken on the POSYII at ANL).

7.2.2 Ellipsometry

Thickness of the as-prepared silanes cured at different temperature tested by ellipsometry is shown in Table 7-1. Ellipsometry also shows a dramatic decrease in thickness for bis-sulfur silane film cured at 180 °C (112 Å) compared to 80 °C (226 Å). For bis-amino silane and mixed silane film, the change in thickness is smaller. This result further confirms the conclusions obtained by neutron and X-ray reflectivity.

Table 7-1: Thickness of the as-prepared silanes cured at 80 °C and 180 °C measured by ellipsometry

Thickness (Å)

80 °C 180 °C

Bis-amino silane 274 270

Bis-sulfur silane 227 113

Mixed silane 258 240

7.2.3 Contact Angle

Contact angle tests show an interesting result (Table 7-2). When cured at 80 °C, contact angle does track corrosion performance in that the mixed silane has the highest contact angle, therefore the lowest surface energy. So mixed silane is more hydrophobic than each individual silane. It is not surprising that films cured at 180 °C have higher contact angle than films cured at 80 °C.

Surprisingly, however, after higher temperature treatment, the difference in contact angle

151 between different dry silanes almost disappears. Bis-amino silane, whose bulk film properties did

not change with increasing temperature based on neutron reflectivity, shows an increased contact

angle, approaching bis-sulfur silane when cured at 180 °C. The dramatic increase of contact

angle for bis-amino silane films may be attributed to the oxidation of the amino groups.101 From

SIMS spectra, Zhang et al.101 concluded that the amino group in γ-APS films deposited on a Fe surface was partially oxidized to an amide group after the film was aged at 110 °C for various time. Higher curing temperature or longer curing time might also involve molecular rearrangement in the surface and modified surface structure, leading to the minimization of the influence of the bridging group.

Table 7-2: Contact angles of 3 silanes cured at different curing temperature.

Contact Angle

80 °C 180 °C

Bis-amino silane 62° 87°

Bis-sulfur silane 73° 86°

Mixed silane 81° 90°

7.3 Conclusions

For bis-sulfur and mixed silanes cured at 180 °C, the SLD is higher and thickness is smaller

compared to an 80 °C. Therefore a denser film is formed. For both silanes, substantially less

water is absorbed when cured at 180 °C. The possible reactions in the bis-sulfur silane films that

lead to these observations are: elimination of water and ethanol enclosed in the film, further

152 condensation of residual silanol group and breakdown of polysulfide linkages to di and mono sulfur linkages.

By contrast, the SLD of bis-amino silane films cured at 180 °C is very close to films cured at

80 °C. There is also no significant decrease in water volume fraction absorbed as observed in the case of bis-sulfur silane and mixed silane films. A decrease in thickness is observed due to the dehydration reaction at the curing temperature above 100 °C.

Contact angle measurements show that higher curing temperature also has the effect of modifying surface structure of different silanes, leading to the minimization of the influence of the bridging group.

Based on the study on 180 °C cure, we confirm the conclusion that bis-amino silane is fully hydrolyzed and condensed at the curing temperature of 80 °C. Further increasing in temperature does affect the bulk structure of the film. For bis-sulfur and mixed silane film, however, higher curing temperature accelerates the hydrolysis and condensation, leading to a denser film and better water-barrier performance.

153 Chapter 8. Morphology and Water Barrier Properties of

Silane Films: The Effect of Thickness

Summary: In this chapter, films with thicknesses approaching practical levels are investigated.

Although some characteristics of thick films are similar to thin films regarding to water barrier

properties, there are notable exceptions. Bis-sulfur silane film, for example, provides an adequate

water barrier only for films cured at 180 °C with thicknesses exceeding 1200 Å. For bis-sulfur

silane films, therefore, both larger thicknesses and higher cure temperatures are expected to

improve anti-corrosion performance.

8.1 Introduction

Previously, a comprehensive study has been completed on thin films spun coated from 1%

concentration solution.91, 92 At this concentration, the films are less than 1000 Å and fall nicely within the optimum range for neutron reflectivity. The results on thin films are interesting, but may not be representative of films used in practical application. We found that water completely penetrates and swells 500 Å silane films that are known to be effective anticorrosion agents when applied in thick films (1-5 µm).10, 12, 29 To sort out these issues, we need to understand the

interaction of water with thick silane films, whose thickness is similar to that actually used in

metal coating.

Ellipsomeric measurements showed a linear relationship between the thickness of the silane

films and the concentration of the silane solutions.102 Performance tests show that the minimum

thickness of the silane film for corrosion protection of bare metals is 2500 Å.87 The thickness

154 effect was also demonstrated by Kent et al. by neutron reflectivity.77 The concentration of silane solutions must be high enough to produce films with thickness greater than 2500 Å. But, on the other hand, the film needs to be thin enough to resist crack formation.

The present study was initiated to elucidate the effect of film thickness. Both higher concentration and lower spin speed are used to obtain thicker films. Three silane systems were investigated following the same procedures as was used for thin films. Each film was tested as- prepared and again after exposure to water vapor at room temperature. Thick films applied on

Al substrate or cured at higher curing temperature are reported as well.

To study thick films, we do NR from the wafer-side rather than the air-side. In this scheme, neutrons will penetrate the 7-mm wafer and reflectivity is observed at the wafer-film interface.

Neutron reflection data shown in this chapter was obtained on the SPEAR at Los Alamos

National Laboratory and also on the POSY II reflectometer at Argonne National Laboratory.

8.2 Results and Discussion

8.2.1 Thickness test by ellipsometry

Thickness of film is controlled both by concentration and spin speed. Therefore, to obtain

thicker films, both higher concentration and lower spin speed are used. Spectroscopic

Ellipsometry (SE) was used to characterize film thickness. Ellipsometry measures two values,

Psi (ψ) and Delta (∆), as a function of wavelength and angle of incidence.80-82 Figure 8-1 shows the ψ and ∆ spectra of 1% bis-amino silane film as a representative of 1%-thin samples. Figure

8-2 shows the ellipsometry spectra of the 5%-thick samples. For the concentrated silane solution,

155 the ψ and ∆ spectra exhibit more oscillations that are commonly attributed to the thickening of the silane films.

45 1% bis-amino silane on Si 40 Exp., 60º Exp., 65º 35 Exp., 70º Exp., 75º 30 Model fit )

Ψ (° 25

10 4000 6000 8000 10000 Wavelength (Å)

(a)

160

140

120 1% bis-amino on Si Exp., 60º ∆ (°) Exp., 65º 100 Exp., 70º Exp., 75º Model fit 80

4000 6000 8000 10000 Wavelength (Å)

(b)

Figure 8-1: The spectroscopic ellipsometry ψ (a) and ∆ (b) spectra taken at 60°, 65°, 70° and 75° for 1% bis-amino silane films on silicon wafer cured at 80 °C.

156 5% Bis-amino 80 Exp., 60º Exp., 65º Exp., 70º 60 Exp., 75º Model fit

Ψ (°) 40

0 4000 6000 8000 10000 Wavelength (Å)

(a)

200 5% Bis-amino Exp., 60º Exp., 65º Exp., 70º 150 Exp., 75º Model fit ) ° ( 100 ∆

0 4000 6000 8000 10000 Wavelength (Å)

(b)

Figure 8-2: The spectroscopic ellipsometry ψ (a) and ∆ (b) spectra taken at 60°, 65°, 70° and 75° for 5% bis-amino silane films on silicon wafer cured at 80 °C.

To extract more information from the SE spectra regarding the thickness, refractive index and non-uniformity of the different silane films, an optical model of the presumed surface structure

157 has to be built. The interpretation of the SE data is then carried out by fitting the calculated

response ψ and ∆ of the model to the experimental data by using simulation and non-linear least

squared regression.26-28, 80-82 Table 8-1 shows the fitting results from ellipsometry result of silane films prepared at different concentration and spin speed.

Table 8-1: Ellipsometry results of silanes prepared at different concentration and spin speed.

Preparation Silanes Thickness (Å) Non-uniformity (%) Refractive Parameters Index 1% solution, bis-amino 273.78 0 1.46-1.54 2000 rpm bis-sulfur 226.85 0 1.48-1.62 mixture 241.58 0 1.46-1.58 5% solution, bis-amino 1569.7 2 N/A 2000 rpm Bis-sulfur 1133.9 24 N/A mixture 1422.9 0 N/A 5% solution, bis-amino 1934.5 1 1.52-1.58 1000 rpm bis-sulfur 1638.7 6 1.55-1.64 mixture 1681.0 0 1.53-1.62

Based on this table, the thickness of silane films increases to above 1000 Å with the

concentration increased to 5%. Lowering the spin speed to 1000 rpm leads to an increase of

several hundred Å in thickness. For any preparation condition, the thickness of bis-amino silane

film is larger than bis-sulfur silane, and the thickness of mixed silane is in between. This

conclusion has also been reached by neutron and X-ray reflectivity for the case of thin films.

Based on the value of non-uniformity, bis-amino silane film is more uniform than bis-sulfur

silane, and the mixed silane is more uniform than each individual silane film. This conclusion is

consistent with performance test result that mixed silane is superior to each individual silane.

Improved coverage of substrate by the mixed silane solution has been proposed to explain the

158 superior performance of the mixed system.14 The refractive index of the silane film varies with

wavelength. The refractive index of bis-sulfur silane is larger than bis-amino silane films at the

same condition. When thickness increases, the refractive index also increases. As an example,

Figure 8-3 shows the refractive index of mixed silane film at different concentration as a

function of wavelength.

1.70 Mixed silane film 1.65 n 1%, 2000 rpm x, 5%, 1000 rpm e 1.60 d n

e I 1.55 v

1.50

Refracti 1.45

1.40 3 3 4 5 6 7 8 9 10x10

Wavelength (Å)

Figure 8-3: Wavelength dependence of the refractive index of mixed silane at different thickness.

8.2.2 Water Conditioning

In this section, neutron reflectivity of thick silane films followed D2O vapor conditioning at room temperature will be discussed. Following the same procedure as previously used on thin films, the baseline systems cured at 80 °C on Si wafer as substrate were investigated. Mixed

silane on Al coated Si was chosen to study the substrate effect in the case of thick films. Finally

thick bis-sulfur and bis-amino silane films cured at 180 °C were characterized.

159 8.2.2.1 Si Wafer Substrate and 80 °C Curing Temperature

0 10 5% Bis-sulfur -1 10 As-prepared

D2O 25 °C

y -2 10 it iv t -3 c 10 e l f

e -4 10 R

-5 10

-6 10 8 9 2 3 4 5 6 7 8 9 0.01 0.1 -1 q (Å )

(a)

3.5 5% Bis-sulfur 3.0 As-prepared

) D O 25 °C 2.5 2 -2

2.0 D (Å ϕ L D2O = 11% 1.5 x S

6 1.0 10 0.5

0.0 0 500 1000 1500

Distance from Si (Å)

(b)

Figure 8-4 (a): Neutron reflectivity data from 5% bis-sulfur silane film as-prepared (○), after exposure to

D2O vapor at room temperature. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). The calculated volume fraction of D2O in the swollen film is 11%. (Data were from SPEAR at LANL).

160

Reflectivity data for films of 5% bis-sulfur silane as-prepared and after exposure to D2O vapor

at room temperature are shown in Figure 8-4a. Compared to thin films deposited at 1%

concentration, the difference is a significant increase in intensity of conditioned film compared to

dry film, indicating substantial absorption of water. The best-fit SLD profiles corresponding to

the lines through the data in Figure 8-4a are shown in Figure 8-4b. For the as-prepared state, the

film thickness is 1391 Å. The SLD of the dry film is 0.56 x 10-6 Å-2, lower than that of thin films

-6 -2 (0.61 x 10 Å ). The calculated volume fraction of D2O in the swollen film is 11%, much higher than the amount of 7.8% absorbed in thin films. The lower SLD of dry film and the increased water absorption in the swollen film indicate that the thick film deposited at 5% is less dense

(greater free volume) than 1% film cured at the same temperature. For bis-1,2-

(triethoxysilyl)ethane (BTSE) films on Al, Franquet et al. 26, 27 also found that films became more porous and brittle as the thickness increases.

0 10 5% Bis-amino_Air-side

-1 As-prepared 10 D2O 25 ºC

-2 vity

i 10

-3 10 Reflect

-4 10

7 8 9 2 3 4 5 6 7 8 9 0.01 -1 q (Å )

Figure 8-5 (a): Neutron reflectivity data from 5% bis-amino silane film as-prepared, after exposure to

D2O vapor at room temperature with beam from air side. (Data were from POSYII at ANL).

161 Figure 8-5 shows the reflectivity curve of 5% bis-amino silane film as-prepared and after

exposure to water vapor at room temperature with beam impinging from air side. Based on the

increase in reflectivity the film swells considerably. The thickness is so large that the peaks are

too close to be differentiated. Based on the resolution limit, the film must be more than 2000 Å.

In Figure 8-6, for the conditioned sample, beam was directed from wafer side instead of air side.

In this case no critical edge is observed, since the beam traveled from high SLD bulk (Si) to low

SLD bulk (air). The advantage is that since there is no critical edge, fringes can be observed at

very low q. More information can be obtained for thick film with rough surface in this low q

region. The SLD profile shows that the thickness of the as-prepared film is 1877 Å, consistent

with the value obtained from ellipsometry. And the SLD is 0.8 x 10-6 Å-2, much lower than that of the thin films. The thickness of the swollen film is 2000 Å. The SLD of the swollen film, surprisingly, is 2.3 x 10-6 Å-2, lower than expected based on the value of thin films. The

calculated water volume fraction is only 27%. Higher water volume fraction is expected based

on the lower SLD of as-prepared film compared to thin films and the hydrophilic nature of bis-

amino silane films. Further work is needed to understand this result.

0 10 -1 10 -2 10

ty -3 10 vi -4 10 ecti -5 10

Refl -6 10 5% Bis-amino_wafer-side -7 As-prepared 10 D2O 25 °C -8 10 7 8 9 2 3 4 5 6 7 8 9 2 0.01 0.1 -1 q (Å )

162 (a)

5% Bis-amino_wafer-side 4 As-prepared )

-2 D2O 25 °C

(Å 3 D L 2 x S

6 ϕ D2O = 27%

10 1

0 0 500 1000 1500 2000 Distance from Si (Å)

(b)

Figure 8-6 (a): Neutron reflectivity data from 5% bis-amino silane film as-prepared (○), after exposure to

D2O vapor at room temperature with beam shooting from wafer side. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves

through the data in (a). The calculated volume fraction of D2O in the swollen film is 27%. (Data were from SPEAR at LANL).

Vapor conditioning results for 5% mixed silane is shown in Figure 8-7. The thickness of the

as-prepared film is 1500 Å and the SLD is 0.76 x 10-6 Å-2. Those two parameters are in between bis-amino silane film and bis-sulfur silane films, as in the case of thin films. The thickness of

film after exposure to vapor is 1490 Å, smaller than the as-prepared state even through 12%

water is absorbed. A similar phenomenon is observed for bis-sufur silane films. Such shrinkages

are also observed in the case of thin films and we proposed that it is due to the further hydrolysis

and condensation of bis-sulfur silane film. Bis-amino silane, on the other hand, is highly

hydrolyzed and condensed at the cure temperature of 180 °C. Therefore, a swelling instead of

shrinkage is observed when exposed to water vapor.

163 0 10

-1 5% Mixed silane 10 As-prepared

-2 D2O 25 °C 10 vity i -3 10

-4 10 Reflect

-5 10

-6 10 8 9 2 3 4 5 6 7 8 9 2 3 0.01 0.1 -1 q (Å )

(a)

3.0 5% Mixed silane 2.5 As-prepared ) D O 25 °C

-2 2 2.0 D (Å

L 1.5

x S ϕ 6 1.0 D2O = 12% 10 0.5

0.0 0 500 1000 1500 Distance from Si (Å)

(b)

Figure 8-7 (a): Neutron reflectivity data from 5% mixed silane film as-prepared (○), after exposure to

164 8.2.2.2 Al substrates

Reflectivity data for 5% mixed silane on Al substrate as-prepared and after exposure to D2O vapor at room temperature are shown in Figure 8-8a. The Al layer is around 250 Å based on our previous study using ellipsometry and neutron reflectivity. While the 5% mixed silane film is around 1500 Å as reported in section 8.2.2.1. Therefore, the fringe spacing due to the silane film is much smaller than that of Al layer based on equation d = 2π / ∆q . As seen from the reflectivity

curve, the silane peaks are superimposed on much wider peaks due to Al layer. In addition,

Silane peaks are only observed at small q. The best-fit SLD profiles are shown in Figure 8-8b.

Based on the study of bare Al on silicon wafer, the Al2O3 is very porous; therefore, the SLD of

Al2O3 is allowed to vary during fitting. Similar to the result for the 1% sample, a water

penetration into the oxide layer is required to fit the data. The water distribution in the film as

well as the amount of water absorbed is consistent with the result for mixed silane on silicon

wafer. In both cases of thickness the hydrophilic layer between Al and silane is more pronounced

compared to that between Si and silane due to the large amount of water in the “pores” of the

aluminium oxide layer. The substrate, however, does not affect the top surface or the bulk

structure. Therefore, the conclusions obtained from thin films regarding to the effect of substrate

on the properties of silane film are also applicable to thick films.

165 0 10 5% Mixed silane on Al -1 10 As-prepared

-2 D2O 25 °C 10 ty

vi -3 10

ecti -4 10

Refl -5 10

-6 10

-7 10 8 9 2 3 4 5 6 7 8 9 0.01 0.1 -1 q (Å )

(a)

4 5% mixed silane on Al

) As-prepared -2 D2O 25 °C 3 D (Å L 2 x S 6

10 1

0 0 500 1000 1500 2000

Distance from Si (Å)

(b)

Figure 8-8a: Neutron reflectivity data from 5% mixed silane on Al substrate as-prepared, after conditioned in D2O vapor. The curves through the data correspond to best fits using model SLD

166 profiles in (b). (b) Best-fit scattering length density profiles corresponding to the curves through the data in (a). (Data were from SPEAR at LANL).

8.2.2.3 180 °C Cure

Figure 8-9 showed the vapor conditioning result of 5% bis-sulfur silane film cured at 180 °C.

The significant observation is that the reflectivity curve is almost the same as the as-prepared

film. There is almost no absorption of water. SLD profiles in Figure 10b show that the thickness

of the as-prepared film is 1230 Å and the SLD is 0.9 x 10-6 Å-2. Compared to 5% bis-sulfur silane films cured at 80 °C discussed above, the thickness is smaller and the SLD is higher when cured at 180 °C, indicating a denser film. This conclusion is consistent with observations on 1% films. Higher cure temperature induces a number of effects including elimination of water and ethanol from the film, further condensation of residual silanol groups and breakdown of tetrasulfide bonds into shorter leakage. All those effects result in a higher crosslink density and lower thickness.

0 10

-1 10

-2

y 10 it -3 iv 10 t c

e -4 l

f 10 e

R -5 10 5% Bis-sulfur 180 °C Cure -6 As-prepared 10 D2O 25 °C -7 10 7 8 9 2 3 4 5 6 7 8 9 2 0.01 0.1 -1 q (Å )

(a)

167 3.5

3.0 5% Bis-sulfur 180 ºC Cure As-prepared ) 2.5 -2 D2O 25 °C

(Å 2.0 D L 1.5 x S 6 1.0 10 0.5

0.0 0 500 1000 1500 Distance from Si (Å)

(b)

Figure 8-9 (a): Neutron reflectivity data from 5% bis-sulfur silane film cured at 180 °C as-prepared (○),

after exposure to D2O vapor at room temperature. The curves through the data correspond to best fits using model SLD profiles. (b) Best-fit SLD profiles corresponding to the curves through the data in (a). (Data were from SPEAR at LANL).

Previous study on 1% thin bis-sulfur silane film showed that films cured at 180 °C is denser

compared to films cured at 80 °C and the water volume fraction reduced from 7.8% to 2%. For

5% film, the water volume fraction is reduced to almost zero when the curing temperature is

increased to 180 °C. Therefore, we conclude that for bis-sulfur silane films, both thickness and

cure temperature are critical for its water-barrier properties. With more than 1200-Å thickness

and 180-°C curing temperature, bis-sulfur silane film can act as a water barrier layer, protecting

the substrate from water penetration.

The vapor conditioning result for 5% bis-amino silane film cured at 180 °C is shown in Figure

8-10. In contrast to bis-sulfur, substantial absorption occurs based on enhanced reflectivity of the

swollen film as well as the shift of critical angle to higher q. The decrease in peak spacing also

168 indicates the increase in film thickness. SLD profiles in Figure 8-10b shows that the thickness of the as-prepared film is 1752 Å and the SLD is 1.0 x 10-6 Å-2. The calculated water volume fraction is 41%, similar to the value obtained from 1% bis-amino silane films on both silicon wafer and Al cured and at both curing temperature. So, for bis-amino silane film, increasing thickness and cure temperature has no effect on its water barrier properties. In any preparation condition, water penetrates the film up to substrate and the films absorb about 40% water.

0 10 5% Bis-amino 180 °C Cure -1 10 As-prepared D2O 25 °C -2

ty 10 vi -3 10 ecti

-4 10 Refl

-5 10

-6 10 8 9 2 3 4 5 6 7 8 9 2 0.01 0.1 -1 q (Å )

(a)

169 5% Bis-amino180 ºC Cure 4 As-prepared D2O 25 °C )

-2 3 D (Å L 2 ϕ D2O = 41% x S 6 1 10

0 0 500 1000 1500 2000 2500

Distance from Si (Å)

(b)

Figure 8-10 (a): Neutron reflectivity data from 5% bis-amino silane film cured at 180 °C as-prepared (○),

8.3 Conclusions

Regarding to water-barrier properties, some characteristics of thick films are similar to thin films. The structure and water absorption of mixed silanes is in between bis-sulfur silane and bis-amino silane. Bis-amino films swell substantially when exposed to water vapor, whereas for mixed silane and bis-sulfur silane film, films shrink due to further hydrolysis and condensation.

Higher curing temperature leads to denser films. On Al substrate, the hydrophilic layer between substrate and silane is more pronounced compared to silicon wafer substrates due to the large amount of water in the “pores” of the Al oxide layer. Finally, the substrate does not affect the top surface or the bulk structure.

170 There are also some important effects of film thickness. Bis-sulfur silane deposited from 5% concentration solution and cured at 80 °C is more porous compared to a 1% thin film. As a result the film absorbs more water. When the same film is cured at 180 °C, however, almost no water is absorbed. Considering the fact that the 180 °C 1% thin film is not an effective water barrier, we conclude that bis-sulfur silane film provide an adequate barrier to water penetration only with more than 1200 Å thickness and 180-°C curing temperature. Therefore, for bis-sulfur silane films, both larger thickness and higher cure temperature are critical for effective water- barrier properties.

By contrast, bis amino films showed no improvement of performance with increased thickness and curing temperature. About 40% water is absorbed for bis-amino film under all preparation condition studied. The more hydrophilic nature of the bis-amino films and the fact that the films are fully condensed at 80 °C accounts for these properties.

171 Chapter 9. General Conclusions and Suggested Future

Work

9.1 General Conclusions

Based on study reported here, the following conclusions can be drawn regarding to the relationship between silane molecular structure, processing parameters, silane film morphology and water-barrier properties:

1: Bridging group is the key factor that controls the morphology and water-barrier properties of silane films. Bis-sulfur silane is not as condensed as bis-amino silane, but it swells less in water because of the hydrophobic nature of bridging group. By contrast, bis-amino film is more hydrophilic since the secondary amine group hydrogen bonds with water. Bis-amino silane films are thicker and smoother than bis-sulfur silane films prepared at the same concentration.

172 3: The improved anti-corrosion performance of the mixed film is traced to modification of the chemistry in both the film and the precursor solution. Based on the enhanced shrinkage that occurs following water-vapor conditioning of the mixed film, condensation is accelerated in the mixed silane. Regarding to the precursor solution, bis-amino silane may act as a catalyst in the hydrolysis of bis-sulfur silane leading to more silanols groups in the solution, which in turn improves the wettability of the solution. Therefore, accelerating the hydrolysis of residual unhydrolyzed group of hydrophobic silane is the key to improving wettability and anti-corrosion performance.

Although the redried films show enhanced shrinkage in the mixed film, the vapor-swollen state does not show any unusual character. The mixed silane film swells to an extent that is only slightly less than that of both components weighted by their volume fraction. Thickness and roughness of mixed silane in the swollen state is also between that of bis-amino silane and bis- sulfur silane. Contact angle, however, does track corrosion performance in that the mixed silane has the highest contact angle. Presumably the chemistry discussed above leads to an air surface that favors hydrophobicity in the mixed film.

4: The substrate does not play the key role for the water barrier properties of silanes. The films deposited on Al substrate and silicon wafer have similar bulk properties and top surface morphology. We conclude that a 200-Å silane film is thick enough that the substrate does not affect the top surface or the bulk structure.

5: Higher curing temperature leads to denser film for bis-sulfur and mixed silane film, whereas the temperature effect on bis-amino silane film is negligible. This result confirms that bis-amino silane is highly hydrolyzed and condensed at a curing temperature of 80 °C. Further increasing in

173 cure temperature does not affect the bulk structure of the film. Based on contact angle test,

however, higher curing temperature does modify the surface structure, leading to the

minimization of the influence of the bridging group.

The possible reactions in bis-sulfur silane films cured at 180 °C that lead to denser films are:

elimination of water and ethanol retained in the film, further condensation of residual silanol

group and breakdown of polysulfide bonds to shorter linkage.

6: Film thickness is an important variable that controls water-barrier performance: Bis-sulfur

silane deposited at 5% concentration and cured at 80 °C is more porous compared to the thinner

film deposited at 1% concentration. As a result, the thicker film absorbs more water. Bis-sulfur

silane provides an adequate barrier to water penetration only for films cured at 80 °C whose

thickness exceeds 1200 Å. Therefore, for bis-sulfur silane films, both larger thickness and

higher cure temperature are critical for effective water-barrier properties. By contrast, bis-amino

films show no performance improvement with increased thickness and curing temperature.

9.2 Suggested Future Work

The following areas are suggested future work for silane studies:

For almost all silane films studied in this work, NR finds a hydrophilic layer between substrate and silane. But because of the resolution limit of neutron reflectivity, as well as the fact that no direct information of a chemical nature is obtained with neutron reflectivity, further study on the

interface is required using other powerful characterization tools to clarify the composition and

morphology of the interface.

174 Previous studies of the hydrothermal response of silane coatings involved exposure to water

vapor at 100% humidity environment. But, in real application, various conditions, including

exposure to liquid water or even hot water or different humidity level, are expected for silane

coating. Therefore, the response of these films in direct contact with liquid water (D2O) at both

room temperature and 80 °C needs to be studied. Long-term water exposure also needs to be

performed to monitor the degradation of silane films. The response of film under normal

humidity level (50%) is of interest in realistic application. Therefore, the humidity level in the conditioning holder needs to be controlled for further study.

Water penetration is part of the reasons of corrosion. The corrosion of Al by pitting is due to

Cl- attack. The metal won’t corrode in pure water without Cl- and O2. So, the barrier properties of silane films to Cl- need to be investigated. The experimental scheme needs to be carefully designed so that the penetration of Cl- into silane film can be monitored by neutron reflectivity.

Finally, the goal of the SERDP project is to integrate organosilane with resin into a full water-

borne coating system. Therefore, next step is to study polymer-toughened silane system. Of

interest is the interaction between polymer and silane, the effect of polymer on the hydrolysis

and condensation of silane as well as the morphology of the polymer silane system. The

interaction of water with this integrated system is still an important issue. The same

hydrothermal conditioning and solvent swelling scheme can be applied to the polymer toughened

silane system. The foremost objective is to prepare high quality polymer-toughened silane films

suitable for neutron reflectivity.

175

References

[1] Van Ooij, W. J.; Child, T. Chemtech 1998, 28 (2), 26-35.

[2] Van Schaftinghen, T.; Le Pen, C.; Terryn, H.; Horzenberger, F. Electrochimica Acta

2004, 49 (17-18), 2997-3004.

[3] Subramanian, V.; Van Ooij, W. J. Surf. Eng. 1999, 15 (2), 168-172.

[4] Puomi, P.; Fagerholm, H. M. J. Adhesion Sci. Technol. 2001, 15 (5), 509-533.

[5] Van Ooij, W. J.; Zhu, D. Corrosion 2001, 57 (5), 413-427.

[6] Susac, D.; Sun, X.; Mitchell, K. A. R. Appli. Surface Sci. 2003, 207 (1-4), 40-50.

[7] Sundararajan, G. P., M.S. Thesis, Department of Materials Science and Engineering,

University of Cincinnati, Cincinnati, 2000.

[8] Plueddemann, E. P., Silane Coupling Agents. 2nd ed.; Plenum Press: New York, 1991.

[9] Yim, H.; Kent, M. S.; Hall, J. S. J. Phys. Chem. B 2002, 106 (10), 2474-2481.

[10] Zhu, D.; Van Ooij, W. J. J. Adhesion Sci. Technol. 2002, 16 (9), 1235-1260.

[11] Jayaseelan, S. K.; Van Ooij, W. J. J. Adhesion Sci. Technol. 2001, 15 (8), 967-991.

[12] Zhu, D. Q.; Van Ooij, W. J. Electrochimica Acta 2004, 49 (7), 1113-1125.

[13] Yuan, W.; Van Ooij, W. J. Journal of Colloid and Interface Science 1997, 185 (1), 197-

209.

[14] Zhu, D., Ph.D dissertation, Department of Materials Science and Engineering, University

of Cincinnati, Cincinnati, 2003.

[15] Song, J.; Van Ooij, W. J. Journal of Adhesion Science and Technology 2003, 17 (16),

2191-2221.

176 [16] Bexell, U.; Olsson, M. Surface Interface Anal. 2001, 31 (3), 223-231.

[17] Bexell, U.; Olsson, M. Surface and Interface Analysis 2003, 35 (11), 880-887.

[18] Abel, M. L.; Rattana, A.; Watts, J. F. Journal of Adhesion 2000, 73 (2-3), 313-340.

[19] Abel, M. L.; Watts, J. F.; Digby, R. P. J. Adhesion 2004, 80, 291-312.

[20] Abel, M. L.; Digby, R. P.; Fletcher, I. W.; Watts, J. F. Surface and Interface Analysis

2000, 29 (2), 115-125.

[21] Johnsen, B. B.; Olafsen, K.; Stori, A. International Journal of Adhesion and Adhesives

2003, 23 (2), 157-165.

[22] Bertelsen, C. M.; Boerio, F. J. Progress in Organic Coatings 2001, 41 (4), 239-246.

[23] Turner, R. H.; Boerio, F. J. Journal of Adhesion 2002, 78 (6), 465-493.

[24] Bexell, U.; Grehk, M.; Olsson, M.; Gelius, U. Surface and Interface Analysis 2004, 36

(7), 624-631.

[25] Horner, M. R.; Boerio, F. J.; Clearfield, H. M. J. Adhesion Sci. Technol. 1992, 6, 1-22.

[26] Franquet, A.; Le Pen, C.; Terryn, H.; Vereecken, J. Electrochimica Acta 2003, 48 (9),

1245-1255.

[27] Franquet, A.; De Laet, J.; Schram, T.; Terryn, H. Thin Solid Films 2001, 384 (1), 37-45.

[28] Franquet, A.; Terryn, H.; Bertrand, P.; Vereecken, J. Surface and Interface Analysis

2002, 34 (1), 25-29.

[29] Zhu, D.; Van Ooij, W. J. Corrosion Science 2003, 45 (10), 2177-2197.

[30] Zhu, D.; Van Ooij, W. J. Progress in Organic Coatings 2004, 49 (1), 42-53.

[31] Trabelsi, W.; Dhouibi, L.; Triki, E.; Ferreira, M. G. S.; Montemor, M. F. Surface &

Coatings Technology 2005, 192 (2-3), 284-290.

177 [32] Cabral, A.; Duarte, R. G.; Montemor, M. F.; Zheludkevich, M. L.; Ferreira, M. G. S.

Corrosion Science 2005, 47 (3), 869-881.

[33] Zhu, D.; Ooij, W. J. V. In Adhesion Aspects of Polymeric Coatings, Newark, NJ, 2000.

[34] Van Ooij, W. J.; Zhu, D.; Prasad, G.; Jayaseelan, S.; Fu, Y.; Teredesai, N. Surface

Engineering 2000, 16 (5), 386-396.

[35] Subramanian, V.; van Ooij, W. J. Corrosion 1998, 54 (3), 204-215.

[36] Sundararajan, G. P.; van Ooij, W. J. Surface Engineering 2000, 16 (4), 315-320.

[37] Pohl, E. R.; Chaves, A. In Silanes and Other Coupling Agents, Mittal, K. L., Ed. VSP:

Orlando, FL, 2004; Vol. 3, p. 3-9.

[38] Osterholtz, E. D.; Pohl, E. R. J. Adhesion Sci. Technol. 1992, 6, 127-149.

[39] PU, Z.; Van Ooij, W. J.; Mark, J. E. J. Adhesion Sci. Technol. 1997, 11, 29.

[40] Franquet, A.; Biesemans, M.; Willem, R.; Terryn, H.; Vereecken, J. Journal of Adhesion

Science and Technology 2004, 18 (7), 765-778.

[41] Arkles, B.; Steinmetz, J. R.; Zazyczny, J.; Mehta, P. J. Adhesion Sci. Technol. 1992, 6,

193.

[42] Schaefer, D. W. Science 1989, 243 (4894), 1023-1027.

[43] Subramanian, V., Ph.D Dissertation, Department of Materials Science and Engineerging,

University of Cincinnati, 2000.

[44] Kent, M. S.; Yim, H. In Silanes and Other Coupling Agents, Mittal, K. L., Ed. VSP:

Utrecht, The Netherlands, 2000; Vol. 2, p. 115-125.

[45] Bexell, U.; Olsson, M. Surface Interface Anal. 2001, 31 (3), 212-222.

[46] Franquet, A.; Terryn, H.; Vereecken, J. Applied Surface Science 2003, 211 (1-4), 259-

269.

178 [47] Song, J., Ph.D Dissertation, Department of Materials Science and Engineering,

University of Cincinnati, Cincinnati, 2003.

[48] Banhegyi, G., Dielectric spectroscopy. In Characterization of Composite Materials,

Ishida, H., Ed. Butterwurth-Heinemann, Boston, 1994.

[49] Nguyen, T.; Byrd, E.; Bentz, D. Journal of Adhesion 1995, 48, 169-194.

[50] Russell, T. P. Physica B 1996, 221, 267-283.

[51] Compton, A. H. Phil. Mag. 1923, 45, 1121.

[52] Kiessig, H. Ann. Phys. 1931, 10, 769.

[53] Parratt, L. G. Phys. Rev. Lett. 1954, 95 (2), 359-369.

[54] Yoneda, Y. Phys. Rev. B 1963, 131, 2010.

[55] Croce, P.; Devant, G.; Sere, M. G.; Verhaeghe, M. F. Surf. Sci. 1970, 22, 173.

[56] Penfold, J.; Thomas, R. K. J. Phys. Condens. Matter 1990, 2, 1369-1412.

[57] Russell, T. P. Materials Science & Engineering. R, Reports: a review journal 1990, 5,

171.

[58] Russell, T. P. Annu. Rev. Mater. Sci. 1991, 21, 249-268.

[59] Satija, S. K.; Majkrzak, C. F.; Russell, T. P.; Sinha, S. K.; Sirota, E. B.; Hughes, G. J.

Macromolecules 1990, 23 (16), 3860-3864.

[60] Penfold, J.; Lee, E. M.; Thomas, R. K. Molecular Physics 1989, 68, 33-47.

[61] Born, M.; Wolf, E., Principle of Optics. 6th ed. ed.; Pergamon Press, New York: 1980.

[62] Roe, R. J., Methods of X-ray and Neutron Scattering in Polymer Science. Oxford

University Press: New York, 2000.

[63] Foster, M. Critical reviews in analytical chemistry 1993, 24, 179.

179 [64] Lovell, M. R.; Richardson, R. M. Current Opinion in Colloid & Interface Science 1999,

4, 197-204.

[65] Zhou, X. L.; Chen, H. Phys Rep 1995, 257, 223-348.

[66] Sivia, D. S.; Hamilton, W. A.; Smith, G. S. Physica B 1991, 173 (1-2), 121-138.

[67] Lekner, J., Theory of Reflectionn. Martinus Nijhoff Publishers, Dordrecht: 1987.

[68] Daillant, J.; Gibaud, A., X-ray and neutron reflectivity: principle and application.

Springer: Verlag Berlin Heidelberg, 1999.

[69] Névot, L.; Croce, P. Rev. Phys. Appl. 1980, 15, 761.

[70] Pynn, R.; Baker, S. Physica B 1994, 198, 1.

[71] Steyerl, A.; Malik, S. S.; Iyengar, L. R. Physica B 1991, 173 (1-2), 47-64.

[72] Penfold, J. Current Science 2000, 78 (12), 1458.

[73] Wu, W. L.; Orts, W. J.; Majkrzak, C. J.; Hunston, D. L. Polymer Engineering and

Science 1995, 35 (12), 1000-1004.

[74] Yim, H.; Kent, M.; McNamara, W. F.; Ivkov, R.; Satija, S.; Majewski, J.

Macromolecules 1999, 32 (23), 7932-7938.

[75] Kent, M. S.; McNamara, W. F.; Fein, D. B.; Domeier, L. A.; Wong, A. P. Y. J. Adhesion

1999, 69 (1-2), 121-138.

[76] Kent, M. S.; McNamara, W. F.; Fein, D. B.; Domeier, L. A.; Wong, A. P. Y. J. Adhesion

1999, 69 (1-2), 139-163.

[77] Kent, M. S.; Smith, G. S.; Baker, S. M.; Nyitray, A.; Browning, J.; Moore, G.; Hua, D.

W. Journal of Materials Science 1996, 31 (4), 927-937.

[78] Lang, J. C.; Srajer, G.; Wang, J.; Lee, P. L. Review of Scientific Instruments 1999, 70

(12), 4457-4462.

180 [79] Schaefer, D. W.; Beaucage, G. B.; Loy, D. A.; Ulibarri, T. A.; Black, E.; Shea, K. J.;

Buss, R. J. In Mater. Res. Soc. Symp. Proc., Materials Research Society: Pittsburgh, PA,

1996; Vol. 435, p. 301.

[80] Woollam, J. A.; Hilfiker, J. N.; Bungay, C. L.; Synowicki, R. A.; Tiwald, T. E.; W., T. D.

In Characterization and Metrology for ULSI technology: 2000 International Conference,

Seiler, D. G.; Diebold, A. C.; Shaffner, T. J.; McDonald, R.; Bullis, W. M.; Smith, P. J.;

Secula, E. M., Eds. American Institute of Physics: 2000; p. 511.

[81] Woollam, J. A.; Johs, B.; Herzinger, C. M.; Hilfiker, J.; Synowicki, R.; Bungay, C. L. In

SPIE proc., SPIE: Denver, Colorado, 2000; Vol. CR72, p. 29.

[82] Woollam, J. A.; Johs, B.; Herzinger, C. M.; Hilfiker, J.; Synowicki, R.; Bungay, C. L. In

SPIE proc., SPIE: Denver, Colorado, 2000; Vol. CR72, p. 3.

[83] Ward, L., The Optical Constants of Bulk Materials and Films. Adam Hilger: Bristol and

Philadelphia, 1988.

[84] Palik, E. D., Handbook of Optical Constants of Solids. Academic Press, Inc.: Orlando,

Florida, 1985.

[85] J. A. Woollam Co., Wollam Ellipsometry Tutorial. http://www.jawoollam.com/tutorial,

February 4th, 2004.

[86] Wang, D.; Jones, F. R. J. Mater. Sci. 1993, 28 (1), 36.

[87] Premachandra, J. K.; Van Ooij, W. J.; Mark, J. E. Journal of Adhesion Science and

Technology 1998, 12 (12), 1361-1376.

[88] Van Ooij, W. J.; lamar, A.; Zhu, D. In triservice corrosion conference, San Antonio, TX,

2002.

[89] Flory, P. J., Principles of Polymer Chemistry. Cornell University Press: Ithaca, NY, 1953.

181 [90] Fowkes, F. M.; Dwight, D. W.; Cole, D. A. J. Non-Cryst. Solids 1990, 120 (1-3), 47-60.

[91] Pan, G.; Yim, H.; Kent, M. S.; Majewski, J.; Schaefer, D. W. Journal of Adhesion

Science and Technology 2003, 17 (16), 2175-2189.

[92] Pan, G.; Yim, H.; Kent, M.; Majewski, J.; Schaefer, D. W. In Proceedings of the Fourth

International Symposium on Silanes and Other Coupling Agents, VSP: Orlando, FL,

2004; Vol. 3, p. 39-49.

[93] Schewchun, J.; Rowe, E. C. J. Appl. Phys. 1970, 41, 4128.

[94] Hu, M. H.; Noda, S.; Okubo, T.; Yamaguchi, Y.; Komiyama, H. Applied Surface Science

2001, 181 (3-4), 307-316.

[95] Jones, D. A., Principles and Prevention of Corrosion. Second ed.; Prentice Hall: Upper

Saddle River, NJ, 1995.

[96] Bertelsen, C. M., Ph.D Dissertation, Materials Science and Engineering, University of

Cincinnati, Cincinnati, 1997.

[97] Mark, J. E.; Erman, B.; Eirich, F. R., The Science and Technology of Rubber. third ed.;

Elsevier Academic Press: 2005.

[98] Dierkes, W. K.; Reuvekamp, L.; Ten Brinke, A. In Silane and other coupling agents,

VSP: Orlando, FL, 2004; Vol. 3, p. 89-103.

[99] Tobolsky, A. V.; MacKnight, W. J., Polymeric Sulfur and Related Polymers. Interscience

Publishers: 1965.

[100] Aklonis, J. J., Introduction to Polymer Viscoelasticity. John Wiley &Sons, Inc.: 1972.

[101] Zhang, C., Ph.D Dissertation, Materials Science and Engineering, University of

Cincinnati, Cincinnati, 1997.

[102] Zhu, D., Van Ooij, W.J. In Adhesion Aspects of Polymeric Coatings, Newark, NJ, 2000.

182