.uhl; - u.-

,.:.: .,.a...:.
.113; 1

335... a

 

mfi‘i‘

5755‘;

This is to certify that the

thesis entitled

THE HYDROTHERMAL BAROHYGROUS CORROSION
OF ALUMINA CERAMICS

presented by

Bejeir Delayne Brooks

has been accepted towards fulfillment
of the requirements for

 

M. S. degree in Chemical Engineering
& Materials Science

Date W01

MSU is an Affirmative Action/Equal Opportunity Institution

 

0-7639

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN REIURN Box to remove this checkout from your record.
TO AVOID FINES retum on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDm.p65-p.15

 

THE HYDROTHERMAL BAROHYGROUS CORROSION OF ALUMINA
CERAMICS

BY

Bejeir Delayne Brooks

A THESIS
Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Chemical Engineering and Materials Science

2002

Abstract

THE HYDROTHERMAL BAROHYGROUS CORROSION OF ALUMINA
CERAMICS

By
Bejeir Delayne Brooks

Hydrothermal corrosion is a process wherein both water and temperature
are simultaneously involved in the corrosion of a material. Prior to this work,
there has not been an extensive study of the hydrothermal corrosion phenomena
in alumina. For ceramics, much of the hydrothermal corrosion literature deals
with hydrothermal acidic or alkaline attack and not corrosion due to a nominally
neutral solution of deionized water. Although, hydrothermal corrosion only
occurs below the boiling or evaporation point of the liquid media, for the present
work and most other work dealing with hydrothermal corrosion of any sort
(alkaline thru acidic) the substrate to be corroded is placed within a closed
pressure vessel adding the elements of pressure (baro-) and steam (hygrous) to -
the hydrothermal corrosion environment. This thesis involves hydrothermal
barohygrous corrosion of two different grades of alumina (94% and 99.5% purity)
Alzoa. Specimens were treated over the first 50 hours of corrosion at 250°C, with
a majority of the AD-94 specimens scrutinized for <75 hours and ADS-995
specimens scrutinized from O to 50 hours. The specimens were analyzed
through atomic force microscopy with focused investigations performed through
other methods as needed (scanning electron microscopy, electron dispersive

analysis and x-ray diffraction).

Copyright
Bejeir Delayne Brooks
August 2, 2002

For Reginald, Rebecca, Telaekah and my thirteen times removed
grandparents on both sides, because without them I would not be here. To Doris

without whom I would have been content with my Bachelors degree.

ACKNOWLEDGEMENTS

I would like to acknowledge the aid and assistance given me on my thesis
from: Benjamin Simkin, Benjamin Tyszka, Boon-Chai Ng, Chee Kuang Kok, '
Craig Carrier, Dr. Darren Mason, Dr. David Grumman, Dr. Gary Cloud, Dr.
Haiping Geng, Dr. John Heckman, Dr. K.N. Subramanian. Dr. Ki-Yong Lee, Dr.
Martin Crimp, Dr. Patrick Kwon, Dr. Per Askeland, Dr. Richard Schalek, Dr.
Ronald Averill, Dr. Stanley Flegler, Dr. Thomas Bieler, Ewa Danielewicz, John
Hulbert, Jong-gi Lee, Kenneth Geelhood, Martin Traub, Raelynn Deller, Richard
Stabley, Theodore Manko, Timmothy Hoepfner, Timmothy Monroe, Wangyang
Ni. I would especially like to thank William Havens and David Enslin who wrote
the programs I used to section the AFM micrographs and of course Dr. Eldon

Case who helped make this thesis possible.

TABLE OF CONTENTS

LIST OF TABLES vi
LIST OF FIGURES viii
CHAPTER 1:

INTRODUCTION 1
1.1. Grain Boundary Attack 1
1.2. Hydrothermal corrosion of alumina ceramics [5] 2
1.3. Delayed failure of a porous alumina [8] 6
1.4. Reaction of alumina ceramics with saturated steam [9] 10
1.5. Grain Boundary Segregation under Hydrothermal Corrosion 14
1.6. Grain boundary segregation on A|203 [15] 15
1.7. Surface segregation of calcium in dense alumina exposed to steam 17

and steam +CO [10]

1.8. Grain boundary segregation in MgO-doped Al203 [16] 20

1.9. The Effects of Hydrothermal Corrosion on Non-Oxide Ceramics 22

1.10. Corrosion behavior of silicon carbide in 290°C water [20] 22

1.11. Degradation of SiC fibres in high-temperature, high-pressure 28

water [21]

1.12. Hydrothermal oxidation of Si3N4 powder [26] 34

1.13. The Effects of Hydrothermal Corrosion on Glasses 37

1.14. Hydrothermal corrosion of Iithia disilicate glass and glass- 37

ceramics [28]

1.15. Summary of the Literature 42
CHAPTER 2: 44
EXPERIMENTAL PROCEDURES

2.1. Material Properties 44

2.2. Sectioning Procedure 44

2.3. Mounting Procedure 47

2.4. Polishing Procedure 48

2.5. Procedure for the Loading and Unloading of the Digestion Bomb 49

[32]

2.6. Digestion Bomb Furnace Procedure 56

2.7. Atomic Force Microscope Examination [33] 59
CHAPTER 3:

RESULTS AND DISCUSSION 63
3.1. Hydrothermal Attack 63
3.2. Roughness Data 141
3.3. X-Ray Diffraction 147
3.4. Mass Changes 158

3.5. Sectioning of the Atomic Force Microscopy Micrographs 158

vi

3.6. Electron Dispersive Study

3.7. Scanning Electron Microscope Analysis
3.8. Surface Morphology

3.9. Summary and Conclusions

3.10. Future Work

REFERENCES
APPENDIX
APPENDIX A: Slip Systems for Alumina

APPENDIX B: Elastic Moduli of Alumina
APPENDIX C: Sectioning Data

vii

1 89
1 92
201
202
205

206

211
212
214
216

LIST OF TABLES

Table 1.2.1: Characteristics of different purities of Alumina Ceramics [5]
Table 1.3.1: Breaking strengths of Alumina bars [8]

Table 1.3.2: Composition of 65.1% dense alumina [8]

Table 1.3.3: Mean fatigue life for porous alumina bars [8]

Table 1.3.4: Values of a0 and a1 [8]

Table 1.4.1: Chemical Properties 96% alumina [9]

Table 1.4.2: Modulus of Rupture data for a 96% pure alumina [9]

Table 1.10.1: Compositions of testing solutions [20]

Table 1.11.1: Properties of the Tyranno fibers [21]

Table 2.1.1: Properties of Coors AD-94 and ADS 995 [31]

Table 3.3.1: X-Ray Diffraction Analysis of Non-Hydrothermally Attacked
Specimens [35]

Table 3.3.2: X-Ray Diffraction Analysis of 10-hour Hydrothermally Attacked
Specimens [35]

Table 3.3.3: X-Ray Diffraction Analysis of 50-hour Hydrothermally Attacked
Specimens [35]

Table 3.3.4: X-Ray Diffraction Analysis of Aluminum Mounting Slide [35]
Table 3.3.5: X-Ray Diffraction Standards for Diaspore and Boehmite [35]

Table 3.3.6: X-Ray Diffraction Standard for Wairakite (Pseudo-Cubic,
Monoclinic) [35]

Table 3.3.7: Correlation between the d-spacings of the or-Alumina
Standard and different stages of the Alumina Specimen Hydrothermal
Attack [35]

Table 3.3.8: Correlation between the d-spacings of the Boehmite Standard
and different stages of the Alumina Specimen Hydrothermal Attack [35]

Table 3.3.9: Correlation between the d-spacings of the Diaspore Standard
and different stages of the Alumina Specimen Hydrothermal Attack [35]

viii

149

150

151

152
153
154

155

156

157

Table 3.3.10: Correlation between the d-spacings of the Wairakite
Standard and different stages of the Alumina Specimen Hydrothermal
Attack [35]

Table 3.5.1: Regressions from the grain boundary etching of ADS-995
attacked at 250°C for 10 to 15 hours as a function of the depth from the
specimen surface

Table 3.5.2: Regressions from the grain boundary etching of AD-94
attacked at 250°C for 1 to 2 hours as a function of the depth from the
specimen surface

Table 3.5.3: Regressions from the grain boundary etching of ADS-995
attacked at 250°C for 10 to 15 hours as a function of the depth from the
specimen surface and the average grain size

Table 3.5.4: Regressions from the grain boundary etching of AD-94
attacked at 250°C for 1 to 2 hours as a function of the depth from the
specimen surface and the average grain size

Table 3.5.5: Regressions from the grain boundary etching of ADS-995
attacked at 250°C for 10 to 15 hours

Table 3.5.6: Regressions from the grain boundary etching of AD-94
attacked at 250°C for 1 to 2 hours

Table 3.1: Single crystal elastic constants of alumina in the trigonal system
[40]

Table 8.2: The reduced trigonal elasticity matrix representing
crystallographic classes 32, -3m and 3m, using compliance constants the
matrix is symmetric about the main diagonal [39]

Table C.1: Area present for the sectioning data of ADS-995

Table 0.2: Area present for the sectioning data of AD-94

ix

157

163

164

165

166

167

168

215

215

216
216

LIST OF FIGURES

Figure 2.2.1: lsomet Low-speed diamond saw: A) Diamond blade, B)
Rotating specimen holder, C) Lever Arm, D) U.S. Quarter

Figure 2.5.1: Side view of the Parr Digestion Bomb inner part, a) Top
View, b) Side View, 0) Bottom View: A) Stainless bomb body; B) Teflon
Cup; C) Corrosion disc (0.05 mm thick); D) Rupture disc (0.254 mm
thick); E) Teflon cover; G) U.S. quarter.

Figure 2.5.2: Bottom view of the Parr Digestion Bomb outer parts, a)
Bottom View, b) Side View, c) Top View: A) Bronze plated screw cap; B)
Stainless steel pressure plate; C) Spring; D) Nickel bottom plug; E)
Stainless steel top cap; G) U.S. quarter.

Figure 2.5.3: A) Spanner-Jack, used to tighten down the top and bottom
portions of the digestion bomb; B) U.S. quarter.

Figure 2.6.1: Machined lower brick A) 6.4 cm diameter X 1.35 cm tall
cylinder; B) 2.5 cm X 1.7 cm tall cylinder; C) 0.9cm X 3.23 cm tall
cylinder; D) U.S. quarter.

Figure 2.6.2: Bottom view of upper refractory brick A) 22.75 cm X 11.35
cm X 2.73 cm tall; B) Inner raised semi-cylinder 14.75 cm diameter X 3.7
cm tall; C) Through-hole for thermocouple D) U.S. quarter.

Figure 2.6.3: Side view of upper refractory brick A) 22.75 cm X 11.35 cm
X 2.73 cm tall; B) Inner raised semi-cylinder 14.75 cm diameter X 3.7 cm
tall; C) U.S. quarter.

Figure 2.6.4: Reverse side view of upper refractory brick A) 22.75 cm X
11.35 cm X 2.73 cm tall; B) Inner raised semi-cylinder 14.75 cm diameter
X 3.7 cm tall; C) U.S. quarter.

Figure 2.7.1: Silicon Nitride Probe Schematics [34]

Figure 2.7.2: Micrographs of the silicon nitride probe a) 800x b) 5000X
[34]

Figure 3.1.1: AFM micrograph of Coors ADS-995, with no hydrothermal
treatment, polished to a 1 um finish, 5.000 urn scan size, 4.069 H2 scan
rate, 256 samples; a) surface view - z-range: 1 uni/division, b) top view.

Figure 3.1.2: AFM micrograph of Coors AD-94, with no hydrothermal
treatment, polished to a 1 pm finish, 9.667 um scan size, 15.26 Hz scan
rate, 512 samples; a) surface view - z-range: 1 urn/division, b) top view.

46

51

52

55

57

57

58

58

62
62

65

67

Figure 3.1.3: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 30 minutes at 250°C, 10.00 pm scan size, 12.21 Hz scan
rate, 512 samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1 .4: AFM micrograph of Coors AD-94, hydrothermally attacked
for 30 minutes at 250°C, 5.000um scan size, 6.104 H2 scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1 .5: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 1 hour at 250°C, 10.00pm scan size, 16.28Hz scan rate, 512
samples; a) surface view - z-range: him/division, b) top view.

Figure 3.1 .6: AFM micrograph of Coors AD-94, hydrothermally attacked
for 1 hour at 250°C, 20.00pm scan size, 12.21 Hz scan rate, 512 samples;
a) surface view - z-range: writ/division, b) top view.

Figure 3.1.7: AFM micrograph of Coors AD-94, hydrothermally attacked
for 1.5 hours at 250°C, 20.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1um ldivision, b) top view.

Figure 3.1 .8: AFM micrograph of Coors AD-94, hydrothermally attacked
for 2 hours at 250°C, 10.00pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: Ium Idivision, b) top view.

Figure 3.1.9: AFM micrograph of Coors AD-94, hydrothermally attacked
for 2.5 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 um Idivision, b) top view.

Figure 3.1.10: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 3 hours at 250°C, 10.00 pm scan size, 9.766 Hz scan rate,
512 samples; a) surface view - z-range: 1 pm Idivision, b) top view.

Figure 3.1 .11: AFM micrograph of Coors AD-94, hydrothermally attacked
for 3 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 um /division, b) top view.

Figure 3.1 .12: AFM micrograph of Coors AD-94, hydrothermally attacked
for 3.5 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.13: AFM micrograph of Coors AD-94, hydrothermally attacked
for 4 hours at 250°C, 10.00 um scan size, 16.28 Hz scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

xi

69

71

73

75

78

80

82

85

87

90

92

Figure 3.1.14: AFM micrograph of Coors AD-94, hydrothermally attacked
for 4.5 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.15: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 5 hours at 250°C, 10.00 pm scan size, 4.883 H2 scan rate,
512 samples; a) surface view - z-range: 1 rim/division, b) top view.

Figure 3.1.16: AFM micrograph of Coors AD-94, hydrothermally attacked
for 5 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 um/division, b) top view.

Figure 3.1.17: AFM micrograph of Coors AD-94, hydrothermally attacked
for 5.5 hours at 250°C, 10.00 um scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 rim/division, b) top view.

Figure 3.1.18: AFM micrograph of Coors AD-94, hydrothermally attacked
for 6 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 uni/division, b) top view.

Figure 3.1.19: AFM micrograph of Coors AD-94, hydrothermally attacked
for 6.5 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 rim/division, b) top view.

Figure 3.1.20: AFM micrograph of Coors AD-94, hydrothermally attacked
for 7 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.21: AFM micrograph of Coors AD-94, hydrothermally attacked
for 7.5 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.22: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 10 hours at 250°C, 10.00 pm scan size, 6.104 Hz scan rate,
256 samples; a) surface view - z-range: 1 rim/division, b) top view.

Figure 3.1.23: AFM micrograph of Coors AD-94, hydrothermally attacked
for 10 hours at 250°C, 10.00 pm scan size, 9.766 H2 scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.24: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 12.5 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan
rate, 512 samples; a) surface view - z-range: 1 rim/division, b) top view.

xii

95

97

99

102

104

106

108

110

113

115

117

Figure 3.1.25: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 15 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate,
512 samples; a) surface view — z-range: 1 urn/division, A) grain ~2p.m in
size, B) grain larger than 2 pm, b) top view, c) top view without lines.

Figure 3.1.26: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 20 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate,
512 samples; a) surface view - z-range: 1 urn/division, b) top view-

Figure 3.1.27: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 25 hours at 250°C, 10.00 um scan size, 12.21 Hz scan rate,
512 samples; a) surface view - z-range: 1 uni/division, b) top view.

Figure 3.1.28: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 30 hours at 250°C, 10.00 um scan size, 16.28 Hz scan rate,
512 samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.29: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 35 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate,
512 samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.30: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 40 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate,
512 samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.31: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 45 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate,
512 samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1.32: AFM micrograph of Coors ADS-995, hydrothermally
attacked for 50 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate,
512 samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.1 .3: AFM micrograph of Coors AD-94, hydrothermally attacked
for 50 hours at 250°C, 10.00 pm scan size, 9.766 H2 scan rate, 512
samples; a) surface view - z-range: 1 urn/division, b) top view.

Figure 3.2.1: Normalized roughness analysis as a function of the
topographic depth for Coors ADS-995, with no hydrothermal treatment,
polished to a 1 pm finish.

Figure 3.2.2: Normalized roughness analysis as a function of the
topographic depth for Coors AD-94, with no hydrothermal treatment,
polished to a 1 pm finish.

xiii

120

123

125

128

130

132

134

137

139

143

143

Figure 3.2.3: Normalized roughness analysis as a function of the
topographic depth for the of Coors ADS-995, hydrothermally attacked for
5 hours at 250°C,

Figure 3.2.4: Normalized roughness analysis as a function of the
topographic depth for Coors AD-94, hydrothermally attacked for 5 hours
at 250°C.

Figure 3.2.5: Normalized roughness analysis as a function of the
topographic depth for Coors ADS-995, hydrothermally attacked for 10
hours at 250°C.

Figure 3.2.6: Normalized roughness analysis as a function of the
topographic depth for Coors AD-94, hydrothermally attacked for 10 hours
at 250°C.

Figure 3.2.7: Normalized roughness analysis as a function of the
topographic depth for Coors ADS-995, hydrothermally attacked for 50
hours at 250°C.

Figure 3.2.8: Normalized roughness analysis as a function of the
topographic depth for Coors AD-94, hydrothermally attacked for 50 hours
at 250°C.

Figure 3.5.1: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a
depth of -5.4 to —25.4 nm.

Figure 3.5.2: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a
depth of -25.4 to -45.4 nm.

Figure 3.5.3: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a
depth of —45.4 to —65.4 nm.

Figure 3.5.4: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a
depth of —65.4 to —85.4 nm.

Figure 3.5.5: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a
depth of -85.4 to —105.4 nm.

Figure 3.5.6: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 10 hours

Figure 3.5.7: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 12.5 hours

Figure 3.5.8: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 15 hours

xiv

144

144

145

145

146

146

169

169

170

170

171

171

172

172

Figure 3.5.9: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 10-12.5 hours

Figure 3.5.10: Grain boundary etching as a function of depth for ADS-
995 hydrothermally attacked at 250°C for 10-15 hours

Figure 3.5.11: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1 hour

Figure 3.5.12: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1.5 hours

Figure 3.5.13: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1.5 hours (-5 to —275 nm)

Figure 3.5.14: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 2 hours

Figure 3.5.15: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1-1.5 hours

Figure 3.5.16: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1-2 hours

Figure 3.5.17: Grain boundary etching as a function of depth and
average grain size for ADS-995 hydrothermally attacked at 250°C for 10
hours

Figure 3.5.18: Grain boundary etching as a function of depth and
average grain size for ADS-995 hydrothermally attacked at 250°C for
12.5 hours

Figure 3.5.19: Grain boundary etching as a function of depth and
average grain size for ADS-995 hydrothermally attacked at 250°C for 15
hours

Figure 3.5.20: Grain boundary etching as a function of depth and
average grain size for ADS-995 hydrothermally attacked at 250°C for 10-
12.5 hours

Figure 3.5.21: Grain boundary etching as a function of depth and
average grain size for ADS-995 hydrothermally attacked at 250°C for 10-
15 hours

Figure 3.5.2: Grain boundary etching as a function of depth and
average grain size for AD-94 hydrothermally attacked at 250°C for 1 hour

XV

173

173

174

174

175

175

176

176

177

177

178

178

179

179

Figure 3.5.23: Grain boundary etching as a function of depth and
average grain size for AD-94 hydrothermally attacked at 250°C for 1.5
hours

Figure 3.5.24: Grain boundary etching as a function of depth and
average grain size for AD-94 hydrothermally attacked at 250°C for 1.5
hours (-5 to -275 nm)

Figure 3.5.25: Grain boundary etching as a function of depth and
average grain size for AD-94 hydrothermally attacked at 250°C for 2
hours

Figure 3.5.26: Grain boundary etching as a function of depth and
average grain size for AD-94 hydrothermally attacked at 250°C for 1-1.5
hours

Figure 3.5.27: Grain boundary etching as a function of depth and
average grain size for AD-94 hydrothermally attacked at 250°C for 1-2
hours

Figure 3.5.28: Grain boundary etching as a function of depth and
average grain size for ADS-995 (10-15 hours) and AD-94 (1-2 hours)
hydrothermally attacked at 250°C

Figure 3.5.29: Grain boundary etching as a function of root time for ADS-

995 hydrothermally attacked at 250°C for 10-15 hours (-10 to —30 nm)

Figure 3.5.30: Grain boundary etching as a function of root time for ADS-

995 hydrothermally attacked at 250°C for 10-15 hours (-30 to —50 nm)

Figure 3.5.31: Grain boundary etching as a function of root time for ADS-

995 hydrothermally attacked at 250°C for 10-15 hours (-50 to -70 nm)

Figure 3.5.32: Grain boundary etching as a function of root time for ADS-

995 hydrothermally attacked at 250°C for 10-15 hours (-70 to -90 nm)

Figure 3.5.33: Grain boundary etching as a function of root time for ADS-

995 hydrothermally attacked at 250°C for 10-15 hours (~90 to -110 nm)

Figure 3.5.34: Grain boundary etching as a function of root time for ADS-

995 hydrothermally attacked at 250°C for 10-15 hours (-10 to -110 nm)

Figure 3.5.35: Grain boundary etching as a function of root time for AD-
94 hydrothermally attacked at 250°C for 1-2 hours (-5 to —25 nm)

Figure 3.5.36: Grain boundary etching as a function of root time for AD-
94 hydrothermally attacked at 250°C for 1-2 hours (-25 to —45 nm)

xvi

180

180

181

181

182

182

183

183

184

184

185

185

186

186

Figure 3.5.37: Grain boundary etching as a function of root time for AD-
94 hydrothermally attacked at 250°C for 1-2 hours (-45 to —65 nm)

Figure 3.5.38: Grain boundary etching as a function of root time for AD-
94 hydrothermally attacked at 250°C for 1-2 hours (-65 to —85 nm)

Figure 3.5.39: Grain boundary etching as a function of root time for AD-
94 hydrothermally attacked at 250°C for 1-2 hours (-85 to —105 nm)

Figure 3.5.40: Grain boundary etching as a function of root time for AD-
94 hydrothermally attacked at 250°C for 1-2 hours (-5 to 105 nm)

Figure 3.6.1: EDS analysis of Coors ADS-995 with no hydrothermal
anack.

Figure 3.6.2: EDS analysis of Coors AD-94 with no hydrothermal attack.

Figure 3.6.3: EDS analysis of Coors ADS-995 hydrothermally attacked
for 5 hours at 250°C.

Figure 3.6.4: EDS analysis of Coors AD-94 hydrothermally attacked for 5
hours at 250°C.

Figure 3.7.1: Scanning electron image of Coors ADS-995, with no
hydrothermal treatment, polished to a 1 pm finish.

Figure 3.7.2: Scanning electron image of Coors AD-94, with no
hydrothermal treatment, polished to a 1 um finish.

Figure 3.7.3: Scanning electron image of Coors ADS-995,
hydrothermally attacked for 5 hours at 250°C.

Figure 3.7.4: Scanning electron image of Coors AD-94, hydrothermally
attacked for 5 hours at 250°C.

xvii

187

187

188

188

190

190
191

191

193

195

197

199

Chapter 1: Introduction

The goals of the current research are the analysis of the first 2 days (50
hours) of alumina under hydrothermal corrosion. The aluminas of interest are
ADS-995 and AD-94 (Coors Ceramic Company: Boulder, Colorado) as the
effects of a corrosive atmosphere are believed to attack a material differently
based on the purity of the material. The reason for this study is that alumina is
often placed within a hydrothermal situation (as pump seals pumps in the nuclear
industry for instance) without investigating the effects of the hydrothermal

conditions the alumina experiences.

1.1 Grain Boundary Attack

In general, impurities segregate to the grain boundary phase of a material
[1]. This trend is more pronounced when the material has an exact chemical
composition than when the material exists as a solution [1]. An example of this is
steel, where there is a certain amount of alloying elements within the iron matrix
residing in interstitial and substitutional sites, in this case the grain boundary
phase of the iron is not necessarily high in alloying elements [1]. Conversely, for
alumina (AI203) where there is a desired chemical composition the grain
boundary phase will be high in impurities, leaving the rest of the matrix nominally
pure [2]. Additionally the grain boundary phases of ceramics are often
amorphous [3], as the grain boundary lacks any long-range order the grain

boundary will melt first and solidify last.

However, the grain boundaries are locations of grain mismatch and therefore
have a higher energy state than the grain boundary phase or the matrix material
[2]. The grain boundary phase is equivalent to a thin film [3], however the
thickness of a grain boundary may vary greatly. Under corrosive conditions, a
material is most likely to be attacked at higher energy locations first therefore the
grain boundaries are usually the first to be attacked [1].

Of importance in the current research is the resistance to chemical attack [4].
If the chemical resistance of the grains surpasses the chemical resistance of the
grain boundary phase, corrosion damage will occur within the grain boundary
phase. If the damage is not occurring at the grain boundaries but instead within

the grains then the purity may not be of as much interest.

1.2 Hydrothermal corrosion of alumina ceramics [5]

The corrosion behavior and the corresponding degradation of alumina
ceramics, by low temperature water were examined by Oda. The bulk alumina
(Table 1.2.1) was sectioned into testing coons (3 mm X 4 mm X 35 mm) with a
diamond, cutting wheel. The coons were then tested within an autoclave for up
to 10 days. The specimens were tested at a temperature of 300°C with a
corresponding pressure of 8.6 MPa.

The concentrations of ions dissolved in the water during the hydrothermal
treatment were determined by inductively coupled plasma atomic spectroscopy
(ICP-AES). X-ray diffraction was used to determine the crystalline phases

present at the surface, and scanning electron microscopy (SEM) was utilized for

surface microscopy. Room temperature fracture strengths, of the corroded and
un-corroded samples, were determined using a three-point bending test (30 mm

span, 0.5 mm/min crosshead speed).

Table 1.2.1: Characteristics of different purities of Aluminat Ceramics [5]

 

 

 

 

 

 

99% pure 99.9% pure 99.99% pure
Specimen alumina alumina alumina
Al203 (wt%) * 98.7 99.92 99.98
SI02 (wt%) * 1.17 0.04 0.01
N320 (wt%) * 0.09 0.02 0.01
M90 (wt%) * 0.01 0.01 <0.01
CaO (wt%) * 0.03 0.01 <0.01
F9203 (wt%) * <0.01 <0.01 <0.01
Density (g/cma) 3.34 3.93 3.95
Flexural Strength (MPa) * 304 490 588
7 Kyocera Corporation, Kyoto, Japan * Room temperature
* EDS Analysis
Yi ><Mi ‘ v
AW/Acalc = 2 meIian x1000; (1.2.1)
AW/A = 1000 X V X (Si concentrat ion X SiO 2 (mass) ) (1.2.2)
A x S] (mass)

The ions dissolved within the solution were assumed to arise from their
corresponding oxides within the AI203; thus, the associated weight loss (AW/A) of
the specimen was calculated from the mass of the dissolved oxides, in the
solution (Equations 1.2.1-2). Where, AW/Amauc is the calculated weight loss of
specimen, yi is the concentration of i elements in solution (ug/mL), MimjIn is the
formula weight of imjn constituents, both m and n are the number of moles of i and
j elements for imjn constituents. Furthermore, A is the geometric surface area
(cm2) of the specimen prior to corrosion, the atomic weight of i elements is
represented by Mi, and V is the volume of the solution (ml). The calculated
weight loss agreed well with observed experimental weight loss within the
experimental error.

The weight loss due to corrosion was mostly attributed to the loss of SiOz
and Na20 impurities in the Al203. In or-Al203, ceramics there is preferential
degradation of the Ca and Na in the grain boundary [6,7]. The grain boundary
contains a large degree of glass and intergranular SIOz phases. The effect of
impurity on the weight loss of these specimens was also investigated and was
found to linearly decrease as the purity level increased.

X-ray analysis before and after corrosion found that boehmite (AIOOH)
formed on the 99% pure alumina specimen after a period of 10 days. The
boehmite was never found in this particular study for either the 99.9% pure
alumina or 99.99% pure alumina specimens, or that of the 99% pure alumina for

a time span of less than 10 days.

As the time span of the hydrothermal attack increases, the thickness of
the corresponding corrosion layer also increases. As the corrosion layer thickens
the strength of the specimen decreases. The fracture strength, under 3-point
bending of the 99% pure alumina specimen, over a span of 10 days dropped to
around 50% of its original value while that of the 99.99% pure alumina specimen
decreased by about 10%. Thus, the higher impurity levels with the higher

corrosion rate have a correspondingly higher loss of strength.

1.3 Delayed failure of a porous alumina [8]

Alumina bars (76.2 mm X 6.35 mm X 6.35 mm) were prepared from
powders. The bars were sintered to a density of 2.59 g/cm3 or 65.1% of the
theoretical density. At a loading rate of 2.26 kg/s the modulus of rupture was
determined in air, distilled water and in liquid nitrogen under 4-point bend testing.

X-ray diffraction verified that the specimens were a-Al203. The average
grain size was ~2 pm. The distribution of the impurities could not be determined
yet silicon and calcium were probably present at the grain boundary. Iron, which
was the most abundant impurity present (F9203), with a high solubility within
alumina (~10 mol%), was probably present within the grains.

Specimens were statically loaded within distilled water at stresses 80 -
100% of those determined for air given in Table 1.3.1. The temperature was held
constant by controlled temperature water running through copper coils
surrounding the flexural apparatus. Twelve bars (Table 1.3.2) were tested at

each of the following temperatures 27, 38, 44, 55, 65 and 76 Celsius (Table

1.3.3). Specimens that immediately broke under loading were not counted. The
“cut side” was under compression (facing upwards) in the loading of all of the
bars. Microswitch-controlled clocks automatically recorded the failure times to
the nearest 6 seconds.

The corresponding cross sectional fracture surface areas were measured
to the nearest 0.0254 mmz. A linear regression of the time to failure data was
performed and was found to correspond to Equation 1.3.1, where t. is the time to
failure, and T is the absolute temperature. The experimental activation entropic
energy is related to ac and a1 is related to the experimental activation energy.
The entropic and activation energy were determined from a least-squares
analysis, Table 1.3.4, which shows that both ac and a1 demonstrate a stress

dependency with ac having the stronger dependency.

Table 1.3.1: Breaking strengths of Alumina bars [8]

 

 

Medium Mean Strength (MPa)
Air (22°C) 41.4 i 7.64
Distilled water (22°C) 39 :t 6.24
Liquid Nitrogen (-203°C) 48.9 :I: 9.22

 

 

Table 1.3.2: Composition of 65.1% dense alumina [8L

 

compound AI203 SIOz F9203 080

 

Concentration (wt% i10%) 99.68 .095 0.12 0.07

Compound M90 3203 NaZO TiOz

 

Concentration (wt% :10%) 0.003 <0.003 <0.003 0.01

 

Table 1.3.3: Mean faggue life for porous alumina bars [8]

 

 

Temperature c/crm: 0.8 o/cm: 0.9 clam: 0.94
(i 1°C) (minutes) (minutes) (minutes)
27 24,0447 i 2,753.9 488.0 i 461.4 Not reported
38 9,833.6 i 2,448.5 97.8 i 55.3 57.6 i 22.4
44 2,449.4 i 996.0 61.7 i- 16.5 Not reported
55 648.7 i 336.1 11.2 i 7.4 5.6 i 3.8
65 350.0 : 268.6 Not reported 2.4 i 1.4
76 95.0 i- 54.7 Not reported Not reported

 

o = applied stress, em = mean breaking strength

 

Table 1.3.4: Values of a0 and a1 [8]

 

 

 

 

 

 

 

o/crm a0 a1
0.80 -12.2 4.96
0.90 -18.3 6.33
0.94 -19.6 6.67
a0 and a1 determined from Table 1.3.3 by least squares method
logtf =a11000+ao (1.3.1)
=l= _
tf = to exp[E b0] (1 .3.2)
E * —b0
1) = v — 1.3.3
f o expl RT ] ( )

The median experimental activation energy was determined to be 111.4 kJ/mol.
No other Arrhenius activation energy could be found that was associated with
delayed failure in porous polycrystalline alumina in the presence of water.
Further analysis of the data yielded the Equations 1.3.2-3, where E* is the zero-
stress experimental activation energy, 1) is the crack velocity, and the following
are system dependent constants: b, to and no.

Emission spectrography of the water is consistent with grain boundary
attack, within the alumina. A marked difference was displayed in the changes of
calcium and silicon concentrations. During a fatigue experiment, the
concentration of calcium in the water increased from a pre-fatigue value of 2 ppm
to a post fatigue value of 50 ppm. Initially, the silicon concentration in the water
was undetectable, but the silicon concentration increased to 15 ppm after the
experiment. Scanning electron microscopy analysis revealed that the fracture
was mostly intergranular in nature, with some transgranular characteristics.
There was no perceptible difference between the fracture surfaces of as-fired or
fatigued to fracture specimens under SEM analysis.

The substantial stress dependence apart from the activation energy
implies that domination by a transport mechanism operating between the
material and the environment is unlikely. Since the stress is applied to the solid
phase and not the interfacial liquid film, neither the diffusion coefficient nor the
effective free energy of transport mechanism activation should be affected.
Most etchants attack and weaken polycrystalline alumina at grain boundaries.

Porous aluminas are weakened by aging in water, as supported by the water and

SEM analysis, which suggests that the grain boundaries are attacked. The

activation entropy is probably the source of the strong stress dependence of a0.

1.4 Reaction of alumina ceramics with saturated steam [9]

In this study, a commercially produced dense 96% alumina, Table 1.4.1,
(AlSiMag 614, SM Corporation, St. Paul, MN) was reacted with saturated steam.
In a saturated steam environment, the atmosphere will not allow any more water
to be present in its gaseous state. Day chose to test the saturated steam attack
of a lower purity alumina (96%) because the researchers believed that the
degree of attack would be higher than that of prior work on 99% pure alumina
(Coors AD-99) that had been investigated in prior work [10]. X-ray diffraction
analysis detected (X'AI203 and trace amounts of sapphirine
(4MgO-5Al20302Si02) and cordierite (2MgO-2Al203058i02) in the as-received
plates.

The 96% alumina was sectioned into plates (50.8mm X 15.0mm X 1.5mm)
and tested in their as-received surface condition. Specimens were placed into an
autoclave and subjected to saturated steam at 286°C with the corresponding
pressure of 6.9 MPa. The modulus of rupture was then measured at room
temperature using a three-point bending apparatus, using a test span of 25.4 mm
was and a loading rate of 15 kg/s. Scanning electron microscopy (SEM) and
XRD was then used to examine the as-treated surfaces. The MOR decreased by

roughly 67% after an exposure at 286°C and a pressure of 6.9 MPa (Table 1.4.2)

10

After, an 8-day exposure the SEM analysis showed that the shape of the
alumina grains was much less distinct than the shape of the control specimens.
Control specimens were AlSiMag 614 specimens that had not been exposed to
the autoclave conditions. Additionally, a drop of ink placed on the exposed
surface of a specimen was absorbed and spread much faster on a treated
surface rather than when the ink was placed on an untreated surface. As
exposure time increased, cube-shaped crystals appeared on the surface, in an
irregular dispersal pattern. As the duration of the exposure increased, the
number of cube-shaped crystals also increased. The shape of the crystals
changed with continued exposure, however the once cube-shaped crystals were
still not distributed evenly about the exposed surface.

XRD analysis showed a higher concentration of silicon and calcium on the
surface after hydrothermal treatment. The same XRD showed little or no calcium
or silicon on the surface, which contained the cube-shaped crystals. The cube-
shaped crystals were calcium rich and identified as a wairakite compound

(CflO'AI203'4SI0202H20) .

ll

Table 1.4.1: Chemical Properties 96% alumina [9]
Al203 SiOz MgO CaO Density

(Wt%) (Wt%L awe/oi (Wt%) (g/cm3)
96 2.0 1.8 0.2 3.7

 

 

Table 1.4.2: Modulus of Rupture data for a 96% pure alumina [9L
Time in saturated Steam Average MOR * Standard Deviation

 

(Days) (MPa) (MPa)
0 199.1 i 26.5
8 99.7 i 22.0
15 70.5 i- 12.1
21 66.9 i 14.3
35 65.2 i 9.0

 

* Average of 10 specimens, except that 15 were tested at 35 days;
uncertainties = one standard deviation.

12

Wairakite is usually found in geothermal wells at temperatures between
ZOO—250°C and pressures of 5.5-26.5 MPa [11.12.13]. The steam pressure and
temperatures of this study [9] fall within these limits. Hydrotherrnaliy synthesized
wairakite can exist in either the tetragonal or cubic forms. Wairakite may form at
temperatures as high as 275°C if the precursor compositions are
CaO-AlzoaonSioz where n varies between 2 and 5. Naturally occurring,
wairakite is monoclinic. However, since the lattice dimensions for each of the
three, wairakite phases vary by less than 1.2 X 10'2 nm, CaO-AI20304Si02-2H20
is described as being pseudo-cubic. The XRD analysis showed that the crystals
from the 35-day specimens were cubic or pseudo-cubic in structure. The
formation of CaOoAl20304Si0202H20 was the only chemical reaction detectable
by either XRD or SEM examination of the external specimen surface.

The sapphirine and the cordierite were more resistant to saturated steam
corrosion than the phases containing calcium in the 96% alumina. The
preferential attack of the calcium containing phases is consistent with (1)
changes in the microstructure of the external surface and (2) the reduction in the
MOR. XRD failed to detect calcium-containing phases other than
CaOoAI20304Si0202H20 (Wairakite). A few volume percent of a phase must be
present in order for a phase to be detectable through XRD. If additional calcium -
containing calcium phases were present then those phases would most likely be
amorphous or glassy structures.

CaO-Al203-Si02 ternary compounds tend to form glasses [14]. The

microstructures observed on the fracture surfaces displayed grains that appeared

13

to be coated with a “glassy (liquid) film,” of unknown composition. The porous
microstructure of the corroded surface is reminiscent of the dissolution of a
glassy phase. The glassy phase is usually found along the grain boundaries and
within pores. Irregular distributions of glassy phases near the surface or within
pores could explain why the CaO-Al20304Si02-2H20 crystals were found in
patches instead of uniformly distributed on the surface. A lower MOR could
result from the chemical dissolution of a glass containing calcium.

Previous attempts to hydrothermally synthesize wairakite were successful,
only when there was an excess of SIOz in a CaO-A|203-4Si02 ternary
composition between ZOO—560°C and at a pressure of 100 MPa. The wairakite

of this study developed from a compositional range of CaO-Al20304.58i02 to

CaO-A1203-558i02.

1.5 Grain Boundary Segregation under Hydrothermal Corrosion

Grain boundary segregation occurs when one or more elements are more
likely to be found in the grain boundary than elsewhere within the material.
Some elements are more likely to form precipitates or reside in the bulk material.
Grain boundary segregation is important in ceramics because the grain boundary
phase often melts before the bulk material and depending on the type and
quantity of the constituents in the grain boundary phase, the lower that melting

temperature may be.

14

1.6 Grain boundary segregation on Argo. [15]

Photon stimulation of both photoelectron and Auger electron peaks, in X-
ray photoelectron spectroscopy (XPS), were used to observe grain boundary
segregation of MgO and CaO in Al203 ceramics (Lucalox, General Electric
Company, Schenectady, NY.). The grain boundary phase contained ~0.1%
magnesium, 0.0005 — 0.0015% calcium and the bulk material had a density of
100% of the theoretical density. XPS has a greater sensitivity to MgO
concentrations than the 3keV stimulation of Auger electron spectroscopy (AES),
which has been used in other studies [16].

Rods of Al203 (5.08 mm diameter X 38.1 mm long) were notched and then
fractured under vacuum immediately prior to examination. Unnotched specimens
(12.7 mm diameter X 38.1 mm long) were fractured outside of vacuum; this
resulted in a higher sensitivity, under AES, due to the larger surface area than
the notched specimens. Scanning electron microscopy verified the intergranular
nature of the fracture surface.

An X-ray source and analyzer coupled with incident radiation from an
aluminum target (1486.6 eV) was used to examine the energy spectrum of the
emitted electrons. The process is less sensitive to depth than AES but the
process is identical to AES for Auger electrons such as magnesium.

Both MgO and 030 electron energy spectra were detected at the as-
fractured specimen surface. The magnesium fracture surface concentration was
around 0.3 atomic percent and the calcium concentration was approximately 0.58

atomic percent where the atomic percent concentrations are accurate to ~20%.

15

Thus while the bulk concentration of magnesium is higher than that of calcium
the segregation of calcium to the grain boundaries is more extreme. The
elevated calcium concentration is present within ~5 nm of the grain boundary.
However, in sintered alumina, calcium is commonly strongly segregated to the
grain boundary phase.

Fracture surfaces were bombarded with argon in order to investigate the
fracture surface at greater depths. Subsequent to bombardment, there was no
calcium peak present on the original surface. At a depth of ~10 nm the
concentrations of magnesium and calcium are identical to that of the bulk
material.

The width of the segregation region is consistent with that of solid solution
segregation arising from ion charge differences between impurity (Ca2+ and
Mg“) and host (Al°+) ions. The depth of segregation is expected to be on the
order of the Debye length or ~1 nm for Al203 [17]. Therefore, the rapid
decreases in concentrations within 5nm of the fracture surface are attributable to
the existence of solid segregation to the grain boundary. However, a second
phase existing as islands at the grain boundaries, which has been observed with
M90 in Al203 [18]. The XPS signal would therefore be an average of the area
concentration of the second phase islands as the surface was abraded by

progressive ion bombardment.

l6

1.7 Surface segregation of calcium in dense alumina exposed to steam and
steam + CO [10]

When exposed to steam and steam + carbon monoxide (CO) gases at
high temperature (~227, ~265 and ~285°C) and pressures (~2.65, ~5.17 and ~7
MPa) dense alumina rods experiences a considerable loss of flexural strength.
Carbon monoxide was chosen as a constituent with the steam because other
ceramics [19] had lost stability under a steam and carbon monoxide atmosphere.
Previous work [19] had attributed this loss to the formation of either boehmite or
diaspore on the surface. Auger electron spectroscopy (AES) was used by
Sinharoy to determine changes in the elemental composition of the surface of
corroded alumina specimens. The AES data was then compared with modulus
of rupture data for these specimens.

Coors AD-99 disks (12 mm diameter X 1 mm thick) were sectioned from
the end of the exposed rod. Some of the disks were then fractured in air before
being mounted for use in the analysis of the fracture surfaces. A small single-
crystal of sapphire was used as reference for the AES. A primary energy of 2
keV coupled with a primary beam current of 6pm was used to minimize charging
problems. The AES was conducted in an ultra high vacuum (10'9 torr) chamber
with a double-pass cylindrical mirror analyzer. Sputtering rate was approximately
30 i5 Almin with, a defocused argon ion beam at 2 keV yet some of the
specimens still had to be sputtered for 5 minutes before Auger spectrums could

be realized.

17

Aluminum, oxygen and trace amounts of carbon contaminants were
detected on the surface of the sapphire in its AES analysis. After being sputtered
for 2 minutes, the carbon was no longer detected.

In the AD-99 there was little to no carbon detected on the fracture
surfaces of either exposed or unexposed specimens. However, under the
influence of steam (or steam + CO) there was a carbon detected on the surfaces
exposed to those environments. This carbon / oxygen signal ratio quickly
dwindled during depth analysis yet there was some carbon present to a depth of
1 pm. Depth profiling revealed no alterations in the aluminum / oxygen Auger
signal ratios.

The surface calcium content depended on the history of the commercial
specimens, the calcium / oxygen signal ratios from steam corroded alumina
showed an apparent segregation of the calcium from the grain boundaries to the
surface of the specimen. The amount of calcium present on the surface
increased as the pressure increased to ~5.17 MPa with the corresponding
temperature. As the pressure increased to ~7MPa the highest calcium
concentration was found at a depth of ~45 nm.

An environment of steam + CO (48% by volume CO), with a pressure of
~2.65 MPa, yielded a depth profile almost identical to that of the pure steam. At
~227°C there was no effect on the surface segregation of calcium due to carbon
monoxide. When the pressure was increased the corroding environment
became steam + CO (49% by volume CO). This environment had an effect on

the medium pressure (~5.17 MPa) specimens in that the calcium I oxygen ratio

18

peaked at a depth of ~33 nm. Whereas the ~7 MPa specimen had a much lower
concentration of calcium on the surface than the other two exposures, Sinharoy
[10] suggested that the calcium concentration may have been lower due to
improper mounting within the spectrometer.

In order to determine the exposure time kinetics of the calcium migration,
the alumina rods were exposed to saturated steam at ~5.17 MPa for different
time-spans. The calcium concentration increased as exposure time increased at
~5.17 MPa.

When the fracture surfaces were compared (exposed and unexposed to
hydrothermal conditions), there was a higher concentration of calcium in the
exposed specimens than the unexposed specimens. Calcium has a larger ionic
radius when compared to aluminum or magnesium, which may explain calcium
mobility and segregation on the surface of MgO doped Al203 [15]. Due to this
larger size, the calcium is less stable within the AI203-MgO matrix and in order to
relieve the strain energy the calcium migrates to both grain boundaries and free
surfaces [16]. Sinharoy’s results infer that the presence of steam or steam +
carbon monoxide increases the segregation process to the fracture surfaces.

Each specimen that was analyzed by AES was also tested for a
corresponding modulus of rupture (MOR). There was a decrease in flexural
strength for both exposed and unexposed to hydrothermal condition specimens.
The MOR change was more pronounced for the case of pure saturated steam.
When directly compared with the AES data there is a linear correlation between

the MOR decrease and the calcium increase near the surface. At ~7 MPa the

19

(steam + CO) specimen had less calcium near the surface than the lower
pressure specimens (~2.65 and 5.17 MPa) for which there is no clear

explanation.

1.8 Grain boundary segregation In MgO-doped AI203 [16]

An alumina mixture containing ~0.1% magnesium and between 0.0005 —
0.0015% calcium was sintered to almost 100% density. Magnesium was added
to the alumina because magnesium inhibits grain growth by segregating at the
grain boundaries. The magnesium inhibition theory has been inferred from the
change in the lattice parameter of the Al203 grain on the addition of MgO to the
matrix. Mass spectroscopy was performed via a spark source at an average
depth of 300nm from the fractured grain boundary. The spectroscopy indicated
an excess concentration of the dopants: MgO and CaO. Auger electron
spectroscopy (AES) with a primary electron beam (2-3 keV) was used in this
study to determine the fracture-surface chemistry.

The cylindrical electron-velocity analyzer within the AES chamber, coupled
with a 10‘10 torr vacuum system was used to analyze the specimens. Alumina
specimens (2.54 mm X 3.56 mm X 2.54 mm) were notched to produce desired
fracture surfaces and fracture surfaces at room temperature within the vacuum
system under impact loading. The fracture surface was then maintained at 10'9

torr and chemically analyzed with the Auger spectrometer.

20

Analysis of the as-fractured primary Al203 yielded a calcium concentration
at the fracture surface (2 - 5 atomic%) and no detectable magnesium. The
detectable range for the spectrometer is roughly 0.1%, which is also the bulk
concentration of the magnesium ion. However, the calcium concentration at the
fracture surface is approximately 3-orders of magnitude higher than the bulk
concentration of calcium. About 6.4 nm of the fracture surface was then
sputtered off so that a depth profile could be performed. The results of the profile
were that almost all of the calcium was absent at that depth, agreeing with the
surface analysis that the calcium had migrated from the bulk material to the
surface.

As calcium and magnesium have identical ion charges, there is little
reason for one to segregate faster than the other from that basis. Previous, AES
work has assumed that more electronegative ions segregate to grain boundaries
more efficiently than others. Numerous elemental groups (3A, 4A, 5A and 6A)
have elements that will segregate to grain boundaries in various systems.
Electronegativity and variances in bonding were believed to be the cause of
segregation of these species.

The ionic radius of calcium (~0.09 nm) exceeds that of magnesium
(~0.065 nm) or aluminum (~0.05 nm). The consequence of the differing atomic
radius is that calcium is much less stable within the matrix and will be more
mobile. Therefore, calcium will tend to migrate towards the grain boundaries and

other free surfaces to relieve strain energy [16].

21

Additionally, the preferential segregation of calcium over that of
magnesium, which is present in a higher concentration, may be a result of
magnesium being tied up as MgO-type precipitates or clusters in the matrix.
Therefore, the overall concentration of “solid-solution” magnesium in the matrix

would be considerably lower than that of calcium.

1.9 The Effects of Hydrothermal Corrosion on Non-Oxide Ceramics

The properties of oxide and non-oxide ceramics may be different under
othenivise identical conditions. Therefore, looking at the effects of non-oxide
ceramics under identical conditions may yield that a non-oxide ceramic is better
for a given application or that the response of the non-oxide ceramic is useful in

understanding the response in a oxide ceramic.

1.10 Corrosion behavior of silicon carbide In 290°C water [20]

The oxidation reaction of SiC powders in high-temperature air may be
controlled by the surface reaction in Equation 1.10.1. In the presence of water
vapor, the reaction is accelerated. The reaction can be viewed as a two-step
process, Equations 1.10.2-3. This chain reaction is limited by the amount of CH4,
and thus the diffusion of methane through an amorphous layer of SiOz formed on
the surface of the SiC.

The reaction of SiC to high-temperature water is believed to be similar to

that of a water-vapor phase. The reaction is expected to accelerate due to the

22

high-temperature water dissolving the SIOz, surface layer. The effects of ionic
species on the silicon carbide are also unknown. The purpose of this study is to
determine the effects of dissolved oxygen and pH on the corrosion behavior of
SiC immersed in different test solutions.

A mostly cit-phase sintered SiC with boron and carbon sintering additives
was sectioned (3 mm X 4 mm X 40 mm) from plates (100 mm X 100 mm).
Samples were then polished with diamond abrasives to a surface roughness of
less than 1 pm the samples were then cleaned with acetone and vacuum dried at
120°C for 48 hours and weighed (i 0.01 mg). A 5-liter titanium lined refresher-
type autoclave with a nitrogen or oxygen gas bubbling system was used (Table
1.10.1).

The degree of oxygenation in the system was controlled through two
methods. The first method entailed deoxygenation of the test solution by
bubbling nitrogen gas through the mixture, this kept the oxygen content to below
20 ppb. The other method was to oxygenate the solution by bubbling oxygen
into the mixture to a content of ~32 ppm. In both cases, the amount of oxygen
was determined using a dissolved-oxygen meter. Specimens were held in a
tetrafluoroethylene beaker (100 ml) while in the autoclave. Specimens were
exposed for 72 hours at 290°C and then dried and weighed to calculate the
weight change. Surface structure and morphology were analyzed by X-ray
diffraction and SEM analysis respectively. The exposure times were increased

from 3 to 200 hours in order to determine the corrosion rates of the test solutions.

23

The effects of oxygen in the testing media were such that the weight loss
was higher in all‘oxygenated cases compared with deoxygenated solutions.
Additionally, as the pH of the solution altered from neutral the degree of attack
also increased. The weight loss was greater with a pH further from neutral and
oxygenation than with a similar pH and a deoxygenated media. The degree of
weightless increased as time increased. As the acidity increased, the weight
loss was found to follow a parabolic law, whereas, the alkaline weight loss
seemed to follow a linear relationship.

Microscopy of the specimens revealed no notable changes in the
morphology for the specimens attacked by acid. The high-purity water samples
yielded a low degree of attack at the grain boundaries. Additionally, at higher
magnifications, preferential intergranular attack was indicated by the presence of
individual crystal shapes. The 10-pH solution yielded some large crystals on the
surface after 200 hours of exposure and enlarged pores after the 72-hour
exposure. Furthermore, the intergranular corrosion was found to have reached a
depth of 60 um after 200 hours from a fracture cross-section.

After 200 hours of testing the LiOH specimen had traces of LIZSi03 in the
XRD analysis. The large crystals are probably the source of the 1.123103. The
SiC was found to be hexagonal in structure both before and after immersion. Of
interest is that there was no trace of an amorphous phase on the surface of any
of the specimens in XRD.

There are various methods, which could explain the dissolution of the SiC

surface, some of these methods result in an overall weight gain, and must thus

24

be discounted. The proposed dissolution model is based on a poorly adherent
layer to be formed on the surface of the silicon carbide. A possible nonprotective

layer, but not the only possible layer, is Si(OH)4.

25

Table 1.10.1: Compositions of testing solutions

 

 

Type of Solution pH
Pure water 6.2 at 290°C
0.045M NaZSO4 + 0.005M H2804 4 at 290°C
0.1M LiOH 10 at 290°C

 

SiC + 202 = SiOz + 002

SiC +2H20 = $102 + CH4

CH4 + 202 = 002 +2H2

SiC + 4H20 = Si(OH)4 + CH4

31(01-1)4 = H38i04' + H1 = H2310?" +2H2+

SiC + 202 + 2H20 = Si(OH)4 + 002

Si(OH)4 + 002 + H20 = H38iO4' + HCO3' +2H+

H38i04' + HCOa‘ +2H+ = H28i042' + 4H“

26

1.10.1
1.10.2
1.10.3
1.10.4
1.10.5
1.10.6
1.10.7

1.10.8

Equations 1.10.4-5 are for the dissolution of Si(OH)4 in the case of a
deoxygenated solution. In the case of an aerated solution, the Equations 1.10.6-
8 may be the case. Weight loss is the result of both of these possible reactions.
Twice as many free hydrogen ions are produced in the oxygenated case as in
the deoxygenated situation. Additionally, the aerated solution has a higher
dependency on pH than in the deoxygenated solution. Thus, the above reactions
support the results of this experiment.

Linear regression analyses of the weight loss due to corrosion data
yielded a slope of 1.1 for the alkaline (pH 10) solution and 0.45 for the acidic (pH
4) solution. These slopes or rates of reaction are also explained by the above
equations, as reaction (1 .10.5) is largely dependent on the pH of the solution and
controls reaction (1 .10.4). Si(OH)4 dissolves rapidly in an alkaline solution with
no protective effects on the SiC substrate. The rate of the corrosion reaction is
greater than 1.0 in the pH 10 mixture because the deep intergranular corrosion
greatly increases the surface area susceptible to attack. Conversely, the
dissolution of Si(OH)4 in an acidic solution is delayed. Therefore the Si(OH)4 still
has a protective effect, thus the acidic attack approaches a parabolic rate law
(0.45 vs. 0.50). An outlier, in the 200-hour alkaline corrosion trial exceeded the
result predicted by the linear regression of the data. The presence of the Li(SiO)i
in the XRD analysis suggests that ionic lithium accelerated the SiC dissolution in

the alkaline solution.

27

1.11 Degradation of SiC fibres In high-temperature, high-pressure water
[21]

In order to use Tyranno (Si-Ti-C-O) fibers in composites, the mechanical
properties of the fibers should be fully understood. The oxidation behavior is an
extremely important factor in the design and application of the fibers. At the time
of Gogotsi work there had been no previously published work on Tyranno fibers,
however there had been work done on SiC [22], Nicalon [23.24], powdered and
densified silicon carbides [25]. The oxidation rate of the powdered or densified
SiC was enhanced in the presence of either water or wet air. The influence of
high-pressure, high-temperature water has not been previously reported.

Gogotsi investigated the oxidation of Tyranno fibers by high-pressure, high
temperature water and determined the tensile strength and Young’s modulus of
the fiber following corrosion.

Gogotsi studied commercially available (UBE Industries Ltd, Tokyo,
Japan) amorphous Tyranno (Si-C-Ti-O) fibers (Table 1.11.1). The analysis of the
composition and structure was performed through XRD, SEM, Fourier-transform
infrared spectroscopy (FT/l R3), and energy-dispersive X-ray microanalysis
(EDAX). Tyranno fibers are made through the pyrolysis of organometallic
polymer precursors. The fibers were of a uniform thickness and circular when
viewed on end. At high magnifications, the microfilaments (in small bundles and
individually) were aligned at the surface of the Tyranno fibers. The commericial
characterization of the Tyranno fibers and the experimental characterization were

in much agreement (Table 1.11.1).

28

Table 1.11.1: Properties of the Tyranno fibers [21]

 

 

TY-S1 E08PX TM-SI E08PX TY-F1 1H16PX
Sup Exp Sup Exp Sup Exp
Grade S LoxM F1
Number of fibers / yarn 800 800 1600
Diameter (um) 11:2 11.3i0.4 11:2 10.6:08 8.51:2 7.8:0.4
Tensile Strength (G Pa) 3.43 4.00:0.67 3.73 3.40:0.63 3.27 3.30:0.54
Elastic Modulus (GPa) 173 159:7 187 192i20 171 173:33
Composition (wt%)
Silicon 50 49.2 54 53 50
Carbon 30 29.0 32 31 .0 30
Oxygen 18 17.5 12 12.2 18
Titanium 2 1.9 2 2.0 2

 

Density of all fibers was 2.3 - 2.4 g/cm3
Sup: are values given by the supplier
Egg: Are values determined through x-ray diffractometry

 

SiC + 2H20 = SiOz +CH4

TiC + 2HzO = TiOz + CH4

TiC +4H20 = TiOz +002 + CH4

SiTinyOz + nHZO = $102 + xTiOz + yCH4
SiTinyOz + mHZO = SiOz + XTIOz + yCOz + mH2
SiOz + nH20 = SiOz-nHzo

29

1.11.1

1.11.2

1.11.3

1.11.4

1.11.5

1.11.6

A thin polyethylene oxide (PEO) film was used to coat the as-received fibers; the
coated surface of the fibers was smooth and featureless. The fibers (50 mm
length) were placed with distilled water inside a gold capsule (60 mm X 5 mm
diameter). Bending and flattening the top of the gold capsule prevented fiber or
liquid loss. A test-tube type pressure vessel (100 MPa) was used to heat the
capsule to temperatures ranging from 300 - 600°C for 25 hours.

The PEO layer was removed prior to mechanical testing, through
ultrasonic cleaning with hot water. As-received fibers, fibers that had only had
the PEO layer removed and fibers after hydrothermal experimentation were
mechanically tested at room temperature. An apparatus modeled after American
Society for Testing and Materials (ASTM 03379-75) and Japan Institute of
Standards (R 7601-1980) schematics used a 25 mm gauge length. Paper
frames were used to aid in handling, aligning and supporting the fibers post
exposure during transferal to the testing machine. Epoxy resin on a paper mount
was bonded to a single fiber pulled from the yarn. The jaws of the testing
machine were affixed to the paper mount and then the middle section of the
mount was cutaway. A 5 kg capacity load cell mounted within a Shimadzu
tensile testing machine was used to the filaments (0.5 mm/min). At least 10
fibers were tested to determine the mechanical characteristics.

While no surface damage or change in strength was found, following
300°C corrosion, there was a minimal (<10 GPa) decrease in the Young’s
modulus indicating some change structural changes. There was a significant

decrease in the properties of Tyranno fibers following hydrothermal corrosion

30

between 300 — 400°C (~40 GPa for LoxM grade). The Si-C-O fiber may be
represented as B-SiC nanocrystals separated by a Si01.1500,85 phase, which
mimics a grain boundary. There is an additional intergranular free carbon phase
consisting of randomly oriented basic structural units. At an elevated
temperature (300-400°C), water may dissolve into the oxycarbide phase, or the
oxycarbide phase may be dissolved in the water or a combination of the two
processes may occur. Either process could cause the decrease in the
absorption due to Si-O (820 cm") stretching that was observed for this
temperature range in FT/IR analysis.

After corrosion at 400-450°C, minimal surface damage had occurred on
the LoxM fiber. Multi-layered oxide films were found on the surface of the fibers
following corrosion above 400°C, through SEM observation. The protective
properties of the multi-layered film were low, and coupled with an easy
delamination. The result of the formation of the oxide film was that the effective
diameter of the fiber was much less than that of the outer fiber diameter used in
the strength determination (6.5 pm instead of 10.2 pm for S-grade after 400°C
corrosion). When the diameter of the inner Si-C-Ti-O core is used instead of the
outer diameter, the strength values were near to those of the as-received fibers,
indicating that the core diameter should be used for the initial strength
determination. From the correlation of the as-received strength and that of the
core diameter calculation infers that a uniform corrosion sans pitting or local
damage. Unfortunately, hydrothermal corrosion at temperatures above 400°C

induced radical decreases in the fiber diameter to a value < 6 pm. The LoxM

31

fibers (100 MPa, 500°C) suffered from complete degradation after 25 hours of
exposure. The dilapidation was so relentless that the exposed fibers were too
weak to be mounted in the Shimadzu system.

The fiber, to a first approximation has a core of Si-Ti-C-O within a 8102/
TiOz shell. As the thickness of the corrosion layer increases the Young’s
modulus of the fiber decreases. The experimental values (~90, 130 and 0GPa
for S, LoxM and F1 grades respectively) of the modulus agree with the values of
the modulus calculated by the rule of mixtures.

The corrosion of Si-Ti-C-O, hydrothermally is likely to be controlled by the
diffusion of CH, or H20 through an amorphous silica layer (Equations 1.11.1-3).
Thus, the Si-Ti-C-O degradation may progress as reactions of either equation
1.11.4 or 1.11.5. FT /lR analysis showed that absorption due to the stretching of
silicon-oxygen bonds (1080 cm") became stronger while absorption from the
stretching of silicon-carbon bonds (820 cm") disappeared above 500°C. There
was no crystallization of silica at the observed range of temperatures. Both rutile
and anatase, forms of TiOz, were discovered through XRD at temperatures
above 450°C. Additionally, the titanium content increased on the surface of
corroded specimens (SOD-600°C) compared with as-received fibers. The silica
layer was partially dissolved which may have increased the percentage of titania
on the fiber surface, yet the role of titanium in the dissolution process is unclear.

The fibers exhibited a slight weight gain following corrosion above 300°C

and drying, possibly from water diffusing into the oxycarbide phase. The

32

diameter of the specimens was decreased, suggesting dissolution of the
protective silica layer:

Corrosion rate is enhanced through both compressive stress and
hydrostatic pressure. Therefore, there are two processes at work: one, which
increases the Tyranno fiber oxidation and another that attacks the protective
silica layer. The rate of this reaction in distilled 285°C water (100 MPa) was
approximately 0.8 umh".

The thicker fibers (LoxM and S) remained stronger than the F1 grade fiber
after testing. The thickness of the fibers alone cannot explain the degradation
phenomena, as the S grade is thicker than the LoxM grade, but showed a greater
loss of strength than the LoxM fibers. The S grade fibers have a lower stability,
possibly due to oxygen content, than the LoxM, which links the mechanical
properties after oxidation to the fiber composition.

At temperatures above 450°C, the silica layer cracked and spalled. The
resulting lamellar oxide films are not protective, permitting high-pressure, high-
temperature water to completely oxidize (S and LoxM from SOD-600°C, F1 at
400°C) fibers. The diffusion of water through the SiOz layer may be accelerated
under experimental conditions causing delamination of the once protective layer
during cooling. Transformations according to Equations 4 or 5 induce high
stresses and strains, which cause cracking and flaking of the brittle oxide layer.
However, water dissolution in the SiOz layer at temperatures below 500°C

accounts for plasticity in the layer in those temperature regimes. The Fl'llR

33

investigation revealed the presence of water in the silica glass at temperatures

up to 400°C.

1 .12 Hydrothermal oxidation of Siam powder [26]

Silicon nitride and its derivatives are of keen interest due to their high
temperature application, but the oxidation behavior of the nitrides is of concern.
Contet studied the hydrothermal degradation of 99.9% pure Si3N4 powders
(Cerac, Inc) with a phase distribution of 60% a-Si3N4 and 40 B-Si3N4. When
oxygen is present, the corrosion reaction is believed to progress as Equation
1.12.1. If water is present then the rate of degradation increases as depicted in
Equation 1.12.2. The hydrothermal reaction given in Equation 1.12.2 has been
observed at high temperatures and pressures by SC Singhal [27].

A gold capsule (2.7 mm inner diameter, 2.85 mm outer diameter, 35 mm
long) was filled with about 0.19 of Si3N4 powder, the bottom of the capsule was
flattened and welded closed prior to experimentation. The top of the capsule was
then pressed flat to prevent escape of the solid material. High-pressure
redistilled water filled the cavity space within the test-tube-type pressure vessel
(6.6 mm inner diameter, 16.0 mm outer diameter, 200 mm long) alongside the
golden capsule. The temperature of the experimentation varied form 200-600°C,

with the corresponding pressures of 10-100 MPa, with a duration up to 72 hours.

34

$13N4 + 302 -) 33102 +2N2 1.12.1

$13N4 + 6H20 9 302 +4NH3 1.12.2

1 3 1 3
[1—(1—a,)/12-[1—(1—a0)/]2=(-2—'2‘—)t 1.12.3
r

35

A room temperature quench was performed on the pressure vessel following
each trial. Using Nessler’s reagent ammonia was detected, after the
experimentation was completed.

The solid residue was weighed, after drying the capsule, so that the
oxidation ratio of Si3N4 could be determined. This oxidation ratio, a function of
time, temperature and pressure could be measured from the weight gain. At
high-pressures (100 MPa) the oxidation rate increased with temperature, which
may be represented by a Jander—type function (Equation 1.12.3). Where t is the
time, or is the oxidation ratio at that time, do is the oxidation ratio (at t = 0), k is the
rate constant and r is the radius of the particle. Water transforms from a liquid to
a gaseous phase at 310°C, below at low pressures (10 MPa) the hydrothermal
oxidation reaction parallels the high-pressure reaction but slower. Above the
transformation temperature, the reaction rate decreases further.

XRD was used to examine the residue to assess the rate at which the two
phases oxidized. The analysis indicated that the oxidation reaction formed a
non-crystalline silica product. If the reaction is allowed to complete, the silica will
transform into a crystalline cristobalite.

The corroded surface of Si3N4, when examined through the SEM, is
covered with SiOz. If permitted the grains will begin to grow and sinter.
Combined with the X-ray data suggests that the diffusion process of water
through silica layers controls the hydrothermal oxidation of Si3N4.

The reduction of the or-form decreased as time progressed while that of

the B-form increased. This variation in oxidation rates for the different phases

36

could be detected through X-ray diffraction patterns. The size of the crystallites
formed from the different phases could not be detected through XRD analysis or

transmission electron microscopy.

1.13 The Effects of Hydrothermal Corrosion on Glasses

Grain boundary phases have little to no long —range order and are therefore
amorphous solids. Glass is an amorphous solid. Since the grain boundary
phase of alumina ceramics is preferentially attacked, the hydrothermal corrosion

properties of glasses must be investigated.

1.14 Hydrothermal corrosion of lithium disilicate glass and glass-
ceramics [28]

The compositions of lithium disilicate (LigOoZSioz) glasses are
stoichiometric with respect to crystalline and glassy phases. Coupled with the
simplicity of a binary system as compared to a multi-component system the
LiZOoZSiOZ system was chosen. For study under hydrothermal conditions
(200°C and ~1.28 MPa) using specimens having 20, 60 and 90% crystallinity by
volume, investigations were carried out in vapor and liquid phases. Increases in
temperature and / or pressure increase the rate of a reaction, however the effects
of temperature are generally more pronounced, than the effects of pressure on
the rate of a reaction. Additionally, the devitrification rate of amorphous silica is
enhanced by the presence of hydrothermal conditions. The durability of a glass

increases when crystallization is increased. Thus, if the amorphous surface

37

undergoes crystallization, the crystallized surface may be more resistant to
chemical attack.

A platinum crucible and an electric furnace were used to melt reagent
grade Li2C03 and min-u-sil glasses at a temperature of 1350°C for 24 hours. A
graphite mold was used to cast glass patties at 450°C for 6 hours. Glass
cylinders (33% LIZO) were nucleated at 475°C for 24 hours and crystallized at
550°C. Changing the length of time at the crystallization temperature controlled
the degree of crystallinity such that the three different crystalline volume fractions
20, 60 or 90% were obtained. The actual volume fraction that was crystallized
was determined through quantitative stereology. The cylinders were then
sectioned (~1 cm X ~0.5 cm X ~0.3 cm) using a diamond saw. After polishing to
600-grit on SiC-abrasive papers, the sectioned specimens were rinsed in
acetone and allowed to dry in air.

The specimens were placed within a Teflon container. For the aqueous
condition, the specimens were submerged horizontally in deionized water while
resting on Teflon matting. For the gaseous corrosion, Teflon matting supported
the specimens vertically above the deionized water. The Teflon container was
placed in water within a stainless steel pressure vessel. The pressure vessel
was then placed in an autoclave and corrosion treatment commenced (200°C
and ~1.28 MPa).

Additional specimens were corroded for 10 hours at room temperature
within deionized water. Another set of specimens was attacked for 9 hours at

90°C in deionized water. Finally, sodium (Na20 —33mol%) - silica (SiOz -67

38

mol%) specimens of glass were corroded with steam from deionized water, and
deionized water for 10 hours.

In all of the trials, the specimen surface area to the volume of solution ratio
was 0.1 cm". After the Teflon cells were removed from the autoclave,
specimens were rinsed, dried with acetone and stored in a desiccator until
analysis. Analysis was performed through infrared reflection spectroscopy,
scanning electron microscopy and X-ray diffraction. Subsequently, pH values of
the solutions were obtained at room temperature, as well as the concentrations
of alkali ions and SiOz through atomic absorption, emission spectroscopy and
calorimetry.

Following exposure to deionized water (200°C) for 10 hours, the
specimens of lithium disilicate glass had a severely corroded surface. The
morphology indicated widespread alkali leaching and a nonprotective surface.
The surface was high in silica, yet the silica concentration was not sufficient to
protect the glass from either dealkalization or network dissolution, resulting in a
poor durability.

Samples that were solely exposed to the vapor phase reaction had two
protective surface films. The protective film of interest displayed a surface
coated in plate-like particles. A dual surface film glass is commonly seen in
instances where there are either AI203 or P205 additives. These glasses may be
very durable in both caustic and acidic environments, and may be formed
through dealkalization, precipitation from solution or surface structural

modifications.

39

Varying the volume percentage of crystallinity showed that when
subjected to vapor phase corrosion, higher degrees of crystallinity had a higher
resistance to attack.

Infrared spectroscopy could not be performed on the specimens that were
corroded at 200°C as the extensive corrosion had created a surface too rough for
reliable reflection. The specimens tested at 25°C displayed a large peak due to
the stretching of Si-O-Si bonds (1100 cm") coupled with a stretched Si bond
(900 cm“). When the shape and composition of the preceding peaks from the
corroded specimens were compared with non-corroded specimens, an exchange
between alkali and hydrogen ions was evident, through infrared spectroscopy.

The infrared reflectance spectra of the 90 volume-percent crystalline
glass-ceramic corroded above deionized water (200°C), was very similar to that
of the uncorroded specimen. The two spectra are analogous because the 90
volume-percent crystalline glass ceramic has a higher chemical stability than the
glass. However, the specimens corroded in deionized water (25°C or 200°C) had
peak intensities that were much lower than the uncorroded specimens due to
surface roughness.

An analysis of the major peak positions from the powder x-ray diffraction
data on lithium metasilicate by Chao [28] found that Li200$i02 was the most likely
composition of the major film component. It is difficult to determine the minor
constituents from the remaining peak positions however LiOH-H20 is a likely
candidate. There is a high possibility of other crystalline products within the

surface films in quantities below the detection limits of the x-ray diffractometer.

40

Metastable lithium metasilicates appear as precursors to the crystallization
of lithium disilicate glass in air at temperatures above 410°C. However, when
placed within an autoclave the devitrification of lithium disilicate glasses may be
accomplished after 93 hours at 200°C. Thus, the formation of crystalline lithium
disilicate glasses is enhanced in the presence of steam.

The dissolution rate of the lithium disilicate glasses was affected by the pH
of the corrosive water. As the water becomes more caustic the dissolution rate
of the lithium disilicate glass increases. There was no data reported for a pH less
than 6.81 so all specimens experienced relatively neutral to highly alkaline (11.65
pH) water. The trend in the dissolution rate was observed at varying degrees (0,
20, 60 & 90 volume%) of crystallinity. Following dissolution the possibility exists
for weight gain or loss. Most cases resulted in a weight loss however where
there was a gain it as attributed to a trapping or inclusion of water beneath the
corrosion film on the glass. The analysis of these solution trends coupled with
scanning electron microscopy and infrared reflection microscopy showed that the
lithium disilicate glass and glass ceramics were more successfully attacked by
water than water vapor.

The results of Chao [28] contradict earlier studies where the vapor phase
attacked silicate glasses more severely than the water phase. Prior studies were
conducted on soda-lime glasses and were explained by a more rapid increase in
the pH of the vapor [29,30]. Charles [30] explained the phenomenon in boiling
HCI in terms of a high SAN ratio and “a continuously replenished condensate

which enhanced network dissolution and minimized solubility effects.” In order to

41

investigate this disagreement in findings, Chao [28] conducted trials above and
within deionized water for 10 hours (200°C). The result of these comparison
trials were that in both cases, liquid and vapor phase corrosion the soda-lime
glass was completely dissolved, whereas the lithium disilicate glass was severely
attacked in the water phase yet not in the vapor phase. These marked
differences between lithium disilicate (glass and ceramics) and soda-lime
glasses, in 200°C deionized water, are attributed to a crystallization of the
surface film due to hydrothermal conditions, which thereby imbues a resistance
to vapor phase corrosion. Limited transport to and from the specimens is not
likely as that would lead to a similar type, and degree, of attack on both soda-
lime and lithium disilicate glasses.

The surface crystallization effect becomes less important as the degree of
bulk crystallization increases. Since the crystallinity is protection against
corrosion, the material is protected as long as there is crystallinity and is not

dependent on the form (bulk or surface) of the crystallinity.

1.15 Summary of the Literature
From the literature [5, 8, 9, 10, 15, 16] it may be determined that when
alumina is attacked by an etchant the corrosion occurs at the grain boundaries.
This corrosion progresses faster when the porosity is higher. The apparent rates
of corrosion and the strength loss increase with the impurity concentration.
Additionally, calcium [5, 8, 9, 10, 15, 16] strongly segregates to the grain

boundary region of the alumina in the presence of any combination of steam and

42

pressure. Calcium is believed to have a high mobility due to the electronegativity
[16] and larger size of the calcium ion [10] when compared with the size of other
common dopant ions. Although magnesium is a common dopant within alumina,
magnesium may be present as magnesia-based precipitates or clusters reducing
the amount of solid solution magnesium along with reducing the mobility of the
bound magnesium ions.

Characterization of the porous alumina under hydrothermal fatigue
conditions [8] yields that the service life is reduced as the temperature increases.
The service life also decreases as the applied stress increases.

The work in SiC [20, 21] along with the Si3N4 [26] and Li2002Si02 [28]
studies further emphasize that ceramic systems are adversely affected by
hydrothermal conditions over relatively short time periods (<10 days). However,
for all the specimens the, majority of the damage originated at the grain
boundary.

Precipitates are expected to grow on the surface of the alumina as the
alumina undergoes hydrothermal corrosion; these structures are likely to be

diaspore, wairakite and boehmite [5,9,10].

43

Chapter 2: Experimental Procedures
2.1 Material Properties
Table 2.1.1 gives the physical properties of Coors AD-94 and ADS-995
available from the Coors Ceramics Company. In general, the vendor provides
more data for the lower purity material than the ADS-995. However, material
properties have a tendency to be larger at higher purities for ceramic materials,

thus, the ADS-995 should have larger values for most properties.

2.2 Sectioning Procedure

An lsomet Low Speed Diamond Saw (Buehler LTD: Lake Bluff, Illinois),
with a Norton diamond wheel saw blade Model M4D220-N75M99- 1/8 (Norton
Company: Worchester, Massachusetts), was used to cut specimens from the as-
received billets. The ceramic materials are very brittle, and when cutting there is
a chance of them prematurely breaking due to their own weight. To decrease the
chances of fracture during cutting, the specimens were affixed to glass
microscope slides, using thermoplastic Cement Model 40-8100 (Buehler LTD).
The glass slide supports both sides of the cut and decreases the likelihood of
breakage.

Coors ADS-995 polycrystalline alumina specimens were sectioned from a
large billet of the material (113.95 cm X 114.1 cm X 0.105 cm). Since the original

plate was large, the initial cutting lasted from eight to sixteen hours.

Table 2.1.1: Properties of Coors AD-94 and ADS 995 [31]

94% Pure 99. 5% N20,
7 Units 77Al207s (AD-94) (ADS-995)
‘ '5 ‘55-? 55.. 55Qms7éc -~ 5 " / 5,
Average Crystal Si27e
Water Absorption ‘51’5’2‘5 —
Gas Permeability

     
 

7 77 < 2. 2 5 I
it ’Not reported:
Not reported

  

Qolor " ' ' '- 5 2': WHITE7 WHITEET
Flexural Strength (MOR) 20°C MPa (psi X 10:) 352 (51) 572 (83)
Elastic Modulus ' 5' 20°C GPa (psi X 106 ) : 7303(44) 372 (54) :
Poisson‘s Ratio 20°C 0.21 Not reported
Compressive Strength 20°C ' MPa (psi X 103) 21 03 (305) 5 Not reported:
Hardness GPa (kg/mm2 ) 11. 5 (1175) Not reported
Jensile Strength 25°C MPa (psi X 103) 5193’ (28) Notreported
Fracture toughness K00) MPa m” 4-5 Not reported
Thermal Conductivity 20°C W/m K7 22.4 7 7 33.5 ’:
Coefficient of Thermal 25— -3 0

Expansion_ 7‘ 7_ 10097°C77 7:10 / C 77 3'27 777 7 6'9
51317971731179 Efeat 5 5., EZTETEOT‘ «ti/[KL 2751f}: fluéfitjlfi
Thermal Shock Resistance 7 7 179°C 7 Not reported
5312(1me USE Terftberafli’refg :22» '

 

ac- kV/ mm (ac 8.3 (210) Not reported

Dielectric Strength 77777 77777777 7 7 7m/il 7

0.0004 Not reported

5 i _. H if chm- 5 ' Not re aorted

 

45

           

Figure 2.2.1: Isomet Low-speed diamodawsw
specimen holder, C) Lever Arm, D) U.S. Quarter

DIoam blade, B) Rotating

46

The fixture on the low speed saw, which holds the billets (Figure 2.2.1 ), is rotated
so that the contact angle between the saw blade and the billet is changed
periodically. Following the initial cutting, the alumina was further sectioned to
obtain 0.5 cm X 1.3 cm X 0.11 cm specimens. The partitioning process took
about 15 to 20 minutes per cut.

Coors AD-94 polycrystalline alumina specimens were sectioned from rod
stock (9.1 mm diameter X 100 mm long). Coors AD-94 rod stock is easily
sectioned in 10 to 20 minutes to obtain 9.1 mm diameter X 1.6 mm long

specimens.

2.3 Mounting Procedure

After polishing, the cut specimens were mounted on a disk-shaped
aluminum plate (19.44 cm diameter X 1.93 cm thick). Two removable 1.23 cm
diameter pegs, roughly 4.1 cm from the aluminum disk’s center, attach to the
plate to allow the motorized arm, AP-50 (Leco Corporation), to rotate the plate on
the polishing wheel. The aluminum plate was heated on a Cole-Parmer Hot
Plate, Model 4658 (Cole-Parmer Company: Chicago, Illinois), until the mounting
Thermoplastic Cement (Buehler LTD) melted. Four to six specimens were
symmetrically secured in the thermoplastic cement about 2.6 cm from the edge
of the aluminum plate. The aluminum plate was then removed from the hot plate,
and placed onto a refractory firebrick (Thermal Ceramics, Taylor, Michigan), to
allow the aluminum plate to cool. The maximum service temperature for the

refractory firebrick is 1200°C. The thermoplastic was still viscous at the time the

47

hot plate was turned off, thus the specimens were repositioned if a shift in
position occurred during cooling.

The steps in the mounting process were reversed to remove the
specimens from the aluminum disk. The polished specimens were then washed
in a 50 ml glass beaker, containing about 20 to 30 ml of acetone, to remove the
thermoplastic. The beaker was placed into an Ultramet lll Sonic Cleaner
(Buehler LTD), with the bath containing de-ionized water to a level about a
centimeter above that of the acetone in the beaker. The specimen was
ultrasonically cleaned for 30 to 45 minutes in a fume hood. Specimens were
then rinsed, for ten seconds with additional de-ionized water and dried with a

Kimwipe EX-L (Kimberly-Clark Corporation: Roswell, Georgia).

2.4 Polishing Procedure

The specimens were polished on a VARl/POL VP-50 (Leco Corporation:
St. Joseph, MI) automatic polisher. The rotary wheel speed ranged from 125-
175 rpm while the positioning arm, AP-50, moved the aluminum plate at a preset
speed. The positioning arm repeatedly moves back and forth along a preset
path. Lower rotary speeds, 75 —150 rpm, were used during initial polishing, so
that there was less gouging of the Technotron White (Leco Corporation) polishing
cloths.

The Coors ADS-995 specimens were first polished with 25 um diamond
paste (Leco Corporation or Warren Diamond Powder Company: Olyphant,

Pennsylvania) for 2 hours. The specimens were then inspected for flat surfaces

48

before continuing to the next grade of polishing media. After polishing at the 25
um grit, large scratches indicated additional polishing was required. During the
polishing process, the surface visibly changes as the laps are changed to polish
the specimen with diamond paste having successively finer grit size. The
surface qualities of each of the specimens were inspected after each polishing
step, with an emphasis on decreasing the macro surface flaws until a near
mirror-like surface finish was achieved. Alterations in the surface finish at the
specimen edges and along the surface are the most obvious indicators that it
was time to progress to a finer grit size. When a specimen’s entire surface has a
uniform luster, the grit size of the polishing media was changed. The specimens
were polished for equal amounts of time at abrasive grit sizes: 17, 15, 10, and 6
pm. Final polishing at 1pm lasted longer. Coors ADS-995 specimens were
polished for roughly two hours at each stage with a final polishing time of 3 hours
at 1 pm.

After final sectioning, there were slight differences in height for the Coors
AD-94 specimens. Therefore, polishing at 25 pm for 5-20 minutes was
performed in order to obtain a uniform specimen surface. AD-94 specimens

were then polished for 20 minutes at each intermediate (17, 15, 10, and 6 pm

polishing media) stage with a final polishing time of 40 minutes at 1 pm.

2.5 Procedure for the Loading and Unloading of the Digestion Bomb [32]
In general, a digestion chamber is used to attack a solid with an aqueous

chemical solution, the aqueous solution thereby “digests” the solid. Early

49

digestion chambers were simple containers with a hollow airtight cavity. The
digestion chamber would be placed into a furnace and whatever reaction
between the solution and the solid that was going to take place would occur. The
internal pressures were sometimes such that the digestion chamber exploded,
hence the name: digestion bomb.

The digestion bomb, Parr Model 4746 (Parr Instrument Company), has a
maximum sample volume of 23 ml and a maximum specimen mass of 1 gram
(Figures: 251-3). The specified specimen mass for the digestion bomb
assumes that the specimen is a powder. However, the specimens used in this
study are sintered compacts, so the reactivity of our specimens is lower than the
reactivity of a powder specimen due to a lower surface area of the sintered
ceramic. Thus, a specimen mass, larger than 1 gram was safe to use in this
study. The maximum operating temperature for the digestion bomb is 275°C with

a corresponding pressure of ~5.94 MPa [32].

50

 

b
Figure 2.5.1: Side view of the Parr Digestion Bomb inner part, a) Top View, b)
Side View, c) Bottom View: A) Stainless bomb body; B) Teflon Cup; C) Corrosion
disc (0.05 mm thick); D) Rupture disc (0.254 mm thick); E) Teflon cover; G) U.S.
quanen

51

 

a

Figure 2.5.2: Bottom view of the Parr Digestion Bomb outer parts, a) Bottom
View, b) Side View, 0) Top View: A) Bronze plated screw cap; B) Stainless steel
pressure plate; C) Spring; D) Nickel bottom plug; E) Stainless steel top cap; G)
U.S. quarter.

52

 

Figure 2.5.2: (Continued)

53

Nine milliliters of room temperature deionized water were placed within the
Teflon chamber of the digestion bomb. Using forceps (11.7 cm steel), the
specimen was positioned within the Teflon chamber such that the polished side
of the specimen was placed in contact with the de-ionized water. The chamber
was then closed and sealed with a “spanner-jack” (Figure 2.5.7).

After each hydrothermal treatment run was complete, the sample was
removed from the Teflon chamber with forceps and then set on aluminum foil to
dry. Aluminum foil is very soft and does not scratch the surface of the
hydrothermally attacked specimens. Additionally, the Kimwipe EX-L (Kimberly-
Clark Corporation) leaves microscopic fibers or debris on the specimens, which
adversely affect the microscopy.

The Teflon chamber was emptied and blown dry with pressurized air
(~345 kPa), which runs throughout the Engineering Building (Michigan State
University, East Lansing, MI 48824) and comes out of the laboratory wall (3531
Engineering Building) via a valve. When the specimen was entirely dry, the
specimen was then prepared for SEM or AFM examination. The specimen was

then air cleaned from the buildings supply of air and wrapped in aluminum foil.

54

 

Figure 2.5.3: A) Spanner-Jack, used to tighten down the top and bottom portions
of the digestion bomb; B) U.S. quarter.

55

2.6 Digestion Bomb Furnace Procedure

To minimize the thermal degradation of the digestion bomb (Parr
Instrument Company), and to therefore increase the service life of the apparatus,
the bomb was utilized at 250°C for this experiment. The vertical tube furnace
used to heat the digestion bomb, takes about 1.75 hours to obtain thermal
equilibrium for a test temperature of 250°C. The vertical tube of the furnace is
composed of two semi-circular (15.25 cm inner diameter X 37 cm height) nickel-
chrome heating elements (C.M. Incorporated, Bloomfield, New Jersey) pieced
together in order to create a cylindrical, or tubular, heating cavity. Refractory
bricks were stacked around the vertical furnace tube to serve as thermal
insulation. The refractory bricks in contact with the tube were machined in order
to conform to the shape of the tube.

An Omega Model CN50011C2 controller (Omega Engineering
Incorporated: Stamford, Connecticut), with a K-type thermocouple was used to
determine the temperature. A recess was machined with a mill, within an excess
refractory brick Figure 2.6.1. The dimensions of the first recess are given in
Figure 2.6.1 and there are multiple recesses so that if the digestion was over-
pressurized and vented, the decompression would not be harmful to the furnace
chamber, such was the case when the digestion bomb did vent. The brick was
then placed at the bottom of the furnace tube. Before beginning an experimental
run the digestion bomb was placed in the milled recess of the brick. Two
additional refractory bricks have been machined with a saw and sand paper, so

that each brick covered half of the opening of the tube furnace (Figure 2624).

56

 

Figure 2.6.2: Bottom view of upper refractory brick A) 22.75 cm X 11.35 cm X
2.73 cm tall; B) Inner raised semi-cylinder 14.75 cm diameter X 3.7 cm tall; C)
Through-hole for thermocouple D) U.S. quarter.

57

 

Figure 2.6.3: Side view of upper refractory brick A) 22.75 cm X 11.35 cm X 2.73
cm tall; B) Inner raised semi-cylinder 14.75 cm diameter X 3.7 cm tall; C) U.S.
quanen

 

Figure 2.6.4: Reverse side view of upper refractory brick A) 22.75 cm X 11.35
cm X 2.73 cm tall; B) Inner raised semi-cylinder 14.75 cm diameter X 3.7 cm tall;
C) U.S. quarter.

58

The specimen was loaded into the digestion bomb (according to Section
2.6 Digestion Bomb Procedure: Loading and Unloading) and then the digestion
bomb was placed inside the cylindrical depression of the supporting refractory
brick within the furnace (Figure 2.7.1 ). Then the upper refractory bricks (Figures
2.624) were put in place. The furnace was set to the test temperature of 250°C.
When the furnace reached thermal equilibrium, the test began. After the desired
isotheral hold was accomplished, the furnace was allowed the cool. Once the
furnace temperature reached 20°C, the digestion bomb was removed and the
specimen was unloaded from the digestion bomb. On completion of each
hydrothermal test, the assembly had loosened somewhat due to the internal
pressures (~3.97 MPa) of the experiment. The loosening had no apparent effect
on the specimen or the testing and was simply a byproduct of the pressures

expenenced.

2.7 Atomic Force Microscope Examination [33]

A Nanoscope Ill Scanning Probe Microscope (Digital Instrument
Corporation, Santa Barbara, California) was used to analyze the thermally and
hydrothermally attacked specimens. Prior to mounting, the attacked specimens
were cleaned with acetone as described in the mounting section (2.3). The
specimens were then affixed to a stainless steel disk (1.18 cm diameter X 0.8
cm) via a single adhesive tab (Ted Pella Inc., Redding, California). The steel
disk with the specimen affixed was then placed on the magnetic stage of the

Nanoscope III. For low magnification analysis, the specimen was placed under

59

an atomic force microscopy (AFM) J-type module, high magnification analysis
was performed under a D-type module; both modules were designed by the
Digital Instrument Corporation. The maximum scan size for the J-type module is
125 um X 125 um, and the maximum scan size for the D-type module is 12 um X
12 pm. The Digital Instruments silicon nitride model NP tip had a cantilever

length of 200 um and a nominal spring constant of ~0.12 N/rn. Although,
“Calculated spring constant values are based on the 0. 6pm silicon nitride
thickness; however, this value can actually vary from 0.4pm to 0. 7pm. Thickness
is cubed in the spring constant calculation, thus, actual values can vary
substantially [34].”

The scan sizes of interest ranged from 5 to 20 um. Due to the nature of
scanning probe microscopy, a smaller scan size requires a faster scan rate,
however if the scanning rate is excessive the resolution of the corresponding
image may suffer. When the scan rate is too low, there is a higher likelihood of
instrument noise being present in the resulting image. Depending on the noise
experienced for a given specimen the scan rate was varied from as low as 2 Hz
to as high as 24 Hz.

In general, a scanning height of 200 nm with 256 or 512 samples per line
was chosen. A lower number of samples per line may be chosen, however an
intermediate or higher sample rate is not possible. The samples per line varied
because at 512 samples per line, noise is more apparent than at 256 samples
per line even if identical cleaning methods are used on both specimens. Thus in

some instances there was too much noise to obtain a micrograph through

60

changing the scan rate.

61

 

\ 35' I35“ /
\ l r

V ,5

r” 35.

n... "3;

 

Cantilever

 

 

 

 

Figure 2.7.1: Silicon Nitride Probe Schematics [34].

5 kV
18 mm working distance 17 mm working distance

 

(a) (b)
Figure 2.7.2: Micrographs of the silicon nitride probe a) 800x b) 5000X [34].

62

Chapter 3: Results & Discussion
3.1 Hydrothermal Attack

Specimens were polished to a 1 pm finish so that all specimens would
have a similar surface finish (Figures 3.1.1 and 3.1.2). The samples were then
placed in the digestion bomb and an experimental run was performed. Each
micrograph represents a separate specimen and not a specimen that has been
cyclically heated.

A scale develops on the specimen surface during hydrothermal
treatments. The scale is characterized in terms of the dimensions parallel to the
surface (in-plane) and the direction normal to the surface (peak directions). Peak
heights within the AFM micrographs were determined by comparing the relative
height of the individual peaks to the scale on the “z” axis. As a result, all of the
height measurements are approximations.

Noise in the AFM is apparent by either no image being recorded, wavy
lines in the image, or other discontinuities in the image. Noise may be attributed
to a bad AFM tip or an uneven specimen surface.

Figure 3.1.1a-b depicts the surface of an ADS-995 specimen that has not
been hydrothermally attacked. Some specks of dust (<2 urn across) are visible
on the surface. AFM allows the imaging of objects that are too small to be
resolved through conventional SEM. At that scale, there is difficulty in cleaning
and maintaining the surface of the specimen within the confines of the atomic

force microscope as the apparatus rests in ambient air, at ambient temperature.

63

Additionally, there are no apparent inclusions or surface defects (Figure 3.1.1a-
b).

Figure 3.1.2a-b displays an untreated specimen of AD-94. There are
many scratches on the surface from polishing. Additionally, a surface depression
was missed by the surface polish.

After 30 minutes of hydrothermal corrosion (Figure 3.1.3), the ADS-995
shows minimal surface attack. The arrow indicates the region that has been
preferentially attacked, most likely to be a grain boundary triple point (Figures
3.1.3). The signs of attack are apparent from the increased pitting of the surface.

The surface of AD—94, Figure 3.1.4, that has been hydrothermally
corroded for 30 minutes still has polishing relics along with what appears to be
either dust or precipitates on the surface.

After 1 hour of exposure, the arrows point to pitting that appears on the
surface of the ADS-995 specimen (Figure 3.1.5). There is still some dust on the
specimen surface.

Following an hour of attack on AD-94 (Figure 3.1.6), there is visible pitting
of the surface and the grain boundary triple points appear to have been
preferentially attacked. The distance between the pitted regions corresponds
roughly to the grain size of 12 um specified by the vendor (Coors Ceramics
Company: Golden, Colorado). The surface also displays various particles of dust

and faint polishing scratches.

 

L
r

l I

a
Figure 3.1.1: AFM micrograph of Coors ADS-995, with no hydrothermal
treatment, polished to a 1 pm finish, 5.000 um scan size, 4.069 H2 scan rate, 256
samples; a) surface view - z-range: 1 urn/division, b) top view.

65

 

Figure 3.1.1 : Continued.

 

I
I l l

a
Figure 3.1.2: AFM micrograph of Coors AD-94, with no hydrothermal treatment,
polished to a 1 um finish, 9.667 urn scan size, 15.26 Hz scan rate, 512 samples;
a) surface view — z-range: 1 urn/division, b) top view.

67

 

Figure 3.1 .2: Continued.

68

 

5:
l l l

a
Figure 3.1.3: AFM micrograph of Coors ADS-995, hydrothermally attacked for 30
minutes at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a)
surface view — z-range: 1 uni/division, b) top view.

69

 

Figure 3.1.3: Continued.

70

 

l
l I l

a
Figure 3.1.4: AFM micrograph of Coors AD-94, hydrothermally attacked for 30
minutes at 250°C, 5.000 um scan size, 6.104 H2 scan rate, 512 samples; a)
surface view - z-range: 1 urn/division, b) top view.

71

 

Figure 3.1.4: Continued.

72

I
1

 

I
l I I

a
Figure 3.1.5: AFM micrograph of Coors ADS-995, hydrothermally attacked for 1
hour at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a) surface
view - z-range: 1 rim/division, b) top view.

73

 

Figure 3.1.5: Continued.

74

I
:

 

l
I I I

a
Figure 3.1.6: AFM micrograph of Coors AD-94, hydrothermally attacked for 1
hour at 250°C, 20.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 rim/division, b) top view.

75

 

Figure 3.1.6: Continued.

76

After 1.5 hours of hydrothermal corrosion, significant pitting occurs for AD-
94. The in-plane dimensions of the pits are in the range of 0.75 microns, and
surface growths are present that are 0.5 microns high (Figure 3.1.7). The grain
boundaries are not visible except the grain boundary triple points are likely in the
same locations as the pits.

Finally, in Figure 3.1.8 (2 hours of corrosion) severe pitting (> 0.5 pm) has
outlined the individual grains as the grain boundaries have been partially
dissolved. There are no obvious polishing artifacts although the arrow indicates
an AFM relic (Figure 3.1.8b) along with some dust on the surface. The micron
sized circular to rhomboid shaped structures on the surface are the early stages
of a scale being formed on the surface, this scale will become more prevalent
and apparent in later stages of corrosion.

Following 2.5 hours of hydrothermal corrosion, Figure 3.1.9, a large
percentage (~45%) of the specimen is obscured by a surface layer. The
individual “leafs” of the scale range from 0.25-2 um in-plane lengths and readily

form on each other.

77

10

 

 

I l
l T I

a
Figure 3.1.7: AFM micrograph of Coors AD-94, hydrothermally attacked for 1.5
hours at 250°C, 20.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 um /division, b) top view.

78

 

Figure 3.1.7: Continued.

79

I
:.

    

I I
I T I

a
Figure 3.1.8: AFM micrograph of Coors AD-94, hydrothermally attacked for 2
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 pm /division, b) top view.

80

 

Figure 3.1.8: Continued.

81

 

“\

l
I

IN
I

a
Figure 3.1.9: AFM micrograph of Coors AD-94, hydrothermally attacked for 2.5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 pm /division, b) top view.

82

 

Figure 3.1 .9: Continued.

83

Following three hours of exposure the ADS-995 displays some pitting on
the surface (Figure 3.1.10). In Figure 3.1.103, the grain structure is barely
visible, along with polishing scratches and some specks of dust on the surface.
Grain triple points as associated with roughly triangularly shaped depressions on
the specimen surface. The grains have different crystallographic orientations so
they appear to have different degrees of contrast under the AFM tip, since the
contrast of the AFM image is a function of the electric potential of the surface,
which in turn depends on crystallographic orientation.

AD-94 corroded for 3 hours, yields a more severely damage structure
(Figure 3.1.11). Approximately 98% of the specimen surface is covered in the
surface layer. What appears to be the original surface (as indicated by the
arrows Figure 3.1.11b) may be areas that are relatively flat in comparison to the
other areas of surface scale and not the original specimen surface. At this point,
the specimen is no longer experiencing only hydrothermal grooving, but the
progressive accumulation of a surface layer. Note the number of peaks, in
relation to Figure 3.1.10, within the surface layer indicating a peak growth height
~0.5 um accompanied by valleys of the same scale. However, the grooves that
occur during the hydrothermal process are covered by the surface layer so the
actual depth of the grooves is difficult to determine. Additionally, there are some

artifacts from the AFM on the substrate surface.

84

 

I
l I l

a
Figure 3.1.10: AFM micrograph of Coors ADS-995, hydrothermally attacked for 3
hours at 250°C, 10.00 pm scan size, 9.766 H2 scan rate, 512 samples; a)
surface view - z-range: 1 pm /division, b) top view.

85

 

Figure 3.1.10: Continued.

86

 

 

a
Figure 3.1.1 1: AFM micrograph of Coors AD-94, hydrothermally attacked for 3
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 pm /division, b) top view.

87

 

Figure 3.1.11: Continued.

88

After 3.5 hours of corrosion, the AD-94.has a few AFM discontinuities, yet
the surface is still completely occluded by the surface layer (Figure 3.1.12).
There are some smooth areas as in Figure 3.1.11, however, the size of the
growths is much larger, 1-3 pm in-plane lengths, with an average peak height of
~0.5 pm. The growths are apparently developing as a series parallelograms
growing on each other, with the majority of these parallelograms being square or
rectangular. Day [Day, 1982], noted that for 96% pure alumina, a cubic wairakite
structure developed on the surface after less than 10 days of hydrothermal
treatment at 250°C (Section 1.5).

After four hours of hydrothermal attack an AD-94 specimen lacked pitting
or the signs of corrosion (Figure3.1.13). Grain boundary etching is mostly absent
except for a single grain boundary triple point alongside polishing marks.
However, there were difficulties acquiring AFM micrographs of AD-94 after 3
hours of corrosion: the surfaces have been degraded so severely that there is
considerable noise when the surfaces were being resolved. In particular, the 4-
hour AD-94 specimen was very difficult to image in the AFM. Therefore, the
surface shown is probably not representative of the majority of the surface of the

AD-94 specimen at this level of corrosion.

89

 

 

a
Figure 3.1.12: AFM micrograph of Coors AD-94, hydrothermally attacked for 3.5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

 

Figure 3.1.12: Continued.

91

l
I I 1

 

a
Figure 3.1.13: AFM micrograph of Coors AD-94, hydrothermally attacked for 4
hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a) surface
view - z-range: 1 uni/division, b) top view.

92

 

Figure 3.1.13: Continued.

93

AD-94 that has been hydrothermally degraded for 4.5 hours (Figure
3.1.14), again displays damage similar to that seen in Figures 3.1.11-12. Some
smooth portions of the surface visible (Figure 3.1.14b) yet they are not believed
to be the AD-94 surface when compared to Figure 3.1.14a. The surface layers
peak heights reach 1 pm with troughs on the order of ~0.6 urn deep. The in-
plane lengths vary from 04-3 pm, however the majority of the scales have an in-
plane length above 1.5 pm.

Figure 3.1.15ab portrays polishing scratches, dust, and possible scale
developing on the surface of ADS-995 with 5 hours of exposure. However, the
grains themselves are not readily apparent in this micrograph. There also
appears to be a few (~4) grains that erupt from the surface of the specimen.

A 5-hour trial of AD-94, continues to display complete coverage by the
surface layer with an average peak height of ~1 um (Figure 3.1.16). Figure
3.1.14 is mostly composed of “tall” peaks in comparison to the peak heights in
Figure 3.1.16, which is due to the fact, that the AFM could not resolve an image

in a more disrupted area of the specimen surface.

94

 

l l
I I l

a
Figure 3.1.14: AFM micrograph of Coors AD-94, hydrothermally attacked for 4.5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

95

 

Figure 3.1.14: Continued.

96

 

I .
I I I

a
Figure 3.1.15: AFM micrograph of Coors ADS-995, hydrothermally attacked for 5
hours at 250°C, 10.00 pm scan size, 4.883 Hz scan rate, 512 samples; a)
surface view - z-range: 1 urn/division, b) top view.

97

 

Figure 3.1.15: Continued.

98

 

 

I I l

a
Figure 3.1 .16: AFM micrograph of Coors AD-94, hydrothermally attacked for 5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

99

 

Figure 3.1.16: Continued.

For AD-94 hydrothermally treated for 5.5 hours, AFM distortion could not
be avoided at this or more severe levels of attack as the steepness of the
surrounding areas of the specimen affected the movement of the tip on the
specimen surface (Figure 3.1.17). Notably, the entire surface is still covered in a
film, primarily composed of 1 pm basal scales. There are occasional large grains
(2 x 3 x 1 urn tail) that exceed the average layer level of the specimen. Quite
likely, wairakite-type cubes are forming on the surface and the micrograph is of
an area that rests between the structures.

The 6-hour hydrothermal attack of AD-94 (Figure 3.1.18), has distortions
within the image from the AFM. The average surface layer height is low, as the
surface layer is forming into possibly wairakite-type structures. The surface
pitting is present at the grain boundary triple points.

After 6.5 hours of corrosion displays AD-94 with a mostly low (<0.1 um)
surface layer height (Figure 3.6.19). Also present are structures (~7 x 5 pm with
a height of ~1 pm). There are conglomerations of numerous rectangular scales.

Following 7 hours of hydrothermal attack (Figure 3.1.20), the AD-94
surface is completely coated by a surface layer that contains structures (~1 um
tall) with significant lateral (~5 x 9 pm) dimensions. The large pit present is over
1 pm deep and 2 pm wide may correspond to a grain boundary triple point.

At 7.5 hours of continuous attack the AD-94 surface is revealed that is
completely covered by the surface layer (Figure 3.1.21). The surface layer

reached 1.3 pm in height and has a triple point pit roughly ~1.5 um deep.

101

”M

 

 

l I
I f I

a
Figure 3.1.17: AFM micrograph of Coors AD-94, hydrothermally attacked for 5.5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

102

 

Figure 3.1.17: Continued.

103

 

 

a
Flgure 3.1.18: AFM micrograph of Coors AD-94, hydrothermally attacked for 6
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

 

Figure 3.1.18: Continued.

105

  

L
I

v
i l

a
Figure 3.1.19: AFM micrograph of Coors AD-94, hydrothermally attacked for 6.5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

106

 

Figure 3.1.19: Continued.

107

 

 

I l l

a
Figure 3.1.20: AFM micrograph of Coors AD-94, hydrothermally attacked for 7
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 uni/division, b) top view.

108

 

Figure 3.1.20: Continued.

109

 

l
l I

a
Figure 3.1 .21: AFM micrograph of Coors AD-94, hydrothermally attacked for 7.5
hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a) surface
view - z-range: 1 um/division, b) top view.

110

 

Figure 3.1 .21 : Continued.

lll

Subsequent to 10 hours of hydrothermal attack, the grain structure of the
ADS-995 can be seen along with scratches due to polishing (Figure 3.1.22).
There is some visible pitting (roughly ten ~0.2 um pits per 100 umz), and a small
degree of dust (<10 specks) on the surface. The pits occur mostly at the grain
boundary intersections, or triple points, are roughly triangular depressions and in
this image that are less than 0.25 micron in their in-plane directions.

An AD-94 specimen having experienced 10-hour hydrothermal corrosion
displays complete coverage by the surface layer by a uniform distribution of ~1
um in-plane scales (Figure 3.1.23). The scales have an average height of ~0.4
um, and there are a few pits on the surface.

Figure 3.1.24 depicts the ADS-995 surface exposed to 12.5 hours of
treatment. The grain boundaries and pits may be seen in this micrograph
alongside some specks (<10) of dust and a few grains as indicated by arrows
(Figure 3.1.24a) that appear to be erupting from the surface of the specimen.
However, when compared to the morphology of the AD-94 surface scale what
appears to be an emerging grain may in fact be a precursor of that scale. Such
may be the case as the ADS-995 has a higher purity and the scaling reaction
should take longer to begin as well as progress. In order for a grain to erupt from
the surface, either all the surrounding material would have to be removed or the
grain would have to begin to grow, neither of these scenarios is likely however
grains can still “appear" to emerge by having their crystallographic orientation

being such that their surfaces appear to “come out" more from the bulk.

112

 

l
r I l

a
Figure 3.1 .22: AFM micrograph of Coors ADS-995, hydrothermally attacked for
10 hours at 250°C, 10.00 pm scan size, 6.104 H2 scan rate, 256 samples; a)
surface view - z-range: 1 urn/division, b) 10p view.

113

 

Figure 3.1 .22: Continued.

114

”fl

 

l L
1 1

1

a
Figure 3.1.23: AFM micrograph of Coors AD-94, hydrothermally attacked for 10
hours at 250°C, 10.00 pm scan size, 9.766 Hz scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

115

 

Figure 3.1 .23: Continued.

116

I
:

  
   

|
I l j

a
Figure 3.1.24: AFM micrograph of Coors ADS-995, hydrothermally attacked for
12.5 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view - z-range: 1 urn/division, b) top view.

117

 

Figure 3.1.24: Continued.

118

Individual grains of the ADS-995 are readily apparent as shown within the
outlined areas A and B, in Figure 3.1.25 after 15 hours of corrosive attack. Dust
is not too apparent in this figure. Here the pitting at the grain boundaries is the
most discernable when compared with any of the other ADS-995 figures. The
mean grain size is approximately 2 pm (Table 2.2.1) with some grains larger than
2 pm and others smaller than 2 pm. Unfortunately, there are still some polishing
scratches on the surface. It is expected that polishing artifacts be visible on the
surface of most specimens that have not been completely occluded by a surface
scale. The polishing was only conducted to 1 pm and at the magnification of the
AFM, scratches in the range present (fractions of microns across) should be
visible.

ADS-995 after 20 hours of exposure has a highly corroded surface
morphology displaying surface scale and some pitting. The scale covers the
entirety of the surface and hinders viewing of the surface (Figure 3.1.26). There
are no apparent polishing marks.

In Figure 3.1.27, ADS-995 following 25 hours of hydrothermal corrosion,
the surface remains mostly covered with scale. The corrosion of the surface
appears to be less severe than that of the 20-hour specimen. Thus, either the
surface was not completely covered or the surface scale was not in a particularly

stable state.

119

I
a.

 

h i l

a
Figure 3.1.25: AFM micrograph of Coors ADS-995, hydrothermally attacked for
15 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view — z-range: 1 uni/division, A) grain ~2um in size, B) grain larger than
2 pm, b) top view, 0) top view without lines.

120

 

 

 

 

 

 

Figure 3.1.25: Continued.

121

 

 

Figure 3.1.25: Continued

122

 

L

I I I

a
Figure 3.1.26: AFM micrograph of Coors ADS-995, hydrothermally attacked for
20 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a)
surface view - z-range: 1 urn/division, b) top view.

123

 

Figure 3.1 .26: Continued.

124

'lJll

 

 

a
Figure 3.1.27: AFM micrograph of Coors ADS-995, hydrothermally attacked for
25 hours at 250°C, 10.00 pm scan size, 12.21 Hz scan rate, 512 samples; a)
surface view — z-range: 1 urn/division, b) top view.

125

 

Figure 3.1.27: Continued.

126

Figure 3.1.28 shows the ADS-995 surface after 30 hours of attack. Scale
deposits are present that are approximately 1.5 pm in their in-plane directions
(Figure 3.1.28ba). In the top view of the ADS-995 (Figure 3.1.28b) that has been
treated for 30 hours, there are a few visible artifacts from the AFM. However, the
surface has apparently been completely covered by the growth or surface layer
as seen in the surface view (Figure 3.1.28a).

The surface of the ADS-995 remains coated with the surface layer after 35
hours of hydrothermal exposure (Figure 3.1.29). However, the specimens
surface topology is smoother than that shown in Figure 3.1.28. Fortunately,
there are no visible scratches from the polishing. The scale deposits in this
micrograph are nearly a micron in height and ~6 pm in the in-plane direction.

A large number of growths and possibly dust are present on the surface of
the 40-hour hydrothermally attacked specimen for ADS-995 (Figure 3.1.30). Few
polishing marks are visible on the surface. The topology of the surface layer has
notably changed from the surface layers revealed in previous ADS-995
micrographs (Figures 3.1.26,27,28,29) in that the grain boundaries are again
visible on the surface.

ADS-995 after 45 hours of hydrothermal treatment (Figure 3.1.31) reveals
pits that are again apparent on the surface along with polishing marks and dust.

Again the surface layer has not deposited on the alumina surface.

127

      
 

l
I I I

a
Figure 3.1.28: AFM micrograph of Coors ADS-995, hydrothermally attacked for
30 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view - z-range: 1 um/division, b) top view.

128

 

Figure 3.1.28: Continued.

129

 

l l
I l |

a
Figure 3.1.29: AFM micrograph of Coors ADS-995, hydrothermally attacked for
35 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view - z-range: 1 urn/division, b) top view.

130

 

Figure 3.1.29: Continued.

131

”ll

 

l
l l l

a
Figure 3.1.30: AFM micrograph of Coors ADS-995, hydrothermally attacked for
40 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view - z-range: 1 um/division, b) top view.

132

 

Figure 3.1.30: Continued.

133

 

 

 

 

 

 

a
Figure 3.1 .31: AFM micrograph of Coors ADS-995, hydrothermally attacked for
45 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view - z-range: 1 urn/division, b) top view.

134

 

Figure 3.1.31: Continued.

135

The surface of ADS-995 hydrothermally corroded for 50 hours (Figure
3.1.32) has a surface devoid of any surface layer. Grain boundaries and
polishing artifacts are present on the surface alongside pits and dust akin to
Figure 3.1.22. This supports the theory that the surface layer is deposited during
the 12.5-35 hour regime, and does not deposit during the 40-50 hour regime.

Conversely, for AD-94 following 50hours of hydrothermal treatment, the
specimen surface is completely occluded by the surface layer (Figure 3.1.33).
The largest structure present is pyramidal in shape and is more than 1 pm in
height (Figure 3.1.33b).

Surface layers may precipitate during the cooling cycle of the treatment.
impurities in the grain boundary phase, or those impurities and water, combine to
form the scale that appears on the specimen surface. The thermocouple is
turned off during the cooling cycle of the furnace ands there is no thermocouple
within the confines of the digestion bomb, thus reactions that would form
isotherms in the cooling rate are neither recordable nor observable. Due to
specimen size, isotherms are unlikely to be recordable with a thermocouple as it
would be difficult to place a thermocouple on the specimen surface during
experimentation. However, a thermocouple placed directly in contact with the
digestion bomb may detect that a reaction was taking place. If the reaction is
time dependent, which it appears to be, there would be difficulty in determining a
change in the temperature as a function of time since the furnace experiences

temperature fluctuations.

136

I
:1

 

a
Figure 3.1.32: AFM micrograph of Coors ADS-995, hydrothermally attacked for
50 hours at 250°C, 10.00 pm scan size, 16.28 Hz scan rate, 512 samples; a)
surface view - z-range: 1 ILITl/dIVISIOi'l, b) top view.

137

 

Figure 3.1 .32: Continued.

138

 

I;
I I I

a
Figure 3.1.33: AFM micrograph of Coors AD-94, hydrothermally attacked for 50
hours at 250°C, 10.00 pm scan size, 9.766 H2 scan rate, 512 samples; a) surface
view - z-range: 1 urn/division, b) top view.

139

 

Figure 3.1 .33: Continued.

3.2 Roughness Analysis

An analysis was made of the general surface roughness of each specimen
(Figures 3.2). The roughness data was taken as a measure of the number of
events that occur within a series of depths within an AFM micrograph. The
roughness data was examined as histograms where the full range of depth is
portrayed in 1000 steps. The data was collected and the “zero” point of the
roughness data is a reference point only within a given AFM micrograph. The
number of events was normalized with respect to the total number of events.
The normalized number of events was then plotted with respect to the
topographic depth.

A quantitative analysis of the roughness data is beyond the scope of this
study. However, the roughness data supports the AFM micrograph data that the
degree of corrosion increases for both alumina specimens as the exposure time
increases (Figures 3.2).

Additionally, the shape of the AD-94 histogram curve changes as the
corrosion progresses (Figure 3.2.4,8) indicating that there are multiple regimes
where the depth of the corrosion is uniform. These distortions from the ideal
curve may represent of grain boundaries, pits, scratches and other defects
marring the surface. Due to the asymmetry of the curves, the normal
distributions are believed to be the summations of multiple gaussians to

accurately describe the surface roughness of the specimens.

141

Both AD-94 and ADS-995 specimens displayed a broadening of the
normalization curves as corrosion progresses. Indicating that the longer a

specimen is hydrothermally attacked the more damage is done to the surface.

142

ADS-995 Normalized Event Histogram - 0 hours

 

 

 

0.09 -
. 0.08 -

O
.3 : 0.07 -
% . 0.06 -
'5 ° _
E °. 0.04 -
g 3 0.03 -
2 .: 0.02 ~
12 0.01 J

-60 -40 -20 0 20 40 60

Topographic Depth (nm)

Figure 3.2.1: Normalized roughness analysis as a function of the topographic

depth for Coors ADS-995, with no hydrothermal treatment, polished to a 1 pm
finish.

AD—94 Normalized Event Histogram - 0 hours

 

 

 

0.025 —
g 0.02 -
0 I
5 ' 0 015
.2 3. '
8 1‘
.5 ‘ i 0 01 -
a F ‘
E ' 1
O I I
Z I i 0.005 4
-25 -20 -15 -10 -5 0 5 10 15 20

Topographic Depth (nm)

Figure 3.2.2: Normalized roughness analysis as a function of the topographic
depth for Coors AD-94, with no hydrothermal treatment, polished to a 1 pm finish.

143

ADS-995 Normalized Event Histogram - 5 hours

0.035 -
0.03 a

0.025 -

O

o 0"“.

O

0.02 z

00

0.015 ~

«00000

0.01 .

Normalized # of events

0.005 ‘

 

 

F r i l

-200 -150 -100 -50 0 50 100 150 200
Topographic Depth (nm)

‘
q

I

Figure 3.2.3: Normalized roughness analysis as a function of the topographic
depth for the of Coors ADS-995, hydrothermally attacked for 5 hours at 250°C,

AD-94 Normalized Event Histogram - 5 hours

0.0045 1

- 0.004 -
f. 00035 -
0.003 ~

0.0025 -

0.002 -
0.0015 ~
0.001 -

  
  
  
  
 

Normalized # of events

  

 

 

-600 -400 -200 0 200 400 600
Topographic Depth (nm)

Figure 3.2.4: Normalized roughness analysis as a function of the topographic
depth for Coors AD-94, hydrothermally attacked for 5 hours at 250°C.

144

ADS-995 Normalized Event Histogram - 10 hours
0.012 -
g.‘ .
l
1, 0.008 .
f
“ 0.006 -
i
l
'.

0.004 -

0.002 5
l r l _ r l i

~150 -100 -50 0 50 100 150 200
Topographic Depth (nm)

Normalized # of events

5'

 

 

 

 

Figure 3.2.5: Normalized roughness analysis as a function of the topographic
depth for Coors ADS-995, hydrothermally attacked for 10 hours at 250°C.

AD-94 Normalized Event Histogram - 10 hours

  
 

0.012 -
I 0.01 ~
% f.
E, i: 0.008 ~
0 .I
at - -
I 0.006 -
8 I r
-- t u.
5 0.004 -
8
2
0.002 _

 
  

 

 

 

I l I

-200 -150 -100 -50 0 50 100 150 200
Topographic Depth (nm)

Figure 3.2.6: Normalized roughness analysis as a function of the topographic
depth for Coors AD-94, hydrothermally attacked for 10 hours at 250°C.

145

ADS-995 Normalized Event Histogram - 50 hours
0.012 5

A 0.01 '

: 0.008 ~
I
I
i
i

I
J

l r r l i l l 1 l

-200 -150 -100 -50 0 50 100 150 200
Topographic Depth (nm)

I

E

I

' 0.006 ‘1
I
\ 0.004 -

Normalized # of events

 

 

Figure 3.2.7: Normalized roughness analysis as a function of the topographic
depth for Coors ADS-995, hydrothermally attacked for 50 hours at 250°C.

AD-94 Normalized Event Histogram - 50 hours

0.007 -

' 0.006 -
13 .
o _ P. 0.005 -
> ' I
3 g I
O 0.004 -
at l
‘0
g 0.003 «
a
E
0
Z

 

 

 

I I T

-800 -600 -400 -200 0 200 400 600 800
Topographic Depth (nm)

Figure 3.2.8: Normalized roughness analysis as a function of the topographic
depth for Coors AD-94, hydrothermally attacked for 50 hours at 250°C.

146

3.3 X-Ray Diffraction

The AD-94 and ADS-995 specimens were observed via X-ray diffraction.
Specimen conditions were: as-polished, and hydrothermally attacked for 10 and
50 hours. The specimens were mounted on an aluminum slide to minimize the
sources of foreign artifacts (i.e. an amorphous glass regime form a glass slide, or
iron from a steel slide), within the X-ray analysis. The specimens were then
compared with powder diffraction standard data (Tables 3.3.1-3) [35], data was
reported (Tables 3.3.7-10) in terms of interplanar spacing (d-spacing) and
relative intensities (Rel. lnt.).

All of the specimens contained necessary peaks for both alumina and
aluminum (Tables 3.3.1-4) [35]. Furthermore, specimens had different relative
intensities at given interplanar spacings, when compared with the standards
Tables 3.3.1-6 [35]. The reason for the discrepancies in the intensities is that
both ADS-995 and AD-94 have a preferred orientation or texture [36]. Preferred
orientations are the norm in most materials: metals, polymers and ceramics and
are a result of grain rotations during plastic deformation [36].

Some of the d-spacings of the standards were very similar to those of the
experimental specimens (Tables 3.3.7-10), however there are d-spacings that do
not correspond to the standards: aluminum, a-alumina, boehmite, diaspore, or
wairakite. Additionally, there were traces of boehmite, diaspore, and wairakite in
the non-attacked ADS-995 specimen. As the purity of the ADS-995 contradicts
these readings, the calibration of the diffractometer comes into question.

However it can be readily determined that there is no appreciable growth of

147

either the boehmite, diaspore, or wairakite phases, as the same peaks are found
at each time frame. Since the specimens have a visible growth on the surface

(Figures 3.1) the growths are likely amorphous or below the resolution of the

diffractometer.

148

Table 3.3.1: X-Ray Diffraction Analysis of Non-Hydrothermally
Attacked Specimens [35]

 

 

 

a—Alumina Standard ADS-995 0 Hours AD-94 0 Hours
d-spacing Rel. Int. d-spacing Rel. Int. d-spacing Rel. lnt.
Inc A III, III,
3.479 75 3.471 6 4 6.8805 5
2.552 90 2.5455 10 3.3984 23
2.379 40 2.3741 3 2.5082 16
2.165 1 2.3461 6 2.3433 53
2.085 100 2.2472 2 2.0584 60
1 .964 2 2.0818 1 1 2.0317 75
1 .74 45 2.0568 2 1 .7226 24
1 .601 80 2.0301 100 1 .5873 20
1 .546 4 1 .7381 6 1 .5336 3
1 .514 6 1.5999 16 1 .5025 4
1.51 8 1.5081 2 1.4344 31
1 .404 30 1 .4338 42 1 .3948 33
1 .374 50 1 .4036 5 1.3646 100
1 .337 2 1 .3727 7 1 .2228 22
1.276 4 1.2386 4 1.1839 13
1 .239 16 1 .2339 2 1 .1423 8
1.2343 8 1.2224 28 1.1211 6
1.1898 8 1.0988 2 1.0743 11
1.16 1 1.0425 3 1.0393 15
1.147 6 1.013 5 1.0145 4
1.1382 2 0.9977 3 1.0132 4
1 .1 255 6 0.9297 2 0.9954 4
1.1246 4 0.9078 3 0.9064 25
1.0988 8 0.9046 4
1.0831 4 0.898 17
1.0781 8
1.0426 14
1.0175 2
0.9976 12
0.9857 1
0.9819 4
0.9431 1
0.9413 1
0.9345 4
0.9178 4
0.9076 14
0.9052 4
0.8991 8
0.8884 1

 

149

Table 3.3.2: X-Ray Diffraction Analysis of 10-hour
Hydrothermally Attacked Specimens [35]

 

 

 

 

oz-Alumina Standard ADS-995 10 Hours AD-94 10 Hours
d-spacing Rel. Int. d-spacing Rel. lnt. d-spacing Rel. Int.
A m, A m, A m,
3.479 75 3.467 8 6.7412 5
2.552 90 2.5444 18 3.3984 23
2.379 40 2.3736 6 2.5082 16
2.165 1 2.3458 7 2.3433 53
2.085 100 2.0812 19 2.0584 60
1.964 2 2.0302 100 2.0317 75
1.74 45 1 .7375 1 1 1.7226 24
1 .601 80 1 .5996 28 1 .5873 20
1 .546 4 1 .5095 3 1.5336 3
1 .514 6 1 .4339 35 1.5025 4
1.51 8 1.4031 10 1.4344 31
1 .404 30 1 .3724 13 1 .3948 33
1 .374 50 1 .2383 7 1 .3646 100
1.337 2 1 .2336 2 1 .2228 22
1 .276 4 1 .2224 24 1 .1839 13
1 .239 16 1 .0987 3 1 .1423 8
1.2343 8 1.0778 2 1.1211 6
1 .1898 8 1 .0424 5 1 .0743 1 1
1.16 1 1.0131 4 1.0393 15
1.147 6 0.9976 5 1.0145 4
1 .1382 2 0.9295 2 1.0132 4
1 .1255 6 0.9077 3 0.9954 4
1 .1246 4 0.9057 5 0.9064 25
1.0988 8 0.9046 4
1.0831 4 0.898 17
1.0781 8
1.0426 14
1.0175 2
0.9976 12
0.9857 1
0.9819 4
0.9431 1
0.9413 1
0.9345 4
0.9178 4
0.9076 14
0.9052 4
0.8991 8
0.8884 1

 

150

 

Table 3.3.3: X-Ray Diffraction Analysis of 50-hour
Hydrothermally Attacked Specimens [35]

 

 

 

 

a-Alumina Standard ADS-995 50 Hours AD-94 50 Hours
d-spacing Rel. Int. d-spacing Rel. lnt. d-spacing Rel. Int.
A m, A m, A mg
3.479 75 3.4199 3.4199 7.0098 9
2.552 90 2.5205 2.5205 3.5229 7
2.379 40 2.3479 2.3479 3.4626 1 7
2.165 1 2.0657 2.0657 2.5416 14
2.085 1 00 2.056 2.056 2.3722 50
1 .964 2 2.0299 2.0299 2.3462 7
1 .74 45 1 .7273 1.7273 2.0801 49
1.601 80 1.5914 1.5914 2.03 96
1 .546 4 1 .5026 1 .5026 1.7365 19
1 .514 6 1 .4339 1 .4339 1.5988 24
1 .51 8 1.3975 1.3975 1.5443 4
1.404 30 1 .3671 1.3671 1.5124 4
1 .374 50 1 .2346 1 .2346 1 .4338 27
1 .337 2 1 .2295 1 .2295 1 .4028 32
1 .276 4 1 .2223 1 .2223 1 .3721 100
1 .239 1 6 1 .096 1 .096 1 .2222 17
1.2343 8 1 .0403 1 .0403 1 .1887 1 1
1.1898 8 1.013 1.013 1.1465 7
1.16 1 0.9959 0.9959 1.125 6
1.147 6 0.9061 0.9061 1.0777 10
1.1382 2 0.9048 0.9048 1.0423 13
1.1255 6 1.0173 4
1.1246 4 0.9078 18
1.0988 8 0.9057 4
1.0831 4 0.8993 12
1.0781 8
1.0426 14
1.0175 2
0.9976 12
0.9857 1
0.9819 4
0.9431 1
0.9413 1
0.9345 4
0.9178 4
0.9076 14
0.9052 4
0.8991 8
0.8884 1

 

151

Aluminum Standard

Table 3.3.4: X-RayDiffraction Analysis of Aluminum Mountiqu Slide [35]

Aluminum Slide

 

 

d-spacing Rel. Int. d-spacing Rel. lnt.
A In, A III.)
2.338 100 2.3266 5.831241
2.024 47 2.0157 100
1 .431 22 1 .4277 43.61628
1.221 24 1.2184 34.11352
1.169 7 1.0112 6.373178
1.0124 2 0.9281 4.685145
0.9289 8 0.9048 6.328016
0.9055 8
0.8266 8

 

152

Table 3.3.5: X-Ray Diffraction Standards for Diaspore and Boehmite [35]

 

Basic Aluminum Oxide (Diaspore) Aluminum Oxide Hydroxide (Boehmite)

 

 

d-spacing Rel. lnt. d-spacing Rel. lnt.
A III.) A l/lo
4.71 13 6.11 100
9.99 100 3.164 65
3.214 10 2.346 55
2.558 30 1.98 6
2.434 3 1.86 30
2.386 5 1.85 25
2.356 8 1.77 6
2.317 56 1.662 14
2.131 52 1.527 6
2.077 49 1.453 16
1.901 3 1.434 10
1.815 8 1.412 2
1.733 3 1.396 2
1.712 15 1.383 6
1.678 3 1.369 2
1.633 43 1.312 16
1.608 12 1.303 4
1.57 4 1.224 2
1.522 6 1.209 2
1.48 20 1.178 4
1.431 7 1.171 1 2
1.423 12 1.1609 4
1.4 6 1.1337 6
1.376 16 1.1 152 2
1.34 5 1.0917 2
1.329 6 1.0459 2
1.304 3 1.0281 2
1.289 6 0.9903 2
1.279 1 0.9818 2
1.256 4 0.9506 2
1.243 5 0.931 2
1.218 2 0.9247 2
1.204 4 0.9105 2
1.1783 1 0.9023 2
1.1739 7 0.8937 2
1.1408 3 0.8907 2
1.1003 1 0.866 2
1.0923 3 0.8607 2
0.8316

 

153

Table 3.3.6: X-Ray Diffraction Standard for
Wairakite (Pseudo-Cubic, Monoclinig [3g

 

Wairakite (Pseudo-Cubic, Monoclinic)

 

d-spacing Rel. Int.
A l/lo
6.85 40
5.57 80
4.84 40
3.64 30
3.42 60
3.39 100
2.909 50
2.897 30
2.783 10
2.77 10
2.68 40
2.67 10
2.5 5
2.489 40
2.418 30
2.215 40
2.17 5
2.147 10
2.1 15 10
2.095 5
1.996 20
1.867 10
1.857 30
1.844 10
1.708 5
1.696 5
1.595 5
1.586 20
1.487 10
1.354 10
1.343 10

 

154

Table 3.3.7: Correlation between the d-spacings of the a-Alumina Standard
and different stages of the Alumina Specimen Hydrothermal Attack [35]

 

a-Alumina ADS-995 0 ADS-995 ADS-995 AD-94 0 AD-9410 AD-94 50
Standard Hours 10 Hours 50 Hours Hours Hours Hours

d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing
A A

 

3.479
2.552
2.379
2.085
1.740
1.601
1.546
1.514
1.510
1.404
1.374
1.239
1.234
1.190
1.147
1.138
1.125
1.099
1.078
1.043
1.018
0.998
0.935
0.908
0.905
0.899

 

3.472
2.546
2.374
2.082
1 .738
1 .600

1 .508
1.404
1.373
1.239
1 .234

1 .099

1.043
1.013
0.998
0.930
0.908
0.906

3.467
2.544
2.374
2.081
1.738
1 .600

1.510
1.403
1.372
1.238
1.234

1.099
1.078
1.042
1.013
0.998
0.930
0.908
0.906

1.591

1.503
1.398
1 .367
1.235
1 .230
1 .096

1 .040

1.013
0.996

0.906
0.905

 

3.490

1 .593

1 .507
1 .398
1 .368

1.186
1.144

1.122

1.075
1.040
1.015
0.996

0.907
0.905
0.898

1 .503
1 .395
1 .365

1.223
1.184
1.142

1.121

1.074
1.039
1.015
0.995

0.906
0.905
0.898

2.542
2.372
2.080
1 .737
1 .599
1 .544
1 .512

1 .403
1 .372

1.222
1.189
1.147

1.125

1.078
1.042
1.017

0.908
0.906
0.899

 

155

Table 3.3.8: Correlation between the d-spacings of the Boehmite Standard
and different stages of the Alumina Specimen Hydrothermal Attack [35]

 

 

 

 

Boehmite ADS-995 0 ADS-995 ADS-995 AD—94 0 AD-9410 A094 50
Hours 10 Hours 50 Hours Hours Hours Hours
d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing

2.346 2.346 2.346 2.348 2.343 2.346

1.527 1 .539 1 .534

1 .434 1 .434 1 .434 1 .434 1.429 1 .434 1 .434

1.412 1.404

1 .396 1 .403 1.398 1 .398 1 .395 1 .403

1 .369 1 .373 1 .372 1.367 1 .368 1 .365 1 .372

1 .224 1 .222 1 .222 1 .222 1 .219 1 .223 1.222

1.178 1.186 1.184 1.189

1.134 1.122 1.121 1.125

1 .092 1.099 1 .099 1 .096

1 .046 1 .043 1 .042 1.040

0.990 0.998 0.998 0.996

0.982 0.996

0.951 0.995

0.931 0.930 0.930

0.911 0.908 0.908 0.906 0.908

0.902 0.906 0.906 0.905 0.907 0.906

0.894 0.905 0.906 0.899

0.891 0.898 0.905

0.866 0.898

 

156

Table 3.3.9: Correlation between the d-spacings of the Diaspore Standard
and different stages of the Alumina Specimen HLdrothermal Attack [35]

 

 

 

 

Diaspore ADS-995 0 ADS-995 ADS-995 AD-94 0 AD-9410 AD-94 50
Hours 10 Hours 50 Hours Hours Hours Hours
d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing
4.710 3.420
2.558 2.546 2.544
2.356 2.357 2.343 2.346
2.077 2.066 2.069 2.080
1 .733 1.738 1 .738 1 .727 1 .729 1 .723 1.737
1.608 1.600 1.600 1.591
1.522 1.512
1 .431 1 .434 1 .434 1 .434 1 .429 1 .434 1.434
1 .400 1 .404 1 .403 1 .398 1 .398 1 .395 1.403
1 .376 1.373 1 .372 1.367 1 .368 1.372
1 .243 1 .239 1 .238 1 .235
1.218 1.222 1.222 1.222 1.219 1.223 1.222
1.141 1.144 1.142 1.147
1.100 1.099 1.099
1.092 1.096

 

Table 3.3.10: Correlation between the d-spacings of the Wairakite Standard
and different stages of the Alumina Specimen Hydrothermal Attack [35]

 

 

 

 

Wairakite ADS-995 0 ADS-995 ADS-995 AD-94 0 AD-9410 AD-94 50
Hours 10 Hours 50 Hours Hours Hours Hours
d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing d-spacing
3.420 3.420
2.500 2.508
2.350 2.346 2.346 2.348 2.357 2.343 2.346
2.095 2.082 2.081 2.080
1.732 1.738 1 .738 1.727 1.729 1.723 1.737
1 .595 1.600 1.600 1.591 1.593 1.599
1.586 1.587
1 .437 1 .434 1 .434 1 .434 1 .434 1.434
1 .407 1 .404 1 .403 1.398 1 .398 1 .395 1.403
1.354 1.365
1.215 1.222 1.222 1.222 1.219 1.223 1.222

 

157

3.4 Mass Changes

The change in mass resulting form hydrothermal corrosion for the Coors
aluminas was determined for the given research. Twin specimens of AD-94 were
exposed for 100 hours at 250°C within the same hydrothermal environment,
resulting in twice the amount of surface area available to attack. The initial mass

of the specimens was 0.1949 g and the final mass was 0.2040 g.

mf - m,“
Am% = x100 3.4.1

mi

 

Using Equation 3.4.1, where Am°/o is the percentage mass change, m is the
initial mass and m, is the final mass, the percentage mass gain was 4.66%. The

observed mass gain is consistent with the formation of a hydrated phase.

3.5 Sectioning of the Atomic Force Microscopy Micrographs

The raw data of the AFM micrographs was obtained from the DI software.
That data was then imported into a Micorsoft Access database and placed into a
3-column format of (x,y,z), where ‘x’ and ‘y’ are the cartesian coordinates of the
image, and ‘z’ is the depth or height of the point relative to the average height of
the specimen. Each axis is either 256 or 512 entries long depending on the
image. Next, the data was imported into a Microsoft Excel spreadsheet.
Simultaneously, using another software package, WSxM 1.1 (Nanotec
Electronica: Cantoblanco, Madrid.) the maximum specimen depth in meters was
determined for each micrograph. The maximum depth in meters was equated to

the maximum depth in the ‘z’ axis. The data was sectioned by retrieving the

158

range of data points corresponding to the depths of interest in nanometers from
the database and placing the data into a spreadsheet where the data was plotted
(Figures 3.5.1-5). In order to interpret, either the sectioned plot or the original
image must undergo a pair of transformations: mirroring and a 90° counter-
clockwise rotation (these transformations have already been done for every
image within this thesis). The reason for the transformations is that the
spreadsheet program plots the data in a different manner than the imaging
software.

The total area present at a given depth range was calculated from the
number of data points present within a given depth range and the area
corresponding to each pixel. The area was then analyzed with respect to: (i) the
midpoint of the depth intervals (Figures 3.5.7—16), (ii) the midpoint of the depth
intervals normalized with the average grain size (Figures 3.5.17-28) and (iii) the
square root of the hydrothermal corrosion times (Figures 3.5.29-40).

Specimens were analyzed prior to formation of the hydrothermal scale.
Specimens were also analyzed after hydrothermal corrosion had etched the
surface of the specimen. For the ADS-995 specimens the hydrothermal
exposure times analyzed were 10, 12.5 and 15 hours. However, as the
hydrothermal corrosion progressed faster in the AD-94 specimens the
hydrothermal exposure times analyzed were 1, 1.5 and 2 hours.

For Figures 3.5, the regression relationships have been determined through
various methods: linear, power law, logarithmic, exponential, Gumbel or Cauchy.

The regressions are contained within Tables 3.5.1-6, the regressions represent

159

the amount of area present at a given depth “ad” as a function of ‘dmid’ Which
represents: the midpoint of the depth intervals, the midpoint of the depth intervals
normalized with the average grain size and the square root of the hydrothermal
corrosion times. The coefficient of determination (r2) value is a measure of how
accurate the regression line depicts the data sets, with the closer the r2 value is
to 1 the better the fit. The regression equations and the r2 values for ADS-995
and AD94 specimens are given (Tables 3.5.1-6).

For the grain boundary etching as a function of the depth (Figures 3.5.7-16)
and as a function of the depth and the average grain size (Figures 3.5.17-28) two
types of distributions were determined: Gumbel (Equation 3.5.1) and Cauchy
(Equation 3.5.3).

The Gumbel distribution (Equation 3.5.1-2) is a doubly exponential
distribution, while the Cauchy distribution (Equation 3.5.3-4) is the exponential of
a power expression, where “x” is a random variable with constants “01-3,” and “k.“
Gumbel and Cauchy distributions are each a form of extremal-value statistical
theory. The Gumbel distribution has been used to analyze the distribution of pit
depths [37] while the Cauchy distribution is used to explain that the pit depth
increases with increasing surface area. Note the r2 values of the Gumbel and
Cauchy functions are independent of grain size (Tables 3.5.1-4).

Tables 3.5.2,4 have additional regressions that correspond to Figures
3513,24 that are valid for the shallow depths (-5 to -275 nm). The Cauchy
distribution has a better coefficient of determination while the Gumbel distribution

is worse when compared to the coefficient of determination relationship near the

160

specimen surface. The Gumbel distribution is a representation of the corrosion
depths, while the Cauchy distribution is descriptive of the pit depth distribution.

There is more corroded surface present (Figure 3.5.11) closer (1742 A2
from -5 to —25 nm) to the surface of the specimen than in the pits (0.34 A2 from -
85 to-105 nm) as collected in Tables C.1-2. As the hydrothermal attack
progresses (Figures 3.5.6.8), there is more corroded surface area present
(470.75 A2) at the 10-hour trial than the 15—hour exposure (2021.37 A"). The
ADS-995 (Figure 3.5.8) was corroded more at 15-hours (2021.37 A2), than the
AD-94 (Figure 3.5.12) was attacked after 1.5 hours of corrosion (1504.78 A2).
The ADS-995 (Figure 3.5.8) has a narrow area distribution between 665.70 A2
and 70.04 A2 independent of the grain size (Figure 3.5.19), for the depth of the
trial (-10 to -110 nm). While, the AD-94 (Figure 3.5.12) has a wide area
distribution between 1742.02 A2 and 0.88 A2, independent of the grain size
(Figure 3.5.23), for the depth of the trial (-5 to -105 nm). Thus, the hydrothermal
corrosion performs a deep, narrow etch of the AD-94 surface and a wide, shallow
etch of the ADS-995 surface. These results correspond to previous work“:10
relating a decrease in strength to an increase in the impurity concentration as
hydrothermal corrosion progresses when coupled with the knowledge that a
material will fail under the deepest surface flaw. The AD-94 has deeper surface
flaws than the ADS-995.

The surface pitting of the ADS-995 showed a predominantly power-law

relationship with the square root of time whereas the AD-94 had a mostly linear

relationship with the square root of time. In both ADS-994 and AD-94, the

161

exponential and the power-law relationships are very similar (Figures; 3.5.29-40).
Therefore, the ADS-995 and the AD-94 the nature of the grain boundary etching
as a function of the square root of time was probably mixed mode (power,

exponential and linear) as the coefficients of determination at the various depths

are similar. Inferring that, at different depths a different relationship is dominant.

162

ad = exp(-cxp[—x]) 3.5.1

 

—c ex [—c d - ] 3.5.2
ad =exp( 1 (1432: Mid )
ad :exp(_x—k) 3.5.3
-k 3.5.4
= C3dmid
ad em 0.4329 )

Table 3.5.1: Regressions from the grain boundary etching of ADS-995 attacked
at 250°C for 10 to 15 hours as a function of the depth from the specimen surface

 

 

 

 

 

 

 

 

Conditions Regression Type Regression r2
8 —4.39
10 hours Cauchy-type ad =exp(3'10 dmid ) 0.9814
0.4329
0.45 —0.075d -
10 hours Gumbel-type ad =exp(85 ex“ ""d ) 0.9224
0.4329
-1.0328
12.5 hours Cauchy-type ad = exp(3442'ldmial ) 0.6315
0.4329
.44 - . 2 -
12.5 hours Gumbel-type ad =exp(240 exp“ 00 2”de )) 0.8310
0.4329
-0.8001
15 hours Cauchy-type ad : exp(6860'4dmid ) 0.8197
0.4329
. . 4 -
15 hours Gumbel-type ad =exp(478 76ex%(;:20908 dm'd)) 0.6670

 

163

Table 3.5.2: Regressions from the grain boundary etching of AD-94 attacked at
250°C for 1 to 2 hours as a function of the depth from the specimen surface

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Conditions Relession Type Regression r2
1.0 hour Cauchy-type ad = exp( 0:31:19 ) 0.7873
1.0 hour Gumbel-type ad = exp(23254e"1’(‘°'“°5dmid ) 0.9496

0.4329
—1.8859
1.5 hours 310203d .
(001) Cauchy-type ad : ex“ 0 4:31: ) 0.9339
1.5 hours _ _ 441.786xp(—0.013ldmid)
(001) Gumbel type ad — exp( 0.4329 ) 0.8943
—0.9346
1.5 hours 5314.7d .
(012) Cauchy-type ad = exp( 0 4321;! ) 0.8756
15 hours _ _ 247.93 exp(-0.0083dmid)
(012) Gumbel type ad — exp( 0.4329 0.9547
—0.9643
14443d .
2.0 hours Cauchy-type ad = exp( 0 4:36; ) 0.8734
2.0 hours Gumbel-type ad = exp(481'756xP(—0'0068dmid )) 0.5996
0.4329
(-5 to -275 nm)
Conditions Regression Type Regression r2
-l.6014
1.5 hours 102741d .
Cauch - e _ mtd 0.9683
(001) Y WP ad — exp( 0.4329 )
1.5 hours 528.4exp(—0.0155dmid)
Gumbel- e = 0.7626
(001) ‘y" “d “9‘ 0.4329 ’
—0.7329
1.5 hours 2472.4d .
Cauch - e _ mid 0.9274
15 hours 239.216xp(—0.0078dmid)
Gumbel- e = 0.8940
(012) typ ad em 0.4329 )

 

164

Table 3.5.3: Regressions from the grain boundary etching of ADS-995 attacked
at 250°C for 10 to 15 hours as a function of the depth from the specimen surface
and the average_grain size

 

 

 

 

 

 

 

 

 

Conditions Regression Type Regression 12
6-10‘7d‘4-39
10 hours Cauchy-type a d = exp( mzd ) 0.9814
0.4329
. —1 . -
10 hours Gumbel-type ad = exp(850 45 ”1:43;: OSdm‘d )) 0.9224
—1.0328
12.5 hours Cauchy-type ad : exp(l'2151dmid ) 0.6315
0.4329
24 .44 —52.128 -
12.5 hours Gumbel-type ad =exp( 0 ex“ ""d )) 0.8310
0.4329
—0.8001
15 hours Cauchy-type ad z exp(l4'524dmid 0.8197
0.4329
4 . —18.412d -
15 hours Gumbel-type ad =cxp 78 766x:(4329 ""01 ) 0.6670

 

165

Table 3.5.4: Regressions from the grain boundary etching of AD-94 attacked at
250°C for 1 to 2 hours as a function of the depth from the specimen surface and

the average grain size

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Conditions Regression Type Regression 12
10—9 d-4.419
1.0 hour Cauchy-type ad = exp( 0 £12“; ) 0.7873
2 2 - 2 . °
1.0 hour Gumbel-type ad =exp( 3 54exl§4$96 ldm’d )) 0.9496
—1.8859
1.5 hours 0.0063d .
(001) Cauchy-type a d = exp( 0 437121: ) 0.9339
1.5 hours _ _ 441.788XP(-156-72dmid)
(001) Gumbel type ad —cxp( 0.4329 0.8943
—0.9346
1.5 hours 0.8189d .
(012) Cauchy-type ad : ex“ 0 4:31:21: ) 0.8756
15 hours _ 247.93cxp(—99.811dmid)
(012) Gumbel-type ad —exp( 0.4329 ) 0.9547
—0.9643
2.0 hours Cauchy-type ad = exp(L6824dmid 0.8734
0.4329
2.0 hours Gumbel-type ad =exp(481'75 exP(_81'722dmid )) 0.5996
0.4329
(-5 to -275 nm)
Conditions Regression Type Regression r2
-1.6014
1.5 hours 0.0302d .
(001) Cauchy-type “cl = exp( 0 4:121: ) 0.9683
2 . — . '
(1631“)w's Gumbel-type ad = exp(5 8 46x93 4138269°2dm'd) 0.7626
-0.7329
1.5 hours 2.5318d .
(012) Cauchy-type ad 2 exp( 0 4:12ch ) 0.9274
15 hours 239.216xp(-94.192dmid)
Gumbel- e = 0.8940
(012) typ ad cm 0.4329 )

 

166

Table 3.5.5: Regressions from the grain boundary etching of ADS-995 attacked

 

 

at 250°C for 10 to 15 hours
Conditions Regression Type Regression 12

RT -10 to —30 nm Power ad = 3430845323 0.0314
RT -10 to —30 nm Exponential ad = 39.962 exP(0-5742dmid) 0.0423
RT —1 0 to -30 nm Linear ad = 326.87dmid — 757.39 0.1674
RT -30 to —50 nm Power ad = 00016.12)?” 0.9999
RT -30 to -50 nm Exponential ad = 0.0152 eXp(2-5464dmid) 0.9986
RT —30 to —50 nm Linear “cl = 334-95dmid - 1026.1 0.9567
RT -50 to -70 nm Power ad = 5 o10“11d;§a968 0.9806
RT —50 to —70 nm Exponential “cl = 10‘8 6XP(6-2461dmid) 0.9718
RT —50 to —70 nm Linear ad = 430.63dmid - 1387.7 0.8924
RT —70 to —90 nm Power ad = 3 010'l4d35a224 0.9800
RT -70 to —90 nm Exponential a d = 3 010-1 1 exp(7.74dm,-d ) 0.971
RT -70 to —90 nm Linear ad = 339.64dmid -1100.8 0.8504
RT -90 to —1 10 nm Power ad = 5 .10‘17 dilifw 0.9931
RT -90 to -110 nm Exponential a d = 2 .10"13 exp(9.0197dm,-d) 0.9875
RT —90 to —1 10 nm Linear ad = 251-53dmid - 820.19 0.7926

 

167

Table 3.5.6: Regressions from the grain boundary etching of AD-94 attacked at

250°C for 1 to 2 hours

 

 

Conditions Refission Type Remession r2
RT —5 to -25 nm Power ad = 1515342127402 0.5179
RT —5 to -25 nm Exponential ad = 5954.7 exP(-l.39ld mid) 0.4692
RT -5 to —25 nm Linear ad = —l849.3d,m-d + 3413.8 0.5651
RT -25 to —45 nm Power ad = 606. 1752-14524 0.0274
RT —25 to —45 nm Exponential ad = 773-66Xp(-0-2694dmid) 0.0138
RT -25 to —45 nm Linear ad = -17l.5dm,-d + 805.05 0.0222
RT -45 to —65 nm Power ad =137.45d31-i1f33 0.7301
RT —45 to -65 nm Exponential ad = 21.1876Xp(1-8681dmid) 0.7722
RT -45 to —65 nm Linear ad = 436.69d mid - 310.89 0.7509
RT —65 to —85 nm Power ad = 6.5302dfn2j49 0.9039
RT —65 to —85 nm Exponential ad = 0.00136Xp(8.6248d mid) 0.8734
RT —65 to -85 nm Linear ad = 385.98dmid — 376.12 0.9865
RT —85 to 105 nm Power ad = 056483565 0.8994
RT —85 to 105 nm Exponential a d = 3 010—7 exp(l4,74dmid) 0.8682
RT —85 to 105 nm Linear ad = 329.06de - 330.14 0.9984

 

168

AD-94 Hydrothen'naly Annealed for 1.5 hours (-5.4nm to -25.4nm)

  
  

  

 

 

504 - , . L, . - ~-
4487 r I f \\
392 . «‘1‘, t" 'M' u
336 - .
A ’0'
g 280 .
$7 224 ; f 1” I '
168 ‘ I a
112 ‘ a
56 (I /
0 r r .Fh—r—r-
0 112 168 224 280 336 392 448 504
X(nm)

Figure 3.5.1: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a depth of
—5.4 to -25.4 nm.

AD-94 Hydrothermaly Annealed for 1.5 hours (-25.4nm to -45.4nm)

 

 

 

 

3 3 .
l 6 I
0 "1—M .3 .1: 3 i r r r
0 56 112 168 224 280 336 392 448 504

x (nm)

Figure 3.5.2: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a depth of
—25.4 to —45.4 nm.

169

ADM Hydromermaily Annealed for 1.5 hours (454nm to 65.41!!!)

504 0 .
448 W\
392 :0 o , .8
00 I I

O

   
 

336

g 280

5

>- 224
168
112 -

56- W

0 c
0 M r r . . r . . .
o 56 112 168 224 280 336 392 449 504
X(nm)

Figure 3.5.3: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a depth of
—45.4 to —65.4 nm.

 

 

AD—94 Hydrothermally Annealed for 1.5 hours (-65.4nm to -85.4nm)

504 W
448
s . ‘

392 , o t s s
336 3 '

,. .. a

e 280 '

5

>. 224

so:

 

 

 

168 1 s
14" . /
112 - {s
56 - o . W '
0 P I r I I I I I I
0 56 1 12 168 224 280 336 392 448 504

X (nm)

Figure 3.5.4: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a depth of
—65.4 to —85.4 nm.

170

AD-94 Hydrothermally Annealed for 1.5 hours (-85.4nm to -105.4nm)

 

 

448 W‘
O
392 ‘° . , so
336 ° ' °
.. “ 3!
e 280
5
> 224 m
I o
168 — 33>"
112 — ’ (i
56 - . . I49
0 I. e I I T I I I I I
0 56 112 168 224 280 336 392 448 504

X (nm)

Figure 3.5.5: AD-94 hydrothermally attacked for 1.5 hours at 250°C at a depth of
-85.4 to -105.4 nm.

Grain Boundary Etching as a Function of Depth for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 hours

1 000

J

100

1

10*

Area (A2)

 

 

0.1

 

0 20 40 60 80 100 120 140
Midpoint of Intervals (nm)

0 A08995 10H -002 — Power (A08995 10H -002) - - Export. (ADSQQS 10H -002)

Figure 3.5.6: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 10 hours

171

Grain Boundary Etching as a Function of Depth for ADS-995
Hydrothermally Annealed at 250 Celsius for 12.5 hours

 

 

 

1000 -
A 100 -
‘33;
(U
9
< 10 ~
1 T i i l i j
0 20 40 60 80 100 120
Midpoint of intervals (nm)
1:: ADSQQS 12H5 —Power (ADSQQS 12H5) — - Expon. (AD8995 12H5)

Figure 3.5.7: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 12.5 hours

Grain Boundary Etching as a Function of Depth for ADS-995
Hydrothermally Annealed at 250 Celsius for 15 hours

 

 

 

 

1000 -
A
A 100 1 A ----- :
‘25.
(U
9
< 10 -
1 i i i i l
0 50 100 150 200 250
Midpoint of intervals (nm)
A ADSQQS 15H0 —- Power (#108995 15H0) - - Expoh. (ADSQQS 151-10)

Figure 3.5.8: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 15 hours

172

Grain Boundary Etching as a Function of Depth for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 12.5 hours

 

 

1000 .
O
100 - a a C,
A O D
(2‘s, 1:]
to 10 -
d)
E o
1 _ O
<> 0
0.1 l i i i i i l
0 20 40 60 80 100 120 140
Midpoint of Intervals (nm)
0 ADSQQS 10H -002 D ADSQQS 12145

Figure 3.5.9: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 10-12.5 hours

Grain Boundary Etching as a Function of Depth for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours

 

 

1000 -
A
O
D. A A A
100 - o D D 4 A A
A O D
‘33; 1:3
to 10 ~
92
< O
1 - <>
0 0
0. 1 i i I i i
0 50 100 150 200 250
Midpoint of Intervals (nm)
0 ADSQQS 10H ~002 1:2 ADSQQ‘S 12H5 A ADSQQS 15H0

Figure 3.5.10: Grain boundary etching as a function of depth for ADS-995
hydrothermally attacked at 250°C for 10-15 hours

173

Grain Boundary Etching as a Function of Depth for AD-94
Hydrothermally Annealed at 250 Celsius for 1 hour

 

 

 

10000 3
1000 -
22‘ 100 -
(v
(D
5.2 10 -
1 _ .-
O
0.1 . , , , T I
0 20 40 60 80 100 120
Midpoint of intervals (nm)
<> A094 1D0 -000 — Power (A094 1D0 -000) - - Expon. (AD94 1DO -000)

Figure 3.5.11: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1 hour

Grain Boundary Etching as a Function of Depth for AD-94
Hydrothermally Annealed at 250 Celsius for 1.5 hours

 

 

 

10000 -
1000 ~
‘2'; 100 -
8
3 10 -
1 _
0.1 i l T i i
0 100 200 300 400 500
Midpoint of Intervals (nm)
a ADQ41DS-001 ~. ADQ41DS -012 —Power(A094 105 -001)

 

- - Expon. (A094 105 -001) Power (A094 105 «012) — - Expon. (A094 105 -O12)

Figure 3.5.12: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1.5 hours

174

Grain Boundary Etching as a Function of Depth for AD-94
Hydrothermally Annealed at 250 Celsius for 1.5 hours (-5 to -275 nm)

10000 -
1000 -

100 ~

Area (A2)

 

104

 

 

 

50 100 1 50

200

250
Midpoint of intervals [Average Grain Size

Cl A094 105 -001 O A094 105 -012
- - Expon. (A094 105 -001)

— Power (AD94 1D5 -001)
— Power (A094 1 D5 -012)

— — Expon. (A094 105 -012)
Figure 3.5.13: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1.5 hours (-5 to —275 nm)

Grain Boundary Etching as a Function of Depth for AD-94
Hydrothermally Annealed at 250 Celsius for 2 hours
10000 5

1000 -

100 -*

Area (A2)

 

A
10-

 

I I I

50 100 150 200 250 300

350 400
Midpoint of Intervals (nm)
e A094 200 -001 — Power (AD94 200 -001) - - Expon. (A094 200 -001)
Figure 3.5.14: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 2 hours

175

Grain Boundary Etching as a Function of Depth for AD-94
Hydrothermally Annealed at 250 Celsius for 1 - 1.5 hours

10000 -

1000 — 8

DO

100 —

Area (A2)

10- ,3:

1‘ 1:1

 

 

0 . 1 I I I I I
0 1 00 200 300 400 500

Midpoint of intervals (nm)

0 A094 100 -000 D A094 105 -001 X A094 105 -012

Figure 3.5.15: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1-1.5 hours

Grain Boundary Etching as a Function of Depth for AD-94
Hydrothermally Annealed at 250 Celsius for 1 - 2 hours

 

 

10000 -
- §
1000 8
‘< A
A > 6 &
12$ 100 — D g ,3 a
(c U Q
9 D ,
< 10 - o C
1 7 1:1
0
0. 1 l i i l I
0 100 200 300 400 500
Midpoint of intervals (nm)
0 AD94 1D0 -000 D AD94 1 D5 -001 A AD94 200 -001 >< A094 105 -012

Figure 3.5.16: Grain boundary etching as a function of depth for AD-94
hydrothermally attacked at 250°C for 1-2 hours

176

Grain Boundary Etching as a Function of Depth and Average Grain Size
for ADS-995 Hydrothermally Annealed at 250 Celsius for 10 hours

1000 -
100 ~

10‘

Area (A2)

 

 

0.1

I I

0 0.01 0.02 0.03 0.04 0.05 0.06

Midpoint of Intervals / Average Grain Size
— Power (AD8995 10H -002) — — Expon. (A05995 10H -002)

Figure 3.5.17: Grain boundary etching as a function of depth and average grain
size for ADS-995 hydrothermally attacked at 250°C for 10 hours

0 A08995 10H -002

Grain Boundary Etching as 3 chtion of Depth and Average Grain Size
for ADS—995 Hydrothermally Annealed at 250 Celsius for 12.5 hours

1000 -

100 ~

Area (A2)

 

10-

 

I

0 0.01 0.02 0.03 0.04 0.05
Midpoint of intervals I Average Grain Size
13 AD8995 12H5 — Power (ADSQQS 12H5) — - Expon. (A08995 12H5)

Figure 3.5.18: Grain boundary etching as a function of depth and average grain
size for ADS-995 hydrothermally attacked at 250°C for 12.5 hours

177

Grain Boundary Etching as a Function of Depth and Average Grain Size
for ADS-995 Hydrothermafly Annealed at 250 Celsius for 15 hours

 

 

 

1000 —
A 100 -
‘33.
(U
9
<
10 -
1 I I I I I I
0 0.02 0.04 0.06 0.08 0.1 0.12

Midpoint of Intervals I Average Grain Size
A ADSQQS 15H0 — Power (ADSQQS 151-10) — - Expon. (AD8995 15H0)

Figure 3.5.19: Grain boundary etching as a function of depth and average grain
size for ADS-995 hydrothermally attacked at 250°C for 15 hours

Grain Boundary Etching as a Function of Depth and Grain Size for ADS-
995 Hydrothermally Annealed at 250 Celsius for 10 - 12.5 hours

 

 

1000 o
O

100 - a U B
A O C]
a 10 -
92
< 0

1 a <>
0 0
0.1 I l I I I 1
0 0.01 0.02 0.03 0.04 0.05 0.06

Midpoint of intervals I Average Grain Size

0 A08995 10H -002 D A08995 12H5

Figure 3.5.20: Grain boundary etching as a function of depth and average grain
size for ADS-995 hydrothermally attacked at 250°C for 10-12.5 hours

178

Grain Boundary Etching as a Function of Depth and Average Grain Size
for ADS-995 Hydrothermally Annealed at 250 Celsius for 10 - 15 hours

 

 

1000 -
A
O
A A A
100- o D D “ A A A
A O U
‘35, 13
(U 10 4
9
< O
1 a 0
O O
0.1 I I I T I I
0 0.02 0.04 0.06 0.08 0.1

0.12
Midpoint of intervals I Average Grain Size

0 A08995 10H -002 1:3 A08995 12H5 A A08995 15H0

Figure 3.5.21: Grain boundary etching as a function of depth and average grain
size for ADS-995 hydrothermally attacked at 250°C for 10-15 hours

Grain Boundary Etching as a Function of Depth and Average Grain
Size for AD-94 Hydrotherrnaliy Annealed at 250 Celsius for 1 hour

 

 

 

10000 .

1000 —
9.4: 100 .
(U
s
< 10 -

1 _
0.1 l i i l l
0 0.002 0.004

0.006 0.008 0.01
Midpoint of intervals I Average Grain Size
0 A094 1D0 -000

— Power (AD94 100 -000) - - Expon. (A094 100 -000)
Figure 3.5.2: Grain boundary etching as a function of depth and average grain
size for AD-94 hydrothermally attacked at 250°C for 1 hour

179

Grain Boundary Etching as a Function of Depth and Average Grain
Size for A094 Hydrothermally Annealed at 250 Celsius for 1.5 hours

 

 

 

10000 -
1000 .
§ 100 «
79’
a 10 .
1 _
0.1 i i l i i
0 0.01 0.02 0.03 0.04 0.05
Midpoint of intervals I Average Grain Size
:1 A094 105 -001 o AD94 105 -012
-— Power (A094 105 -001i — - Exoon. (ADQ4 1D5 -0011

Figure 3.5.23: Grain boundary etching as a function of depth and average grain
size for AD-94 hydrothermally attacked at 250°C for 1.5 hours

Grain Boundary Etching as a Function of Depth and Average
Grain Size for AD-94 Hydrothennaiiy Annealed at 250 Celsius for
1.5 hours (-5 to -275 nm)

 

 

 

 

 

10000 —
1000 a
‘23:
«r 100 ~
8’
<
10 -
1 1 i l I i
0 0.005 0.01 0.015 0.02 0.025
Midpoint of Intervals I Average Grain Size
1: AD94 1D5 -001 o A094 1 D5 -012 — Power (A094 105 -001)
- — Expon. (AD94 105 -001) — Power (A094 105 -012) — — Expon. (A094 1D5 -012)

Figure 3.5.24: Grain boundary etching as a function of depth and average grain
size for AD-94 hydrothermally attacked at 250°C for 1.5 hours (-5 to -275 nm)

180

Grain Bomdary Etching as a Function of Depth and Average Grain
Size for AD-94 Hydrothermally Annealed at 250 Celsius for 2 hours

 

 

 

10000 ~

1000 ~
‘35.

(u 100 .
2
<

10 1

1 i l 17 i F l l
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Midpoint of intervals / Average Grain Size
A A094 2D0 -001 — Power (AD94 2D0 -001) - — Expon. (A094 200 -001)

Figure 3.5.25: Grain boundary etching as a function of depth and average grain
size for AD-94 hydrothermally attacked at 250°C for 2 hours

Grain Boundary Etching as a Function of Depth and Grain Size for
AD-94 Hydrothermally Annealed at 250 Celsius for 1 - 1.5 hours

 

 

10000 —

_ 8-
1000 < 8
(7%: 100 7 >: 8 1:1 ’0 ,.
8 D
g 10 - o 33"
1 7 :1
O
0.1 i i 1 fl 1
0 0.01 0.02 0.03 0.04 0.05
Midpoint of intervals lAverage Grain Size
0 AD94 100 -000 r: AD94 1D5 -001 x A094 1D5 -012

Figure 3.5.26: Grain boundary etching as a function of depth and average grain
size for AD-94 hydrotherrnaliy attacked at 250°C for 1-1.5 hours

181

Grain Boundary Etching as a Function of Depth and Average Grain
Size for AD-94 Hydrothermaly Annealed at 250 Celsius for 1 - 2 hours

 

 

10000 -
- 3
1000 8
x A
A .1 B. A
<25, 100 - D 3 4, a
(U D 9
9 [:1
< 10 7 C3
0
1 _ :1
O
0.1 l i l l I
0 0.01 0.02 0.03 0.04 0.05

Midpoint of intervals / Average Grain Size

0 A094 100 -000 D A094 105 -001 A A094 200 -001 x A094 105 -012

Figure 3.5.27: Grain boundary etching as a function of depth and average grain
size for AD-94 hydrotherrnaiiy attacked at 250°C for 1-2 hours

Grain Boundary Etching as a Function of Depth and Average
Grain Size for ADS-995 (10-15 hours) and AD-94 (1-2 hours)
Hydrothermally Annealed at 250 Celsius

 

 

10000 -

1000 - +
A $5.3“ X + + + +
it, 100 — BEE; 4 :2 ¢ L A o + +
to 0 ~
9 10 - D >: O
< o x D

1 7 X0
0 x >1:
0.1 I f I I I I
0 0.02 0.04 0.06 0.08 0.1 0.12

Midpoint of intervals I Average Grain Size

0 A094 100 -000 D A094 105 -001 A A094 200 -001
X A08995 10H -002 O A08995 12H5 + A05995 15H0

Figure 3.5.28: Grain boundary etching as a function of depth and average grain
size for ADS-995 (10-15 hours) and AD-94 (1 -2 hours) hydrothermally attacked
at 250°C

8 A094105 -012

182

Grain Boundary Etching as a Function of Root Time for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours
(-10 to -30 nm)

 

 

 

0 I T I I I
3 3.2 3.4 3.6 3.8 4

Root Time

0 -10 to -30nm — Power (-10 to -30nm)
- — Expon. (-10 to -30nm) - ° - - Linear (-10 to -30nm)

Figure 3.5.29: Grain boundary etching as a function of root time for ADS-995
hydrothermally attacked at 250°C for 10-15 hours (-10 to -30 nm)

Grain Boundary Etching as a Function of Root Time for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours
(-30 to -50 nm)

 

 

 

o I I I I I
3 3.2 3.4 3.6 3.8 4

Root Time
0 ~30 to -50nm — Power (-30 to -50nm)
- - Expon. (-30 to -50nm) - - - - Linear (-30 to -50nm)

Figure 3.5.30: Grain boundary etching as a function of root time for ADS-995
hydrothermally attacked at 250°C for 10-15 hours (-30 to -50 nm)

183

Grain Boundary Etching as a Function of Root Time for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours
(-50 to -70 nm)

450 a
400 -
350 -
300 -
g 250 1
a 200
9
< 150 a
100 -
50 -
0 .- .
-50 3 3.2 3.4 3.6 3.8 4

Root Time
0 -50 to -70nm — Power (-50 to -70nm)
- - Expon. (-50 to -70nm) - - - - Linear (-50 to -70nm)
Figure 3.5.31: Grain boundary etching as a function of root time for ADS-995
hydrothermally attacked at 250°C for 10—1 5 hours (-50 to —70 nm)

 

 

 

Grain Boundary Etching as a Function of Root Time for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours
(-70 to -90 nm)

350 1
3001
2507
200 a
1501
100 - _
50 -

0 a L, , i

J
-50 3

Area (A2)

 

 

 

O.)
N
0)
p.
9°
0)
00

00
.5

Root Time

A -70 to -90nm — Power (-70 to -90nm)
- — Expon. (-70 to -90nm) - - - - Linear (-70 to -90nm)

Figure 3.5.32: Grain boundary etching as a function of root time for ADS-995
hydrothermally attacked at 250°C for 10-15 hours (-70 to -90 nm)

184

d

Grain Boundary Etching as a Function of Root Time for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours
(-90 to -110 nm)

 

 

 

50 ~
0 * I o t - - I I I I
is ' '3' 2 3.4 3 6 3 8 4
-50 -
Root Time
it -90 to -1 10nm — Power (-90 to -110nm)
- — Expon. (-90 to -110nm) - - - -Linear (-90 to -110nm)

Figure 3.5.3: Grain boundary etching as a function of root time for ADS-995
hydrothermally attacked at 250°C for 10—15 hours (-90 to -110 nm)

Grain Boundary Etching as a Function of Root Time for ADS-995
Hydrothermally Annealed at 250 Celsius for 10 - 15 hours
(-10 to -110nm)

 

 

1000 - D
D
S
100 a Q
A O A
‘25; x
(U 10 -
(D
2 :1
1 d A
)K
0.1 I I I I 1
3 3.2 3.4 3.6 3.8 4
Root Time

0 -10 to ~30nm o -30 to -50nm c1 -50 to -70nm A -70 to -90nm x -90 to -110nm

Figure 3.5.34: Grain boundary etching as a function of root time for ADS-995
hydrothermally attacked at 250°C for 10-15 hours (-10 to -110 nm)

185

Grain Boundary Etching as a chtion of Root Time for AD-94
Hydrothermally Annealed at 250 Celsius for 1 - 2 hours (-5 to -25 nm)

2000 -
1800 a U
1600 ~
1400 -

 

 

0 j I I I I
0.95 1.05 1.15 1.25 1.35 1.45

Root Time

 

U -5 to -25 nm — Power (-5 to -25 nm) — — Expon. (-5 to -25 nm) - - - -Linear (-5 to -25 nm)

Figure 3.5.35: Grain boundary etching as a function of root time for AD-94
hydrotherrnaily attacked at 250°C for 1-2 hours (-5 to —25 nm)

Grain Boundary Etching as a Function of Root Time for AD-94
Hydrothermally Annealed at 250 Celsius for 1 - 2 hours (-25 to —45 nm)

800 -
700 - 0

600 — :- l :— L'_'..'_' ------------------------------------

 

0 I I I I I
0.95 1.05 1.15 1.25 1.35 1.45

Root Time

 

o -25 to -45 nm — Power (-25 to -45 nm)
- - Expon. (-25 to -45 nm) - - - - Linear (-25 to -45 nm)

Figure 3.5.36: Grain boundary etching as a function of root time for AD-94
hydrothermally attacked at 250°C for 1-2 hours (-25 to -45 nm)

186

Grain Boundary Etching as a Function of Root Time for AD-94
Hydrothermally Annealed at 250 Celsius for 1 - 2 hours (-45 to -65 nm)

 

 

0 I I I I
0.95 1.05 1.15 1.25 1.35 1.45
Root Time

 

U -45 to -65 nm -——- Power (-45 to -65 nm)
- - Expon. (-45 to -65 nm) - - - - Linear (-45 to -65 nm)

Figure 3.5.37: Grain boundary etching as a function of root time for AD-94
hydrothermally attacked at 250°C for 1-2 hours (-45 to —65 nm)

Grain Boundary Etching as a Function of Root Time for AD—94
Hydrothermally Annealed at 250 Celsius for 1 - 2 hours (-65 to -85 nm)

300 -
250 -
200 A
150 1
100 -
50 -

0 R-—-

0.95 1.05 1.15 1.25 1.35 1.45

Root Time

Area (A2)

 

 

A -65 to -85 nm — Power (-65 to -85 nm)
- — Expon. (-65 to -85 nm) - - - - Linear (-65 to -85 nm)

Figure 3.5.38: Grain boundary etching as a function of root time for AD-94
hydrothermally attacked at 250°C for 1-2 hours (-65 to -85 nm)

187

Grain Boundary Etching as a Function of Root Time for AD~94
Hydrothermally Annealed at 250 Celsius for 1 ~ 2 hours (~85 to ~105 nm)

350 .
300 1
250 -
200 4
150 7
100 -
50 -

0 x - . - . I. I I I
1
500335 1.05 1.15 1.25 1.35 1.45

Area (A2)

 

 

 

Root Time

1: ~85 to ~105 nm — Power (~85 to ~105 nm)
- - Expon. (~85 to ~105 nm) - - - - Linear (~85 to ~105 nm)

Figure 3.5.39: Grain boundary etching as a function of root time for AD-94
hydrothermally attacked at 250°C for 1-2 hours (~85 to —105 nm)

Grain Boundary Etching as a Function of Root Time for AD-94
Hydrothermally Annealed at 250 Celsius for 1 ~ 2 hours (~5 to ~105 nm)

2000 —
1800 - D
1600 —
1400 -

‘22: 1200 -
E 1000 — a
<

)

800 ‘i o D o

600 '

400 - O

200 ‘ 1:) g A
0 ‘ I I fl I

0.95 1.05 1.15 1.25 1.35 1.45
Root Time

 

 

D ~5 to ~25 nm 0 ~25 to ~45 nm D ~45 to ~65 nm A ~65 to ~85 nm :1: ~85 to ~105 nm

Figure 3.5.40: Grain boundary etching as a function of root time for AD~94
hydrothermally attacked at 250°C for 1-2 hours (~5 to 105 nm)

188

3.6 Electron Dispersive Study

The electron dispersive study (EDS) for both the ADS-995 and the AD-94
was performed at 20 W with a 17 mm working distance, exposure times were
conducted in real time for 10 minutes for each specimens with representative
images captured in Figures 3.6. The EDS 0f the ADS-995 and AD-94 specimens
hydrothermally attacked for 0, and 5 hours displayed no appreciable quantities of
elements besides aluminum. Since neither material is 100% pure and therefore
contains some impurities, the impurity levels of the individual constituents are
believed to be below the level, which is detectable by the available EDS array.

The gold, which was found on the surface of the ADS-995 specimens, is due
to the gold coating used for the analysis. This gold coating may be inhibiting the

detection of the characteristic x-rays from the various impurity elements.

189

     
  

 

Figure 3.6.1: EDS analysis of Coors ADS-995 with no hydrothermal attack.

Figure 3.6.2: EDS analysis of Coors AD-94 with no hydrothermal attack.

  

Figure 3.6.3: EDS analysis of Coors ADS-995 hydrothermally attacked for 5
hours at 250°C.

 

Figure 3.6.4: EDSanalysis of Coors AD-94 hydrothermally attacked for 5 hours
at 250°C.

191

3.7 Scanning Electron Microscope Analysis

Scanning electron microscopy (SEM) was performed for ADS-995 and AD~
94 for the O and 5~h0ur specimens. The SEM images were imaged with a 25 kV
beam and a working distance of 10 mm.

The low magnification SEM images of non-hydrothermally attacked ADS-995
(Figures 3.7.1 a-b) show a surface with many (<50) small (<3 um) surface
depressions (Figure 3.7.1 a). The backscattered electron (BSE) image (Figure
3.7.1 b) reveals many pores not apparent in the SEM image (Figure 3.7.1a). At
higher magnifications (Figures 3.7.1c-d), the pits are still visible in the SEM
image (Figure 3.7.10) and additional flaws appear in the BSE image (Figure
3.7.1 d).

Non-hydrothermally attacked AD-94 (Figure 3.7.2a,c) has fewer (<10)
surface depressions that are larger (<15 pm) than the pits in the ADS-995
(Figure 3.7.1a). The corresponding BSE images (Figure 3.7.1b,d), reveals more
defects than the SEM image, which was also the case in the ADS-995 (Figures
3.7.1 b,d).

Following 5 hours of hydrothermal corrosion, the ADS-995 surface (Figure
3.7.3) displays more surface flaws than the ADS-995 that had not been
hydrothermally attacked (Figure 3.7.1). At higher magnification (Figure 3.7.3c,d),
the surface has been textured by the hydrothermal corrosion.

A 5-h0ur hydrothermal corrosion of AD-94 (Figures 3.7.4) has more (>30)
surface flaws, than, the non-hydrothermally attacked AD~94 specimen (Figure

3.7.2). Additionally, a surface deposit has collected on the substrate.

192

I Ilium I; . _ '

Figure 3.7.1: Scanning electron image of Coors ADS-995, with no hydrothermal
treatment, polished to a 1 pm finish; a) Low magnification SEM image, b) Low
magnification BSE image, c) High magnification SEM image, d) High
magnification BSE image.

 

193

 

Figure 3.7.1: Continued.

194

w

I ”“1".“
, ,.

 

Figure 3.7.2: Scanning electron image of Coors AD-94, with no hydrothermal
treatment, polished to a 1 pm finish; a) Low magnification SEM image, b) Low
magnification BSE image, 0) High magnification SEM image, d) High
magnification BSE image.

195

 

Figure 3.7.2: Continued.

196

 

Figure 3.7.3: Scanning electron image of Coors ADS-995, hydrothermally
attacked for 5 hours at 250°C; a) Low magnification SEM image, b) Low
magnification BSE image, c) High magnification SEM image, d) High
magnification BSE image.

197

 

 

Figure 3.7.3: Continued.

198

 

 

 

b
Figure 3.7.4: Scanning electron image of Coors AD-94, hydrothermally attacked
for 5 hours at 250°C; a) Low magnification SEM image, b) Low magnification
BSE image, 0) High magnification SEM image, d) High magnification BSE image.

199

 

: Continued.

Figure 3.7.4

200

3.8 Surface Morphology

For AD~94 and ADS-995 the Figures 3.1 show variations in the surface
morphology as a function of hydrotherrnai corrosion time. initially the surface is
devoid of any scale but as the hydrothermal corrosion progresses the scale
deposits on the surface.

The majority of the surface scale (Figures 3.1) that was deposited on the
substrates (AD-94 and ADS-995) displayed a mixed morphology of {111} and
{110} planes, which appears as a cauliflower-type structure (Figure. 3.1.16b and
Figure 3.1.28b) A few structures showed the {100} morphology.

The differing morphologies are similar to the synthesis of diamond films
though microwave discharge [38]. Depending on the temperature and the
atmospheric ratios of methane and hydrogen diamond films will develop differing
surface morphologies appearing as: cauliflower-type structures at low
temperatures and low to high ratios of methane to hydrogen, {111}-type facets at
high temperatures and low to high ratios with an intermediate {100}-type facet
developing at moderate temperatures and ratios.

The surface film deposited on the AD-94 and ADS-995 experiences various
morphologies depending on the length of time they were subjected to the
hydrothermal corrosion. A cauliflower-type structure was evident in both AD-94
(Figure 3.1.16b) and ADS-995 (Figure 3.1.28b) while different specimens AD-94

(Figure 3.1.120) and ADS-995 (Figure 3.1.27b) showed {100}-type facet.

201

3.9 Summary and Conclusions

ADS-995 and AD-94 structural alumina ceramics were hydrothermally
corroded for up to 50 hours at 250°C. The ADS-995 was hydrothermally
corroded for 0, V2, 1, 5, 10, 12%, 15, 20, 25, 30, 35, 40, 45, and 50 hours with
times of interest being 0, 5, 10-15 and 50 hours of corrosion. The AD-94 was
hydrothermally corroded in half hour increments between 0 and 7.5 hours, 10
and 50 hours with times of interest being 0, 1-2, 5, 10 and 50 hours. The given
periods are of interest because they correspond to the control specimens (0
hours), the most heavily corroded specimens for this study (50 hours), the ranges
where the hydrothermal corrosion went from surface pitting to the deposition of a
surface layer and reference times of 5 and 10 hours.

For the hydrothemial corrosion of the present study, the entire surface is
simultaneously being corroded by the exact same conditions. The surface
uniformity is best expressed in Figure 3.7.4a-d, where the long-range order of the
corrosion may be seen. Viewing the surface at lower magnification (Figures 3.7)
shows the difficulty in locating and imaging a 100 - 10000 um2 area that was
nominally flat.

The hydrothermal environment initially has little to no effect on the alumina
substrates (Figures 3.1.3-4). As the material is subjected to the hydrothermal
environment, for longer periods the surface begins to be corroded (Figures 3.1.5-
6) evidenced as pitting and etching of the grain boundaries. Eventually the
grains will be discernable form a once finely polished substrate (Figures 3.1.8,

25). The etching process is progressing in a closed environment so the etched

202

material, which comes from the grain boundary phases, is placed into the water
solution. As the time in the hydrothermal environment proceeds the compounds
in the solution precipitate out onto the substrate surface (Figures 3.1.9, 26) as
evidenced through the mass gain (4.66%) experienced by the AD-94 specimens
following 100 hours of treatment. Furthermore, the growth of the precipitate
phase through AFM at different stages of evolution is depicted in Figures 3.1,
with an electron image depiction in Figures 3.7.

The X-ray and EDS analyses of the specimens were inconclusive. The X-
ray peaks present in the non-hydrothennaliy attacked specimens and the
hydrothermally attacked specimens could not be distinguished from one another.
Meanwhile the EDS only detected alumina in the specimens. Both the E08 and
the X-ray analyses should have detected changes in the specimen compositions
as there was a 4.66% mass gain experienced by the AD~94 specimens and the
evident growth of a surface layer on the specimens as evidenced in Figures 3.1
and Figures 3.7.

Additionally, with the overall roughness data of the alumina specimens may
be approximated by a series of normal distributions. However, analysis of the
etched areas as function of the pit depth will require extremal-value statistics:
Gumbel or Cauchy. Sections 20 nm thick, were analyzed to the maximum pit
depth of the specimens (Figures3.5). ADS-995 and AD-94 specimens had
similar relationships where the hydrothermal corrosion progresses from etching

to depositing a surface layer (Tables 3.51 ~4). As time progresses the ADS-995

203

and AD-94 specimens surface corrosion has a mixed-mode relationship with
square root of time.

The surface of the alumina progressively becomes more textured as
corrosion time increases. Texturing of some form is expected, as surface
degradation is the purpose of an etching process. What is of interest is that the
surface corrosion yields a surface that appears as many overlapping scales. The
scales are composed of polyhedra in the form of {100}, {110} and {111} with the
majority of the scales having a mixed {111} and {110} morphology. These
polyhedra occlude increasingly more of the surface as the corrosion progresses
and eventually may transform into the cube structures observed by Day in 96%
pure AlSiMag 614 (Section 1.8) or diaspore and boehmite mentioned by
Sinharoy et. ai. (Section 1.7). Although boehmite, diaspore and wairakite were
not found in the current study, a scale did develop on the specimen surfaces.

The hydrothermal environment degrades the surface of polished aluminas
(ADS-995 and AD-94) at temperatures well below the maximum atmospheric use
temperature (Table 2.1.1) of the ceramic. Thus, there should be a maximum use
temperature for the hydrothermal environment. The progression of this attack
follows extremal-value statistics, which have already been applied to failure
analysis. Understanding, the mechanics and properties of hydrothermal
corrosion, is fundamental in determining the service life or schedule necessary to

prevent failure.

204

3.10 Future Work

Future work in the hydrothermal corrosion of ceramics should encompass an
in-depth X-ray analysis of the ADS-995 and AD~94 specimens hydrothermally
corroded for 0 to 50 hours. Furthermore, an analysis of the scale development
from 10 to 15 hours in ADS-995 and from 1 to 2 hours in AD-94 in 10-minute
intervals (Le. 60, 70, 80 minutes for AD-94) should be completed. in the current
work an analysis was made of the corroded surface area present at a given
depth (Section 3.5), a study of the total pit should also be conducted. Finally, the
exact same study conducted in the present work should be done for a non-oxide

ceramic to compare how different material systems.

205

REFERENCES

206

REFERENCES

. William D. Callister, “Materials Science and Engineering: An introduction,”
Wiley, New York, pages 70-159, 2000.

. W. D. Kingery, H. K. Bowen and D. R. Uhimann, “introduction to Ceramics,”
Wiley, New York, pages 66-70, 188-201, 714, 1976.

. Michel Barsoum, “Fundamentals of Ceramics,” McGraw-Hill, New York,
pages 365-503, 1997.

. J. F. Lynch, Ruderer C. G., and Duckworth W. H., “Engineering Properties of
Selected Ceramic Materials,” American Ceramic Society incorporated,
Columbus, pages 541-28, 1966.

. Kohei Oda and Tetsuo Yoshio, “Hydrothermal Corrosion of Alumina
Ceramics,” Journal of the American Ceramic Society, Volume 80, Number 12,
pages 3222-3236.

. S. Kiotaoka, Y. Yamaguchi and Y. Takahashi, “Tribological characteristics of
a—Alumina in High-Temperature Water,” Journal of the American Ceramic
Society, volume 75,number 11, pages 3075-80, 1992.

. S. Ono, K. Suzuki, M. Kumagai, and K. Hoshi, “Corrosion and Elution
behaviors in Al203 Ceramics in High-Temperature Demineralized Water,”
Ziryo-to-Kankyo, volume 40, pages 655-60, 1991.

. David Avigdor and S. D. Brown, “Delayed failure of a porous alumina,”
Journal of the American Ceramic Society, volume 61, number 3-4, pages 97-
99, 1978.

. Delbert E. Day, “Reaction of alumina ceramics with saturated steam,”
Ceramic Bulletin, volume 62, number 6, pages 624-631, 1982.

10. Samar Sinharoy, Levenson L., Ballard W. and Day D. “Surface segregation of

calcium in dense alumina to steam and steam-CO,” Ceramic Bulletin, volume
57 number 2, pages 231-233, 1978.

11.Fi. M. Barrer and P. J. Denny, Hydrothermal Chemistry of the Silicates: X, A

Partial Study of the Field CaOoAl20304Si0202H20,” Journal of the Chemical
Society (London), pages 983-1000, 1961.

12.0. S. Coombs, “X-Fiay Observations on Wairakite and Non-Cubic Analcine,”

Mineral Magazine, volume 30, number 230, pages 699-708, 1955.

207

13. A. Steiner, ”Wairakite, the Calcium Analogue of Analcine,” Mineral Magazine,
volume 30, number 230, pages 691-98, 1955.

14. D. Bahat, “Compositional Study and Properties of Characterization of Alkaline
Earth Feldspar Glasses and Glass-Ceramics,” Journal of Materials Science,
volume 4, number 10, pages 855-60, 1969.

15. R. l. Taylor, J. P. Coad and R. J. Brook, “Grain boundary segregation in
A1203," Journal of the American Ceramic Society, volume 57, number 12,
pages 539-540, 1974.

16. H. L. Marcus and M. E. Fine, “Grain boundary segregation in MgO-doped
Al203,” Journal of the American Ceramic Society, volume 55, number 11,
pages 568- 570, 1972.

17. F. A. Kroeger, “Chemistry of imperfect Crystals,” North-Holland Publishing
Company, Amsterdam, Page 776, 1964.

18.W. D. Kingery, “Plausible Concepts Necessary and Sufficient for
interpretation of Ceramic Grain Boundary Phenomena: 1” Journal of the
American Ceramic Society, volume 57, number 1, pages 1-12, 1974.

19. P. P. Budnikov and F. l. Chatikonov, “influence of High Pressure Water Vapor
on Some Properties of Corundum Materials,” Silikattechnik, volume 18,
number 8, pages 243-48, 1967.

20. Hideo Hirayama, T. Kawakubo, A. Goto and T. Kaneko, “Corrosion behavior
of silicon carbide in 290°C water,” Journal of the American Ceramic Society,
Volume 72, number 11, pages 2049-2053, 1989.

21 .Yu Gogotsi and M. Yoshimura, “ Degradation of SiC-based fibres in high-
temperature, high-pressure water,” Joumai of Materials Science, volume 13,
number 6, pages 395-399, 1994.

22.0. J. Pysher, K. C. Goretta, R. S. Hodder and R. E. Tressier, “Strengths of
Ceramic Fibers at Elevated-Temperatures,” Journal of the American Ceramic
Society, volume 72, number 2, pages 284 - 288, 1989.

23. R. E. Tressier, N.Y. Jia, J. Zheng and H. Takahashi, in Proceedings of the tst
international Symposium on the Science of Engineering Ceramics by S.
Kimura and K.Niihara, page 167, 1991.

24.T. Yamamura, T. lshikawa, M. Shibuya, T. Nagasawa and K. Okamura, in

Proceedings of the 1st Japan international SAMPE Symposium, page 1084,
1 989.

208

25.Yu G. Gogotsi and V. A. Lavrenko, “Corrosion of high-performance ceramics,”
Springer Veriag, Berlin, pages 40-45, 1992.

26.Ciaude Contet, Kase J. l., Noma T., Yoshimura M., and Shigeyuki Somiya,
“Hydrothermal oxidation of Si3N4 powder,” Joumai of Materials Science
Letters, volume 6, pages 963-964, 1987.

27.S.C Singhal, Journal of American Ceramic Society, volume 59, number 1-2,
pages 81-82, 1976.

28.Yung-Kuo Chao and D. E. Clark, “Hydrothermal corrosion of lithium disilicate
glass and glass ceramics,” Glass Technology, Volume 25, number 3, pages
152-156, 1984.

29. BE. Clark and EC. Etheridge, American Ceramic Society Bulletin, volume
60, number 6, page 646, 1981.

30. R.J. Charles, Journal of Applied Physics, Volume 29, Number 11, page 1549,
1958.

31 .Coors Technical Ceramics Website, www.coorstekcom, 2001.

32. Parr instrument Company, “Operating Instructions: Parr Acid Digestion
Bombs,” No. 249M, Parr instrument Company, December 1997.

33. Digital instruments Nanoscope iii: Atomic Force Microscope intruction Manual
Version 2.2, July 8, 1992.

34. Digital instruments Website, www.c_li.com, 2001.

35.JCPDS--international Centre for Diffraction Data, “Powder diffraction file:
Sets 1-43,” American Society for Testing and Materials. Swarthmore, Pa.:
The Centre, 1960.

36. 8D. Culiity, “Elements of X-ray Diffraction: Second Edition,” Addison-Wesley
Publishing Company, London, page 295, 1978.

37.T. Shibata, “1996 W.R. Whitney Award Lecture: Statistical and Stochastic
Approaches to Localized Corrosion,” Corrosion, volume 52, number 11,
pages 813-830, 1996.

38.Saeid Khatami, “Controlled synthesis of diamond films using a microwave

discharge (non-equilibrium plasma),” Michigan State University, Thesis,
pages 339-455, 1997.

209

39.John Frederick Nye, “Physical Properties of Crystals: Their Representation by
Tensors and Matrices,” Clarendon Press, Oxford, pages 141-281, 1984.

40.J. B. Wachtman Jr., “Elastic Deformation of Ceramics and Other Refractory
Materials,” pages 139 to 168, in Mechanical and Thermal Properties of
Ceramics: Proceedings of a Symposium, Edited by JB Wachtman Jr., US
Government Printing Office, Washington DC, page 146, 1969.

210

APPENDIX

211

APPENDIX A: Slip Systems for Alumina
Optimaliy there are six ({10—11}o<11~20>) independent slip systems in a
hexagonal (a = b at c, a = B = 90°, y=120°) close-packed structure, although most
hexagonal close-packed structures have only three ({10-10}o<11-20> or
{0001}-<11-20>) independent slip systems [1]. In the case of alumina, there are

only two independent slip systems {0001}-<11-20> [2]. The structure of alumina
is not close-packed but a hexagonal close-packed sub-lattice with two-thirds of
the octahedral sites filled as the other sub-lattice [2]. Normally three independent
slip systems are associated with the {0001} slip plane, alumina has “lost” an
independent slip system.

Unfortunately, the hexagonal matrix is a rhombohedral system [39]. The
other rhombohedral system is trigonal (a = b at c, or = B = 90°, y=120°) [39]. Both
of these crystallographic systems use the four coordinate Miller-Bravais indices
{hkil}. if the material in question is not pure then the hexagonal close-packed
lattice becomes a hexagonal sub-lattice, which in some cases may be
represented as a trigonal structure. Alumina crystals are capable of being
represented as trigonal or hexagonal [40]. Crystallographicaily, the major
difference between the trigonal and hexagonal systems is that the trigonal
structure has a single, 3-foid axis of rotation, whereas the hexagonal structure
has a single, 6-fold axis of rotation [39]. in both cases, the axis of rotation is
parallel to the z-axis [39].

As a result, most alumina data is recorded in the hexagonal system but

some is reported in the trigonal system. The system in which the data was

212

recorded determined how the specific analysis was performed. We will treat the
hexagonal system as the norm for alumina and we will note any data based on

the trigonal system.

213

 

 

   

APPENDIX B: Elastic Moduli of Alumina

Much research was done to obtain the anisotropic elastic moduli of alumina
crystals. The single crystal elastic constants, compliance and stiffness
coefficients have been reported in Table B1 in the trigonal system [40]. The
compliance data was then placed within the symmetric elasticity matrix for a
trigonal body, Table 8.2. The compliances may then be placed within the
appropriate expression (Equation B.1) for the reciprocal of the elastic modulus for
a trigonal structure of the appropriate class [39]. The compliances have been
denoted as s,,- and the 1 values refer to the value of the respective Miller indices.
Given the trigonal lattice parameters (a = b at c), the Miller-Bravais indices of
interest are [0001] and either [1010] or [0110], where the boldface denotes a
negative index. For use in equation 8.1 the relative indices are: [001] and either
[100] or [010]. In for the non-basal direction, [001], equation 1.20.1 has been
reduced into 3.2 using the appropriate values for l. The resultant elastic modulus
in the non-basal direction was 460.83 GPa. Furthermore, 33 is the reduction of
8.1 for either [100] or [010] basal directions. As these directions are equivalent,

their elastic modulus was 425.53 GPa.

214

Table 8.1: Single crystal elastic constants of alumina in the monal system [40]

 

1.1 Q . 514; B E .1_4 Units
Compliance 2.35 2.17 6.94 -0.716 -0.364 0.489 10'” m2/N
Stiffness 49.7 49.8 14.7 16.4 11.1 -2.35 1010 N/m2

 

Table 3.2: The reduced trigonal elasticity matrix representing crystallographic
classes 32, -3m and 3m, using compliance constants the matrix is symmetric
about the main diagonal [39]

 

 

 

 

S11 S12 S13 S14 0 0
S11 523 " S14 0 0
$33 0 0 0
S44 0 0
S44 '2(S14)
2(11-12)
— l B ‘l
(1 —1_,2)2 s” +13%33 + 1,2 (1 —132 )(2513 + S“) + 21,1,(31,2 -1§s,,) °
1
E = 4 8.2
’3 533
1
B3

— (1—z§)2s,,

215

APPENDIX C: Sectioning Data

Table C.1: Area present for the sectioning data of ADS-995

 

 

 

A08995 ADS995 ADS995
Depth Range 10H -002 12H5 15H0
(nm) (A2) (A2) (A2)

-10.20 to -30.20 418.32 99.14 665.70
-30.20 to -50.20 47.00 128.90 286.52
-50.20 to -70.20 3.78 74.04 312.92
-70.20 to -90.20 1.03 41.58 245.36
-90.20 to -110.20 0.31 16.33 181.73
-110.20 to -130.20 0.31 117.00
-160.20 to -180.20 70.04
-210.20 to -230.20 142.10
Root time (hoursy“) 3.16 3.54 3.87

 

Table 02: Area present for the sectioninqdata of AD-94

 

 

 

 

AD94 AD94 A094 A094
Depth Range 100 -000 105 -001 105 -012 200 -001
(nm) (A2) (A2) (A2) (A2)
-5.4 to -25.4 1742.02 1202.93 319.02 1009.06
-25.4 to -45.4 758.17 501.02 144.12 710.33
-45.4 to -65.4 153.31 186.12 141.49 339.32
-65.4 to -85.4 4.92 66.99 147.82 163.88
-85.4 to -105.4 0.34 53.33 86.17 136.91
-155.4 to -175.4 29.53 61.80 83.31
-255.4 to -275.4 16.67 33.80 44.29
-355.4 to -375.4 6.94 10.95 96.44
-455.4 to -475.4 0.88
Root time Lhoursy”) 1.0000 1.2247 1.41

 

216