Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Process engineering of thick dielectric films by Chemically Bonded Composite Sol-Gel Kim, Hyungkeun 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-267387.pdf [ 25.2MB ]
Metadata
JSON: 831-1.0078497.json
JSON-LD: 831-1.0078497-ld.json
RDF/XML (Pretty): 831-1.0078497-rdf.xml
RDF/JSON: 831-1.0078497-rdf.json
Turtle: 831-1.0078497-turtle.txt
N-Triples: 831-1.0078497-rdf-ntriples.txt
Original Record: 831-1.0078497-source.json
Full Text
831-1.0078497-fulltext.txt
Citation
831-1.0078497.ris

Full Text

PROCESS ENGINEERING OF THICK DIELECTRIC FILMS BY CHEMICALLY BONDED COMPOSITE SOL-GEL by HYUNGKEUN KIM B.Eng., Myongji University, Korea, 1995 M.S., Myongji University, Korea, 1997 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Materials Engineering) THE UNIVERSITY OF BRITISH COLUMBIA April 2007 © Hyungkeun Kim, 2007 ABSTRACT This research explores new processing methods to decrease residual stress in ceramic films on metallic substrates, and thus to prevent large-scale cracking in the films.  The system of particular concern is alumina-based ceramic coating on aluminum alloy, wherein coefficient of  thermal expansion (CTE) of  the ceramic is about 3x smaller than that of  the alloy. The specific  goal was to achieve relatively thick (~0.2mm) and substantially low density of cracks in dielectric films  of  alumina-based ceramic on aluminum alloy AA5052, by Chemically Bonded Composite Sol-Gel (CB-CSG) process. The principal strategies undertaken in the materials process engineering involved: (i) multi-layer film  deposition - to introduce the intermediate steps of  stress relaxation; (ii) composite sol-gel slurries with bi-modal particle size distribution - to decrease the overall process temperature, and to decrease film  strain during thermal treatment as well as to increase the density and stability of  the slurries; (iii) chemical bonding of  the film  through phosphating - to further  decrease the process temperature to the level of  300°C; and (iv) introduction of  organic-phase (citric acid) derived bond coats at the interface  between the AA5052 substrate and the ceramic coating -to achieve residual stress relaxation through viscoelastic deformation  of  the bond coat. The coatings were processed through spray deposition of  consecutive -40 fj,m  thick layers, heat-treated at 300°C for  10 min after  each deposition. Two size fractions  of  alumina powders (average size of  0.5 |am of  "fine"  and 3 |am of  "coarse") were used in formulation  of the Composite Sol-Gel (CSG) slurry, and the fine/coarse  particle content was optimized based on slurry viscosity and stability, as well as properties of  the final  coating. The coatings were characterized for  microstructure, residual stresses and dielectric strength, as a function of  the process parameters. The most important finding  of  this work is that it is possible to deposit thick ceramic films  on aluminum alloy substrates, if  all four  processing strategies listed above are implemented simultaneously. In particular, the citric acid - derived organo-ceramic bond coats seem to play an important role in relaxing residual stresses resulting from differential  thermal contraction and expansion. It is concluded that the viscoelastically deforming  organo-ceramic bond coat helps to relax residual stresses in the coating layers due to differential  thermal contraction/expansion, and thus allows deposition of  films  of  up to 200 |am thickness. Dielectric strength of  the CB-CSG alumina coated AA5052 aluminum reached a maximum of  15±1 kV/mm for  the first layer, and subsequently decreased to 10.5±1 kV/mm. It is believed that this decrease in the dielectric strength after  the first  layer is caused by increased density of  cracks in the coating, as evidenced by decrease of  the residual stress in the coatings. Table of  Contents Abstract ii Table of  Contents iv List of  Tables .. vi List of  Figures vii List of  Greek Symbols xii List of  Roman Symbols xiii List of  Abbreviations xv Acknowledgement xvi Chapter 1 Introduction 1 Chapter 2 Literature Review 5 2.1 Ceramic Coating Processing 5 2.1.1 Sol-Gel Processing 5 2.1.2 Composite Sol-Gel 7 2.1.3 Packing of  Particles 9 2.1.4 Chemically Bonded Composite Sol-Gel (CB-CSG) 12 2.1.4.1 Transformation  of  Phosphates 15 2.2 Citric Acid in Ceramic Processing 16 2.2.1 Citric Acid 17 2.2.2 Role of  Citric Acid in Aqueous Alumina Suspensions 18 2.2.3 NMR for  Testing Materials 20 2.3 Dielectric Properties of  Ceramics 21 2.3.1 Ceramic Insulators 21 2.3.2 Dielectric Strength 23 2.4 Residual Stresses in Ceramic Coatings 25 2.4.1 Mitigation of  Residual Stresses in Coatings 29 2.4.2 Determination of  Residual Stress in Coating Systems 31 2.4.2.1 Determination of  Residual Stress by Curvature Method 31 2.4.2.2 Determination of  Residual Stress by X-ray Diffraction  35 Chapter 3 Scope and Objectives 39 3.1 Scope of  the Investigation 39 3.2 Objectives 40 Chapter 4 Experimental Methodology 42 4.1 Processing of  Alumina Composite Sol-Gel 42 4.1.1 Alumina Composite Sols 42 4.1.2 Characterization of  Alumina Composite Sols 43 4.1.3 Thermal Behavior of  CA 44 4.2 Processing of  CB-CSG Coatings for  Thick Dielectric Films 45 4.2.1 Surface  Preparation of  AA5052 Substrate..... 46 4.2.2 Spray Coating 47 4.2.3 Chemical Bonding of  the Coatings by Phosphating 49 4.2.4 Heat Treatment 50 4.3 Characterization of  CB-CSG Coatings for  Thick Dielectric Films 51 4.3.1 Dielectric Strength Measurements 51 4.3.2 Curvature Measurements 52 4.3.3 Residual Stress Determination by X-ray Diffraction  53 4.3.4 Young's Modulus Measurement of  Alumina CB-CSG Coatings 53 Chapter 5 Study of  Alumina Composite Sol-Gel for  Thick Dielectric Films 59 5.1 Characterization of  Alumina Composite Sol 59 5.1.1 Viscosity of  Alumina Composite Sols with Various Concentrations of  Fine Alumina Particles 59 5.1.2 Particle Size Distribution in Alumina CSG 65 5.1.3 Settling of  Alumina Particles in the Composite Sol 70 5.2 Characterization of  Alumina Composite Sol-Gel 79 5.2.1 Morphology of  Alumina Composite Sol-Gel with Various Concentration of Fine Size Alumina Particles 79 5.2.2 Effect  of  the Specific  Surface  Area on Density and Viscosity of  Alumina Composite Sol-Gel 83 5.3 Summary 85 Chapter 6 CB-CSG Coatings with Citric Acid-Derived Bond Coat 86 6.1 Cross Sectional Morphology of  CB-CSG Coatings 86 6.2 Effects  of  Citric Acid on CB-CSG Coatings 91 6.2.1 Behavior of  Citric Acid at Elevated Temperatures 92 6.2.2 Role of  Citric Acid in CB-CSG Coatings 97 6.2.3 Stress Mitigation for  CB-CSG Coated Aluminum 106 6.3 Determination of  Residual Stresses in CB-CSG Alumina Coated Aluminum alloy (AA5052) 112 6.3.1 Curvature of  CB-CSG Alumina Coated AA5052 Alloy System 113 6.3.2 Young's Modulus of  CB-CSG Alumina Coated AA5052 115 6.3.3 Strain Measurement by X-ray Diffraction  (XRD) 119 6.3.4 Average Residual Stresses in the Alumina CB-CSG Coating Layer on Aluminum alloy (AA5052) 122 6.4 Electrical Properties of  Alumina CB-CSG Coatings 125 6.4.1 Dielectric Strength of  CB-CSG Alumina Coatings 126 6.4.1.1 Effect  of  the shape of  an electrode : 126 6.4.1.2 Dielectric Strength of  CB-CSG Alumina Coated AA5052 128 6.5 Summary 132 Chapter 7 Conclusions 134 Chapter 8 Recommendations for  Future Work 139 References  141 Appendix 150 List of  Tables Table 1-1 Comparison of  a current stainless steel (430) and proposed aluminum alloy (5052) metal substrate for  heating elements [1.24] 4 Table 2.3-1 Thermal conductivity of  single and complex oxides [1.1] 22 Table 2.3-2 Thermal shock resistance (W/m) for  ceramic materials [2.60] 23 Table 4.1-1 Preparation of  the alumina composite sol slurry by the concentration of  fine alumina particles, ~0.5 fj,m,  in coarse alumina particles, ~3 |j.m 42 Table 4.3-1 Example of  sample geometries and strain used to obtain the Young's modulus by using Dynamic Mechanical Thermal Analyzer, DMTA 56 Table 5.1-1 Summary of  particle size distribution of  the alumina composite sols after  24h ball-milling 69 List of  Figures Fig. 1 -1 High efficiency  heater and its application for  water boiling 4 Fig.2.1 -1 Reaction of  the hydrated surface  of  alumina with H3O+ or OH".[2.13,2.14] 7 Fig.2.1-2 Schematic of  sol particles gluing coated particles into a coherent 3 dimensional structure [2.26] 8 Fig.2.1-3 Packing of  spheres of  different  sizes; (a) theoretical packing and (b) practical packing [2.29] 9 Fig.2.1-4 Calculated and experimental total porosity for  vibrated, two-component mixtures of  tabular alumina fines  (-200+325 mesh) and three different  coarse fractions  [2.31] 10 Fig.2.1-5 Apparent density graded and fired  mixtures of  silica rock. The solid circle indicates the highest density of  silica (-1.74 g/cc) [2.29] 11 Fig.2.1-6 Schematic of  interactions between alumina particles or alumina sols and phosphates in a CB-CSG system [2.33] 14 Fig.2.2-1 Structural formula  of  (a) citric acid and (b) tricarballylic acid [2.44] 18 Fig.2.2-2 Schematic of  hydroxyl group on the surface  of  alumina sol exchange for carboxyl group in a citric acid containing aqueous suspension system 19 Fig.2.4-1 Model for  the expansion of  (a) three rods with different  coefficients  of  thermal expansion [2.74] and (b) alumina CB-CSG multi-coating layers with viscoelastic organic bond coat, as in the present study 27 Fig.2.4-2 Thermal expansion of  (a) stainless-steel-based and (b) NiCrAl-based materials [2.86] 30 Fig.2.4-3 Behavior of  a ceramic coated metal substrate in thermal cycling, (a) stress-free initial stage at elevated temperature,7"/, (b) ceramic coated substrate experiences differential  shrinkage after  cooling to temperature, T2<Tj.  (c) Bending is allowed to balance the stresses in ceramic coating and a metal substrate [2.88] 32 Fig.2.4-4 Profile  of  a coating structure [2.87] 33 Fig.2.4-5 Directions of  stress and strain (ax and cry parallel to the surface  and a z normal to the surface)  at the surface  of  a stressed material 37 Fig.2.4-6 Comparison of  stresses of  Mo coatings on steel measured by (a) X-ray diffraction  and (b) curvature measurement [2.93] 38 Fig.4.2-1 Cross sectional SEM images for  (a) non-sand blasted (b) sand blasted AA5052 vm substrate 46 Fig.4.2-2 Profilometer  of  the sand blasted aluminum substrate used in this work and schematic of  determination of  roughness value, Ra, using a mean line [4.3]. The average roughness, Ra of  the substrate shown above is 5.87 ^m 47 Fig.4.2-3 Spray coating setup. Spraying has been repeated 5 times until the desired thickness of  coating (>200 |xm) is reached 48 Fig.4.2-4 Schematic of  the Chemically Bonded Composite Sol-Gel [2.33] 49 Fig.4.3-1 Setup of  the hipot tester for  the breakdown voltage measurement 51 Fig.4.3-2 Sample location in Talysurf  Profilomer.  52 Fig.4.3-3 Sample location in Dynamic Mechanical Thermal Analyzer, DMTA for  a stiffness  and Young's modulus measurements for  beam samples 54 Fig.4.3-4 DMA for  measuring average Young's modulus at various temperatures in 3-point bending 57 Fig.5.1-1 Viscosity of  alumina composite sols ball-milled for  12 hours. The concentration in the legend box stands for  the amount of  fine  alumina particles (~0.5(j.m) in the mixtures with coarse ones (~3|J.m) 60 Fig.5.1 -2 Viscosity of  alumina composite sols ball-milled for  24 hours. Dot lines show the range of  the viscosity (-200-400 cp) suitable for  spray coatings 62 Fig.5.1-3 Viscosity of  alumina composite sol ball-milled for  48 hours. Dot lines show the range of  the viscosity (200-400 cp) suitable for  spray coatings. The concentration of  A40 offers  longest work time while A45 and A50 are beyond the workable range. A30 and A35 are below the viscosity range suitable for spray coatings 63 Fig.5.1-4 Viscosity and specific  surface  area of  24h ball-milled mixed CSG with varying concentration of  the fine  particles 64 Fig.5.1-5 Particle size distribution of  24h ball-milled alumina composite sols with single particle size fraction:  (a) fine  alumina particles and (b) coarse alumina particles (submicron range in the inset) 66 Fig.5.1-6 Particle size distribution of  the alumina composite sols with varying concentrations of  the fine  particles, after  24h ball-milling, (a) A30, (b) A3 5, (c) A40, (d) A45, and (e) A50 67 Fig.5.1-7 Prediction of  the velocity of  settling alumina particles in the alumina sol suspension in terms of  the radius of  the particles 70 Fig.5.1-8 Settling patterns for  fine  (-0.5 |xm) and coarse (-3 |j.m) alumina particles in alumina sol suspension with a ball-milling of  (a) 0 min (b) 30 min (c) 5h, and (d) 12h. Left  2 cylinders represent fine  alumina particles (with 12h and 24h ball-milling, respectively) while right 2cylinders represent coarse alumina particles (with 12h and 24h ball-milling, respectively) in all pictures (a-d) 72 Fig.5.1-9 The rate of  settlement of  two different  alumina particles, -0.5 |xm and - 3 |J.m in alumina suspension of  viscosity of  10-20 cp. Slurries were ball-milled for  12 h and 24 h 73 Fig.5.1-10 Settlement of  alumina composite sol after  12h ball-milling. Pictures were taken (a) right after  ball-milling, (b) after  lhr, (c) 12hrs, and (d) 24hrs after  ball milling stop. The samples are A30, A35, A40, A45, and A50 from  left  to right in each picture. Vertical arrows in (d) indicate the magnitude of  separation of  the alumina sol and alumina particles 74 Fig.5.1-11 Rate of  settlement of  mixed alumina particles dispersed in alumina sol after  12 h ball-milling as a function  of  (a) time and (b) concentrations of  fine  alumina particles 76 Fig.5.1-12 Settlement pattern of  mixed alumina particles after  24h ball-milling. Pictures were taken (a) right after  ball-milling, (b) after  lh, (c) after  12h and (d) after  24h from  the ball milling stop. The samples are A30, A35, A40, A45, and A50 from the left  in each picture. Vertical arrows in (d) indicate the separation of  the alumina sols and particles. Measurable settling was observed in A30 and A35 only 77 Fig.5.1-13 Rate of  settlement of  alumina CSG suspensions after  24 h ball-milling 78 Fig.5.2-1 Fracture surfaces  of  pelletized alumina samples, (a) 100% coarse alumina particles, - 3 |am and (b) 100% fine  alumina particles, -0.5 |xm 80 Fig.5.2-2 Microstructures of  alumina CSG with the concentrations of  (a) A30, (b) A35, (c) A40, (d) A45, and (e) A50 of  fine  alumina particles (~0.5|am) in gel-cast samples (all in the same magnification  of  3000x) 81 Fig. 5.2-3 XRD spectra of  alumina sol after  heat treatment at 200-500°C 82 Fig.5.2-4 Density of  gel cast CSG including a mixture of  two different  sizes of  alumina powders vs. concentration of  the fine  alumina particles. The size of  the fine  and coarse particles is -0.5 and ~3[xm, respectively 83 Fig.5.2-5 Fig.6.1-1 Comparison of  density, and specific  surface  area of  CSG as a function  of concentration of  fine  particles 84 SEM images of  the 1s t layer of  the A40 coating, (a) overall view, (b) the interface,  and (c) the coating volume, respectively. 87 Fig.6.1 -2 SEM images of  the A40 coating after  deposition of  the 3 r d layer, (a) overall view, (b) the interface,  and (c) the coating volume, respectively 88 Fig.6.1-3 SEM images of  the coating after  deposition of  the 5 t h layer of  the mixture of  A40. (a) overall view, (b) the interface,  and (c) the coating volume, respectively 89 Fig.6.2-1 DTA-TGA data for  (a) as-received CA; (b) alumina CSG with/without CA (TGA only) and (c) alumina CSG with/without CA (DTA only). Heat flow  for  alumina CSG with/without CA overlaps, resulting in a single curve 93 Fig.6.2-2 XRD spectra of  CA at elevated temperatures; (a) as-received and (b) after holding at elevated temperature up to 300°C with varying hold time at 300°C from  10 to 30 min 95 Fig.6.2-3 13C NMR spectra of  CA (a) as-received, (b) heat-treated at 300 °C and (c) the reference  from  the literature [2.51] 96 Fig.6.2-4 Comparison of  the fracture  morphologies of  as-received (a,b) and (c,d) CA-impregnated porous alumina specimens (~3fj.m,  -50 % porosity) followed  by heat treatment at 300°C 98 Fig.6.2-5 EDX analysis of  (a) the as-received and (b) CA-impregnated model porous alumina specimens 99 Fig.6.2-6 Comparison of  alumina coated aluminum (a) without CA and (b) with CA. -40 vol% porosity is observed in (a) while - 35 vol% porosity is observed in (b) due to filling  effect  of  CA. Porosity was measured by water absorption 102 Fig.6.2-7 EDX analysis, by (a) peaks and (b) quantification,  of  the Bond Coat layer between the CB-CSG coating and aluminum (the system includes additionally hydrogen, which is not detected by EDX) 103 Fig.6.2-8 Investigation of  wettibility of  CA melted at ~150°C on (a) a sintered alumina and 105 Fig.6.2-9 Profiles  of  strain recovery after  removal of  the constant force  of  18N, applied to AA5052 substrate, and the substrate with alumina coatings with/without citric ... acid. The force  was removed after  holding for  60 min. CA in the legend refers to citric acid, (a) overview, (b) magnified  between 70 and 120 min, and (c) log scale of  (b) 107 Fig.6.2-10 Relaxation moduli of  CA containing CB-CSG alumina coated AA5052, compared with as received AA5052 and conventional (CA-free)  CB-CSG alumina coatings on AA5052. Thickness of  the coatings was -100 |am. (a) linear axis scale and (b) log-log scale I l l Fig.6.3-1 Radius of  the curvature of  AA5052 aluminum substrate before  (i.e. at zero coating thickness) and after  deposition of  coatings of  increasing thickness 113 Fig.6.3-2 Average Young's modulus of  coated AA5052 aluminum, as determined by calculation and experiment 115 Fig.6.3-3 Young's modulus of  (a) as-received AA5052 and (b)-(f)  CB-CSG alumina coated AA5052 aluminum measured during heating and cooling cycles: (b) 1s t layer deposited (c) 2 n d layer added (d) 3 r d layer added (e) 4 t h layer added and (f) 5 layer added 117 Fig.6.3-4 XRD peaks of  alpha-alumina at (a) stress-free  state and (b)~(f)  under the stressed state in the order of  1st to 5 th layer 119 Fig.6.3-5 Comparison of  (a) strain and (b) angle shift  by XRD vs. coating thickness.... 121 Fig.6.3-6 Comparison of  residual stress obtained from  bending theory, using calculated Young's modulus and experimentally obtained Young's modulus by 3-point bending test for  alumina CB-CSG coatings on AA5052 123 Fig.6.3-7 Comparison of  the residual stresses determined by XRD and curvature methods 125 Fig.6.4-1 Comparison of  the effect  of  the shape of  the electrode (sharp end vs. flat  round end) on the dielectric breakdown voltage of  CB-CSG alumina coatings deposited on aluminum 127 Fig.6.4-2 Open porosity of  the consecutive alumina coating layers 129 Fig.6.4-3 Breakdown voltage of  the CB-CSG alumina coated aluminum as a function  of thickness of  the coating. The average thickness of  one single deposit is ~ 40 |xm and the coating is made of  5 layers 130 Fig.6.4-4 Dielectric strength of  the CB-CSG alumina coated aluminum with the thickness of  the coating 130 Fig.6.4-5 Relationships between residual stress (•) and dielectric strength ( • ) with respect to coating thickness 131 List of  Greek Symbols k  Thermal conductivity (W/m-K) cr Strength (MPa) a Coefficient  of  Thermal Expansion, CTE, (K"1) s Strain (no unit) v Poisson's ratio (no unit) crx Stress in direction of  x-axis (MPa) v Velocity of  a settling particle (mm/s) 77 Viscosity (Pa.s) p\ Density of  a solid (g/cc) pi Density of  fluid  (g/cc) T Relaxation time (sec) List of  Roman Symbols Ri Thermal shock resistance parameter (W/m) E Young's modulus (MPa) h0 Thickness of  substrate (jam) /o Stiffness  (N/m) Zo Point of  the neutral axis (|J.m) M Bending moment (Nra) E Average Young's modulus (MPa) E(z) Young's modulus at position z (GPa) d Lattice spacing (A) Da Apparent density (g/cc) wd Weight of  a sample in air (g) ws Water-soaked weight of  a specimen (g) W t Weight of  a sample immersed in water (g) Ra Average roughness (|J.m) \z I Distance measured from  the mean line in measuring roughness (mm) L Sample length in the x-direction (mm) K e Strain constant (no unit) D Displacement of  a drive shaft  (=0.05mm) t Sample thickness (mm) I Length of  span (= 40mm) F Force on a sphere (N) a Radius of  a particle (|J.m) Gravity acceleration (9.8 m/s ) Volume fraction Breakdown voltage List of  Abbreviations CTE Coefficient  of  Thermal Expansion AA5052 Aluminum Alloy 5052 CA Citric Acid CSG Composite Sol-Gel CB-CSG Chemically Bonded Composite Sol-Gel TDF Thick Dielectric Film NMR Nuclear Magnetic Resonance XRD X-ray Diffraction DS Dielectric Strength DT-TGA Differential  Thermal-Thermo Gravimetric Analyzer DMTA Dynamic Mechanical Thermal Analyzer DMA Dynamic Thermal Analyzer SEM Scanning Electron Microscopy EDX Energy Dispersive X-ray CAD Citric Acid-Derivative Acknowledgement At the outset, I would like to express my sincere gratitude and deep appreciation to my supervisor, Dr. Tom Troczynski for  the opportunity given to me to pursue the interesting research and for  his guidance and constant encouragement throughout this work. His guidance and support have been as helpful  as his technical expertise towards the successful completion of  this PhD program. I particularly acknowledge the financial  freedom  accorded by him at every stage of  the current program. I am also very grateful  to Dr. Olivera Kesler, Dr. Rizhi Wang and Dr. George Oprea for  constructive comments, guidance and support. As experimental work performed,  I also deeply thank Dr. Sing Yick (NRC, National Research Council Canada) for  his valuable assistance of  measuring curvatures as well as my thanks to graduate students, Donna Dykeman and Tomer Curiel in composite group in the department of  Materials Engineering at UBC for  their assistance with mechanical tests. In particular, I thank to Michael Chen and Manus Tsui who support the construction of  the thesis along with all my colleagues in UBCeram group. I also greatly thank to Dr. Andy Hundal Jung for  his invaluable constant encouragement during the program, NSERC for financial  support. I am very indebted to my beloved family  from  my first  day till this achievement. My special thanks go to my lovely wife,  Nayoung Lim, for  her continuous encouragement. CHAPTER  1 INTRODUCTION Ceramic coatings have a variety of  applications, such as electrical and thermal insulation [1.1-1.4], corrosion and wear protection [1.5,1.6], and many other functions,  such as catalysis or gas sensors [1.7]. The factors  which affect  electrical properties of  ceramic coatings for  electrical insulation include the band gap, but also engineering properties such as thickness and microstructure. When a ceramic coating (i.e., AI2O3, CTE=~8xlO"6/°C) on a metal (i.e., AA5052, CTE=~24xlO"6/°C) is considered, heat treatment temperature is a significant  factor  to be taken into account because of  typically large thermal expansion mismatch between ceramic coatings and metals. Upon cooling, a metallic substrate will typically contract significantly  more than the deposited ceramic layer, resulting in tensile stresses in the metal and compressive stresses in the ceramic layer [1.8-1.10]. This thermal mismatch limits processing and frequently  induces failure  of  the coatings. For example, Olding et. al [1.11] studied alumina-alumina composite sol-gel coatings on steel for  electrical insulation. They have found  it very difficult  to coat such a ceramic composite on steel with >100 jam thick coat without delamination due to micro cracking and thermal mismatch -triggered macro-cracking. However, according to their research [1.11] >100 (j,m thick ceramic coating has been required for  the coating breakdown voltage above 2kV. It is therefore  a processing challenge to overcome the above disadvantages, to be able to deposit thick ceramic coatings on metals without failure  of  coatings. This challenge is undertaken as one of  the principal objectives of  the present work. In order to address the effects  of  a large difference  in CTE between the ceramic coating and a metallic substrate, an alternative coating processing methods should be considered. For example, polymers can be applied to a ceramic-metal interface  as a bond coat to mitigate stresses at the interface  [1.12,1.13]. The viscoelastically deforming polymeric bond coat provides the means of  relaxing stresses in the ceramic-metal interface during cooling. Kozuka et.al  [1.12] reported that a sol-gel derived silica and titania coatings with an organic polymer (polyvinylpyrrolidone and polyvinylacetamide) bond coat, deposited on silicon by spin-coating, could be fabricated  crack-free.  Lower heat treatment temperatures also reduce the thermal expansion mismatch stresses [1.14], This is difficult  to achieve if  the coating is to be sintered. In this respect, sol-gel processing substantially allows to reduce the temperature and thus reduce the residual stresses. This aspect is also explored in the present work. In the current work we study (and attempt to solve) the processing limitations for thick ceramic coatings on metals through innovative use of  "bond coats", the intermediate layers between the metallic substrate and the ceramic "top coat". Citric acid (CA) has been used as a precursor for  the organo-ceramic bond coat. The Chemically Bonded Composite Sol-Gel (CB-CSG) and gradual deposition method has been used to lower the heat treatment temperatures for  coatings. All three approaches (i.e. (i) introduction of  the bond coat, (ii) decrease of  process temperature through use of  CB-CSG, and (iii) gradual (multilayer) deposition method, were aimed at decreasing residual stress in the ceramic coating such that relatively thick films  (>0.2mm)can be deposited without spallation on light metals such as aluminum alloys (AA5052). Although CA is a widely used dispersant for  alumina suspension [1.15-1.18], it transforms  to an amorphous, poorly defined  organic phase at elevated temperatures, and is believed to provide polymeric (viscoelastic) behavior [1.19]. The CB-CSG method requires firing  temperatures as low as -300 °C [1.20]. The CB-CSG method has been developed using chemical reaction between sol-derived nano-alumina (, i.e. <0.05jam particles derived from  low-temperature, i.e. 300°C, thermal decomposition of  aluminum hydroxide) and phosphoric acid. This chemical reaction (refer  to Section 2.1.4.1) provides bonding between alumina particles (0.5-3|j.m) and metal substrate when used in coating by forming  aluminum phosphates in the coating and in the interface  in the CB-CSG process [1.21]. The current process engineering study explores the combined use of  CB-CSG with a CA-derived bond coat as an innovative technique to deposit thick dielectric films  (TDF) on metallic substrates with large difference  in CTE between them (alumina: 8xlO~6/°C, AA5052: 24x10 /°C), to insulate heating elements (i.e., carbon based electrical resistive films)  from the base metal substrate. Phosphate-bonded alumina and aluminum alloy (AA5052) are used in the current work as the ceramic coating and the metal substrate, respectively. Phosphate-bonded alumina has a potential for  application as an electrical insulation film  resistant to voltages up to 2kV, providing the thickness of  the coatings is >200 |_im, Fig. 1.1 [1.22]. The resistive film  must be insulated from  the metal substrate to pass the standard safety  test (2kV for  3 sec). Alumina is a promising insulation material for  the heating elements, e.g. in common appliances such as kettles, due to its relatively high thermal conductivity (30 W/mK) and high dielectric strength (10-16 kV/mm) [1.23]. Aluminum alloy, AA5052, is also preferable  because of  its low density (2.68 g/cc) and excellent thermal conductivity (138 W/mK) [1.24]. It has been reported by Datec Coating Corp., supporting this research, that such 3kW heating element system can boil 1 liter of  water in 2 min. Tablel-1 compares the currently used metallic substrates for  heating elements. Even though CTE of  AA5052 (24X10"6/°C) is much larger than that of  stainless steel (SS, 10xl0"6/°C), the heating efficiency  (i.e. low losses due to high thermal conductivity) of  AA5052 presents a great advantage. Therefore,  it is worthwhile to develop a process in which the alumina coating is deposited on AA5052 to provide sufficient  electrical resistance films. Electrical Insulator Heating Element Alumina Coating AA5052 Substrate Fig. 1-1 High efficiency  heater and its application for  water boiling Table 1-1 Comparison of  a current stainless steel (430) and proposed aluminum alloy (5052) metal substrate for  heating elements [1.24], Property Stainless Steel (430) Aluminum alloy (5052) Density (g/cc) 7.7 2.68 Thermal Conductivity (W/m-K) 20 138 Coefficient  of  Thermal Expansion :CTE (xlO"6/°C) 10 23.75 Melting Point (°C) 1425-1510 600 Yield strength (MPa) 310 90-255 CHAPTER  2 LITERATURE  REVIEW 2.1 Ceramic Coating Processing The thickness of  ceramic coatings depends on the intended applications, e.g optical coatings are generally thin (sub-micron), while electrically insulating coatings usually have to be relatively thick (> 100 jam) to achieve sufficient  dielectric strength [2.1-2.2], In the current study, we investigated thick (>200 (am) alumina coatings proposed as electrically insulating films  for  heating elements. Processing of  the thick coatings for  the heating element was studied via the Chemically Bonded Composite Sol-Gel (CB-CSG) technique which can provide much thicker films  than a conventional sol-gel technique [1.5]. The sol-gel technique has been one of  the most promising processes since it does not require complicated equipment and high temperature heat treatment (< 500 °C) [2.3-2.8], 2.1.1 Sol-Gel Processing The sol-gel technique can use a water-based process or an alcohol-based process, a through solution of  metal salt or metal alkoxide, respectively [2.9]. The first  step in the water-based process is to form  a sol by hydrolysis of  the metal ions, M: M n + +nH20-* M(OH)n + nH+ (2.1-1) Gelation of  the sol is achieved by either dehydration or an increase in pH. During the gelation by the dehydration, the energy barrier to gelation is reduced by the increase in the concentration of  electrolyte in the diffuse  layer around the individual particles. Increase in pH reduces the magnitude of  positive surface  charge on the sol particles, leading to the reduction of  repulsive forces  between particles and lowering the height of  the energy barrier [2.10], leading to gelation. In the preparation of  alumina sols, various aluminum compounds can be used, such as aluminum iso-propoxide (Al(OC3H7)), aluminum butoxide (Al(OC4H9)3) and boehmite (AlOOH)) [2.11]. In particular, alumina sol from  boehmite can be formed  via hydrolysis and condensation of  Al3+ ions in water at pH>3 as follows  [2.12]; where h indicates the molar ratio of  hydrolysis. For h>2.46, rapid precipitation of  condensed, amorphous, or weakly crystalline hydrated alumina phases occurs. The most common crystalline phase in the reaction, Eq.2.1-3a is pseudoboehmite, y-Al(OH)3 [2.12], When y-Al(OH)3 is dispersed in water, a polymerized structure forms; Al(OH)3 +H-OH -> [Al(OH)3 (OH)]" +H+ <=> [Al(OH)3 -OH] - + H++ [Al(OH)3 -OH]" [(0H)3Al-0-Al(0H)3]2" +H 3 0 + (2.1-4) Fig 2.1-1 illustrates the surface  of  water-based alumina for  varying pH. For pH<9, H+ is adsorbed on the surface,  creating an effective  positive charge, [0-Al-0-H-H]+. At pH>9, H+ desorbs from  the surface  of  the aluminum oxide in water, creating an effective  negative charge on the surface  [2.13-2.16]. A1(0)-0H +H-OH-* Al(OH)3 [A1(H20)6]3+ +h H 2 0 [Al(0H)h(H20)6.d (3"h)+ + hH 30 . hH 30+ + h0K-^2hR 20 (2.1-3b) (2.1-3a) (2.1-2) Al / I ) o h 2 A l 7 + H 2 ° \ / Al \ OH. H / + / Al Al / \ H O m Al Al / \ HO Al Al / \ 0 pH-9 l O H + 0 H OH 0 H 0 + 2 0 \ / \ Al / Al / Al Fig.2.1-1 Reaction of  the hydrated surface  of  alumina with H3O+ or OH-. [2.13,2.14] 2.1.2 Composite Sol-Gel The sol-gel process technique provides many advantages over conventional ceramic processes, including high purity raw materials (i.e., >99.5 %) and low heat treatment temperatures (i.e.< 500 °C) [2.17-2.21], However, it has a significant  drawback, namely the large processing shrinkage (>90 %), resulting in cracks [2.17-2.22], A composite sol-gel technique has been recently proposed to avoid the disadvantage of  the large shrinkage of  a sol-gel that causes cracks in thick coatings on metallic substrates [2.23-2.25]. The composite sol-gel is fabricated  by dispersing ceramic powders, fibers,  or whiskers as a second phase into the sol. The patent by Barrow et al. [2.23] has disclosed ceramic coatings of  thicker than 100 (am on selected substrates using a composite sol-gel technology. The composite sol-gel coatings did not crack during drying because of  the high weight percentage of  the filler  in the slurry. Murrell [2.26] also reported that coating thickness with the sol alone is limited to less than 10 jam without cracks, while >25 (am can be coated by composite sol-gel. In the case of alumina, the 20 wt% alumina sol with the dispersed alumina powder is suggested for  a coating thickness of  >25 jim [2.26], Figure 2.1-2 presents the schematic of  the formation  of a 3D network of  alumina as the filler  and alumina sol upon drying. Fig.2.1-2 Schematic of  sol particles gluing coated particles into a coherent 3 dimensional structure [2.26]. a-Alumina is a favorable  filler  to make the composite sol-gel with alumina sol because the alumina precursor transforms  to the same chemical compositions as the a-alumina at elevated temperatures, i.e. AIO(OH) ^ y-Al203 (at 450 °C) 5-Al203 (at 600 °C) 9-Al203 (at 800-1000 °C) a-Al203 (at >1200 °C) [2.27,2.28], The alumina composite sol forms  the three dimensional network structure by connecting the adsorbed sols, resulting in a relatively low surface  area (~3 m2/g) material during the gelation [2.26], By having such a low surface  area, packing density of  particles is increased allowing increases the thickness of  a single coating deposition. However, no research for  SG or CSG thick (100-Dispersed alumina . ~0.5 |am Alumina sol -50 nm 200|j,m) alumina coating on aluminum has been reported. 2.1.3 Packing of  Particles In order to minimize the void volume, different  particle sizes should be mixed as shown in Fig.2.1-3, also known in ceramic processing science as "grading" [2.29], Fig.2.1-3 Packing of  spheres of  different  sizes; (a) theoretical packing and (b) practical packing [2.29] "Smaller particles introduced and distributed in the interstices of  packed larger particles will reduce the porosity and pore size. Large particles added to finer  particles displace fines  and pores and reduce the porosity [2.29]." Very high densities may be obtained using this principle, e.g. McGeary [2.30] experimentally achieved a packing density of  -95% for  a quaternary system of  vibrated steel spheres while obtained a packing density of  -70% for  a binary system. Figure 2.1-4 plots the porosity of  two component mixtures of  tabular aluminas [2.31], Fine Fraction (wt.%) Fig.2.1-4 Calculated and experimental total porosity for  vibrated, two-component mixtures of  tabular alumina fines  (-200+325 mesh) and three different  coarse fractions  [2.31], Blending of  different  particle sizes (i.e. grading) to achieve high packing density is common in formulating  coarse-grained refractories.  The commonly used batch for  high density-packing is approximately 60-65 wt.% coarse particles and 35-40 wt.% fine  particles. Another study of  packing density by Hugill and Rees [2.29] provides ternary diagram using silica particles of  three distinctive sizes, Fig.2.1-5. According to their diagram, the most suitable mixture ratio for  the highest density of  the silica bulk (-1.74 g/cc) is 45% coarse, 10% medium, and 45% fine  particles. These results were applied in the current study for  the fabrication  of  coatings with as high packing density as possible. ; COARSE ( 5 - 3 0 MESH) Fig.2.1-5 Apparent density graded and fired  mixtures of  silica rock. The solid circle indicates the highest density of  silica (-1.74 g/cc) [2.29]. 2.1.4 Chemically Bonded Composite Sol-Gel (CB-CSG) Ceramics made by sol-gel are usually sintered at temperatures >500°C to achieve appropriate mechanical strength and density. However, such a high temperature limits the use of  ceramic coatings on some metallic substrates whose melting temperatures are low, such as aluminum and magnesium. If  the properties of  a ceramic coating obtained by high temperature sintering can be obtained by low sintering temperature synthesis, it will serve as an impetus to the improvement of  ceramic coatings. This can be obtained by introducing chemical species to ceramic composite sols, inducing a chemical reaction between the chemical species and ceramics at as low temperature as 300°C. This process is known as a Chemically Bonded Composite Sol-Gel (CB-CSG). The chemical used in the current study for  alumina composite sols is phosphoric acid, commonly used for  processing of  mono-aluminum phosphate (A1(H2P04)3) and aluminum phosphate (AIPO4) [1.2,2.32-2.39]. The chemical bonding via phosphating on sol-gel derived oxides (alumina, AI2O3) or hydrated oxide (aluminum hydroxide, Al(OH)3) provides denser and harder ceramics than ordinary composite gel derived ceramics, after  heat treatment at temperatures as low as 300°C [2.36,2.37], Previous studies [2.32-2.39] indicate that the phosphates can reduce porosity by playing a role as an amorphous binder, and form  a chemical reaction with alumina, lowering the firing  temperature to ~300°C, resulting in strength comparable to that of  ceramics made at conventional (high) sintering temperatures >1000°C. Thus, strong and dense ceramics can be produced through the chemical reaction between alumina and phosphates. This reaction depends on the ratio of  A1:P and temperature. The product, aluminum meta phosphate (A1(P03)3), is favored  by high phosphate-loading conditions (A1/P<1) whereas aluminum orthophosphates (AIPO4) are favored  by low phosphate loading conditions (A1/P>1) [2.33]. The reaction of  alumina or hydrated alumina with phosphoric acid (H3PC>4) is exothermic and produces mono-aluminum phosphate (A1(H2P04)3 ) as an amorphous binder at low temperature (<500°C) [2.40]: A1203 +6H3PO4 -> 2A1(H2P04)3 +3H20 (2.1-1) Al(OH)3 + H 3 P0 4 -> amorphous phase (2.1 -2) Mono-aluminum phosphate produces AIPO4 and AlPCV H3PO4 when heated as well: A1(H2P04)3 ->A1P04- H3PO4 + H3PO4 (2.1-3) A1(H2P04)3 —> AIPO4 + 2 H3PO4 (2.1-4) In order to have a complete reaction, e.g. dehydration and the formation  of  stable aluminum phosphate, a heating temperature above 500°C is required [2.40]. Figure 2.1-6 illustrates schematically bonding between the phosphates, alumina and alumina sol. Reaction according to Eq.2.1-2 takes place on the surface  of  alumina particles and Eq.2.1-3 and Eq.2.1-4 in the solution. Phosphates Alumina particle-surface  reacted with phosphates Alumina particle-surface  reacted with phosphates Alumina sol- reacted with phosphates Fig.2.1-6 Schematic of  interactions between alumina particles or alumina sols and phosphates in a CB-CSG system [2.33]. 2.1.4.1 Transformation  of  Phosphates The chemical bonding reaction depends on the ratio of  A1:P , the reaction temperature, and the reactant's surface  area. AIPO4 has attractive properties, e.g. stability up to ~1800°C [2.34], AIPO4 is the final  product of  the reaction between alumina (anhydrous or hydrated alumina) and orthophosphoric acid (H3PO4). Bothe et al. [2.34-2.35] studied the reactions between phosphoric acid and different  aluminas (anhydrous alumina and hydrated alumina) for  varying temperatures and surface  areas. According to their research, AlPC^-HbO forms  at 113-133°C when alumina reacts with phosphoric acid, for  a molar ratio of  Al/P=l:l. Dehydration of  A1P04-H20 to get A1P04 is nearly complete at about 170°C. Hydrated alumina, y-boehmite and gibbsite, produce berlinite (equivalent to quartz structure) while anhydrous alumina produce cristobalite. Bothe et al. [2.35] studied the chemical bonding reactions for  anhydrous and hydrated alumina. According to their results; in the case of  anhydrous alumina, the heat released during the reaction increases for  a high initial surface  area and high concentration of phosphates. On the other hand, for  the hydrated alumina, although the boehmite is finer  and more crystalline than gibbsite (0.5 fim  particle size), the heat released in the boehmite during the reaction was lower than that for  the gibbsite (e.g., boehmite: -32 J/mol and gibbsite: -40 J/mol). This was explained by the higher solubility of  gibbsite than that of  boehmite [2.35], For both anhydrous and hydrated alumina, the heat released increases with high temperature and concentration of  phosphoric acid. Formation of  type A (tridymite), B (berlinite), and C (cristobalite) phosphates depends on the temperature and the ratio of  Al/P. Type A is favored by high concentration (A1:P=1:6-12) and high temperature. Type C and B are formed  with the ratio of  1:3 and 1:1 of  A1:P, respectively [2.33-2.35]: <140°C >200°C AI2O3+H3PO4 > AIPO4+I/2 AI2O3 +3/2H20 > A1P04(-B,-C)+ 1/2 AI2O3 (2.1-5) 450-600°C 650-800°C, -3/2 H 2 0 10 Al203(<l|am)+H3P04 -> amorphous phase A1P04(-B,-C) + 19/2 A1203 (2.1-6) 500°C, 3 H 2 0 800°C Al(OH)3 + H3PO4 > 1/2 A1P04-B + 1/6 A1(P03)3-B +1/6 A1203 > I/2AIPO4-A+ l/6Al(P03)3-A+l/6 AI2O3 (2.1-7) 200°C, -2 H 2 0 500°C, -H 20 Al(OH)3 + H3PO4 > 1/3A1(H2P04)3 +1/3 A1203 > 1/3 A1(P03)3-B +1/3 A1203 800°C 1200°C > 1/3 Al(P03)3-A+l/3 AI2O3 >• 1/3 AIPO4-C +1/3 A1203+ 1/3P205 (2.1-8) 2.2 Citric Acid in Ceramic Processing Ceramic powders dispersed in liquids are normally unstable due to Van der Waals attraction forces,  resulting in large (>10 um) agglomerates, and fast  sedimentation of  the clusters. Although a certain amount of  agglomeration may be advantageous in deposition of thick ceramic coatings [2.41], excessive agglomeration may cause undesirable viscosity change of  the slurry. Controlling viscosity of  alumina slurry for  spray coating in this study is a significant  processing factor.  The viscosity of  the slurry can be controlled by adding a stabilizing dispersant into the slurry, such as citric acid. Organic dispersants are widely used to control the surface  charge for  stability of  suspensions in aqueous systems [2.42-46, 2.48-50], The quality of  the colloid is usually determined by its state of  dispersion. 'In aqueous colloidal processing of  alumina powders, control of  the suspension properties is usually achieved by adjustment of  the repulsive double layer forces  either by (i) protonation or deprotonation of  the surface  hydroxyl (-OH) group, thereby creating a surface  charge; (ii) addition of  specific  adsorbing organic or inorganic oligo- or polyelectrolytes [2.44].' The key factor  to achieve optimized coating properties with an alumina composite sol is the ability to stabilize the alumina slurry and to control its viscosity during spraying. This can be achieved by adding citric acid (CA) as a dispersant to the slurry [2.41], 2.2.1 Citric Acid The pX"a (i.e. acid dissociation constant; the lower the pK a the more acidic the solution) of  citric acid CA (C6H807), Fig.2.2-la, at the ionic strength (7=0.1) is between 3 and 6 [2.44], CA adsorbs on alumina in an acidic environment (pH<7) and desorbs in an alkaline region [2.45], CA may have somewhat different  structure depending on the pH of the solution, e.g. the form  of  citric acid in the neutral and base solution, and tricarballylic acid in acid solution, Fig.2.2-1 [2.44], At room temperature, CA is a white crystalline powder. It can exist either in an anhydrous form  (C6H807), or as a monohydrate (C6H807 'H20) that contains one water molecule for  every acid molecule. The anhydrous form  crystallizes from  hot water, while the monohydrate forms  when CA crystallizes from  cold water. The monohydrate dissolves readily in water. It can also be converted to the anhydrous form  by heating it above 74°C [2.50] and the melting point is ~150°C and decompose thereafter.  [2.47]. H I O O H O O ^ / o C o r II  I  II ,C C C H c C c / \ / \ / | \ / \ ^ H \ ° O C H C x O \ H H I \ \ l H H H H H (a) (b) Fig.2.2-1 Structural formula  of  (a) citric acid and (b) tricarballylic acid [2.44], 0 1 1 c II c / \ c / 1 \ \ / \ 0 C H | \ 1 2.2.2 Role of  Citric Acid in Aqueous Alumina Suspensions CA is widely used as a dispersant for  aqueous alumina suspensions [2.43-2.45,2.48]; adsorption of  CA on particle surface  causes an increase of  double layer repulsion [2.44], The hydroxyl group at the interface  between alumina and water can be exchanged for  a carboxylate group which might change the surface  charge [2.52,2.53], According to the ligand exchange model, the adsorption of  the CA onto the surface  of  alumina particles can be drawn as in Fig.2.2-2. Citric acid has three functional  groups of  carboxylic acid sites. These carboxylic acid sites can be dissociated (i.e., COO- at pH>3) or not be dissociated (i.e., COOH at pH<3) depending on the solvent conditions of  pH [2.49]. Surface of alumir sol I -OH" C H 2 - C O O H • O H - + n O H - C - C O O H I c h 2 - c o o h OH-nH20 + C H 2 - C O C i t ; O H - C - C O O H C H g - C O O H Cit-C-0 -n 0 Cit- C - 0 11 0 Cit- C - 0 11 0 Surface i l ° f ;^|lumina i sol Fig.2.2-2 Schematic of  hydroxyl group on the surface  of  alumina sol exchange for  carboxyl group in a citric acid containing aqueous suspension system. The adsorption of  the ions in the oxide-solution by the ligand exchange does not influence  the surface  charge because it is a simple exchange of  -OH" group for  -COO". A change of  the surface  charge occurs only when an additional carboxylate group that is not coordinated to the surface  is present in the molecule. Two carboxylate groups coordinate to the same Lewis base center, or -OH 2 + group is exchanged instead of  a hydroxyl group [2.44], It is believed that the surface  charge of  the dispersed alumina particles does not change because a pH of  the alumina slurry is -4.7. [2.43,2.44], Usually, the amount of  the CA in an aqueous system is small (< -0.2 wt%) since it influences  the viscosity of  a slurry. When the amount of  citric acid in a suspension is increased, the isoelectric point (IEP) of  the suspension is changed (i.e., IEP is at pH=7 with 0.1 wt% citric acid and at pH=3.4 with 0.4 wt% citric acid) [2.44], A larger concentration of  the CA than desirable (i.e., >0.15 wt% at 3<pH<7) in the suspension causes the reduction and, finally,  disappearance of  the repulsive double-layer and then agglomeration of  the dispersed particles [2.44], Therefore,  it is important to control the concentration of  the CA in a suspension for  an appropriate viscosity of  the suspension. 2.2.3 NMR for  Testing Materials Nuclear magnetic resonance (NMR) spectroscopy has long been a primary characterization technique for  liquid samples. NMR involves the interaction of  the nuclear magnetic dipole with magnetic fields.  These interactions provide detailed information  on the atomic environment. The frequency  or the magnetic field  is swept to obtain a resonance. The information  in such a resonant spectrum includes line position, often  related to the chemical shift.  This information  can then be interpreted to give insight into the local atomic environment of  those atoms responsible for  the resonance [2.54]. The table of 13C NMR absorptions of  major functional  groups is attached in Table A-3 in the Appendix [2.55], The chemical shift  (expressed in ppm) and spin-spin couplings of  such nuclei as 'H 13 and , J C are used routinely to analyze the chemical composition of  organic compounds [2.54], A material is placed in a magnetic field  (i.e., 1 to 20 T). The magnetic dipole of  certain nuclei, such as hydrogen or carbon, will tend to line up with the magnetic field.  These nuclei will precess, or rotate, at a particular frequency  that is dependent on the nuclide and the magnetic field  strength. Adding energy in the form  of  radiowaves at this specific  resonance frequency  will cause the nucleus to be pushed out of  alignment with the magnetic field.  The nucleus, after  some short period of  time, will then fall  back into alignment with the magnetic field  and emit a radiowave in the process. The emitted radiowaves are monitored, and plotted as a function  of  field  frequency  (ppm). [2.54]. 2.3 Dielectric Properties of  Ceramics 2.3.1 Ceramic Insulators Ceramic materials contribute to the crucial functions  of  a variety of  electronics including packaging devices, sensors, memory storage, and communication devices. The major use of  electronic ceramics includes capacitor dielectrics, thick film  components, and ceramic substrates [1.1,2.56-2.58]. Most ceramic materials are classified  as electrical insulators, i.e. with the energy band gap >7 eV. A good dielectric insulator should have resistance to breakdown under high voltages and it should not draw appreciable power from the circuit. It also must have reasonable physical stability, and none of  its characteristics should vary much over a wide temperature range [2.58]. The advantages of  using ceramics as electrical insulators are their capability for  high-temperature operation without significant degradation in chemical, mechanical, or dielectric properties. Ceramic materials have a wide range of  thermal conductivities (1-300 W/m-K) [1.1], depending both on the composition, microstructure and structural defects  in the material. The highest thermal conduction has dense and single phase oxide such as BeO, or non oxide such as A1N and SiC. The presence of  complex oxides such as Mg0-Al203 and Mg0-Si02 and microstructural features  such as pores, cracks, and second phases, generally lowers the thermal conductivity, Table 2.3-1 [1.1]. Table 2.3-1 Thermal conductivity of  single and complex oxides [1.1]. Materials BeO MgO A1203 Mg0-Al203 Mg0-Si02 A1N SiC Conductivity 210 36 30 15 5.3 250 100 (W/m-K at 100°C) For an electrical ceramic insulator such as alumina, degradation of  the properties upon heating is controlled by thermal conductivity (30 W/m-K at 100°C ), tensile strength (300 MPa at 25°C), the coefficient  of  thermal expansion (8.3 xlO'6/ °C at 25-1000°C), elastic modulus (400 GPa) [2.59,2.60], These properties control the thermal shock resistance, which is an important parameter when choosing a ceramic insulator for  applications in which temperature change may be severe, such as heaters, resistor cores and thick-film  substrates. The thermal shock resistance parameter, R2 (applicable for  low thermal conductivity materials, for  example ceramics), may be defined  as; R2 = (2.3-1) aE where A: is a thermal conductivity (W/m-K), a strength (MPa), a coefficient  of  thermal expansion (K"1), and E Young's modulus (GPa) [2.60,2.61], Alumina (96-99 wt.%) is currently one of  the most widely used ceramic materials for electrical insulators. It has demonstrated superior properties such as high dielectric strength (10-16 kV/mm), high hardness (Vicker's hardness 1600), high mechanical strength (300 MPa), and relatively high thermal shock resistance, as shown in Table 2.3-2 [1.3, 2.59,2.60], Table 2.3-2 Thermal shock resistance (W/m) for  ceramic materials [2.60] Material at 100 °C at 400 °C at 1000 °C AI2O3 2.7 1.1 0.6 MgO 1.98 0.9 0.5 Mullite 1.1 0.8 0.8 Zr02 0.3 0.3 0.3 Porcelain 0.5 0.1 0.5 2.3.2 Dielectric Strength "Dielectric strength is a measure of  the maximum voltage gradient that can be impressed across the dielectric without physical degradation of  its insulating properties, leading to breakdown [1.1]." The dielectric strength of  electrically insulating materials can be influenced  by their intrinsic properties (i.e., the lower dielectric constant, preferably  « 30, the better dielectric), geometry (i.e., the thicker the better) and flaws  (i.e., the lower amount of  the cracks, pores, and second phases the better), but also by the measurement conditions (i.e., less humidity the better) [1.1]. Buchanan [1.1] classified  a material whose dielectric strength is grater than 5 kV/mm at room temperature as good insulator. In that sense, alumina is one of  the best candidates because it has low dielectric constant (~ 8.2-10.2) and high dielectric strength (-10-16 kV/mm) [1.1]. Aluminum phosphate (AIPO4, berlinite) is also considered as a good dielectric material due to its low dielectric constant value of  -4.5-6 at between lOkHz-lMHz [2.62,61] so it presumably improves the dielectric properties of alumina coatings when contained. According to those results [2.61,2.62], H2O contents in AIPO4 also increase the dielectric constant because hydroxyl group (OH") is randomly occupied in sites and associated with P, forming  P-OH. However, in the current study, it is believed that the amount of  aluminum phosphate formed  by reaction between phosphoric acid and alumina sources (both sols and fillers)  during heat treatment is too small (-2 wt %) affects  the alumina coatings. For gases, the breakdown happens due to collision of  electrons with gaseous atoms leading to further  ionization. For liquids, breakdown is caused by dielectric heating and ionization processes by mobile ions in the liquid. For solids, breakdown can be explained by several possible mechanisms: intrinsic, thermal, ionization, and electromechanical breakdown [1.1]. In addition to the environmental factors,  microstructural defects  (i.e., impurities and pores) and a local increase of  temperature (i.e. local hot spots), influence dielectric strength. Increase in the level of  impurities, temperature and porosity decreases dielectric strength [2.64]. Dielectric strength of  alumina has been intensively studied as a function  of  various different  additives (i.e., polymers, Si02, and Mg) in the alumina matrix [2.65-2.68] or as an additive in other oxide materials (i.e., Si02) [2.64], and for  the various forms  for  pure alumina [2.69-2.71]. Olding et al. [2.67] investigated dielectric properties of  alumina and alumina/silica sol-gel composite coatings. They found  it difficult  to deposit coatings thicker than 100|_im without cracks when using small particle sizes (-0.3 jam) alone and in a mixture with large particles (-3.6 |j,m). However, they suggested that a using mixture of  large ( - 3.6 (am) and small (~ 0.3 |am) alumina particles allowed deposition of  thick films  (>100 (am), with high breakdown voltages over 3000VAc- Garcia-Cerda et al. [2.64] also investigated the dielectric strength with respect to the effects  of  alumina doped (1 wt.%) into sol-gel derived silica films  (0.25-0.4 (am) on a silicon substrate at ~ up to 200°C. They found  that alumina-doped silica films  have higher dielectric breakdown strength (-250 V/mm) than that of  pure silica film  (~ 6 V/mm). The dielectric breakdown in solids can be classified  as intrinsic, ionization, thermal, and electrochemical [1.1,2.72], Intrinsic breakdown (by electrons released from  an electrode, knocking down other electrons and leading to avalanche breakdown) occurs rarely because the high electric field  (~106-107 V/cm) is required. Ionization breakdown occurs in inhomogeneous dielectrics through the partial discharges resulting in the pores or cracks in the ceramic. The ionization of  gases inside of  the pores occurs when the ceramic is subject to a high enough electric field,  resulting in the local generation of  heat in the ceramic. This local heat causes a temperature gradient and eventually an increase in the conduction of electrons followed  by locally generated thermal stresses. Thermal breakdown is generally caused by the local heating when the heat conduction rate from  the electrode is faster  than heat dissipation out of  the ceramics. Thermal breakdown depends on the thermal conductivity and geometry of  an electrode [1.1,2.72]. 2.4 Residual Stresses in Ceramic Coatings Thermal expansion mismatch of  two joined materials leads to thermal stresses if  the system temperature changes. A body uniformly  heated to increase its temperature by AT expands, the strain described by a second-order tensor (%). The connecting property is called the linear coefficient  of  thermal expansion, ay, i.e. Sy = ay AT  (2.4.1) A polycrystalline ceramic consists of  a random array of  single crystals. If  these crystals are non-cubic and the temperature is changed, each crystal will attempt to strain differently  from its neighbors. Therefore,  adjacent crystals will push or pull on each other, creating local residual strains in the material. These strains are termed residual because they exist even in the absence of  the applied stresses. In some cases, this effect  can give rise to local micro-cracking. Indeed, this may happen as the material is first  cooled after  high temperature fabrication.  As a reference,  fracture  strain for  polycrystalline ceramic is typically less than 0.001 (0.1%) [2.73-2.76], The CB-CSG coatings studied in the present work are made of  several, e.g. five, layers of  alumina. Thermal expansion of  each alumina layer, which is newly deposited on a previous layer, may differ  from  the previous coating layer due to the presence of  cracks and pores introduced during the heat treatment. The development of  stresses in such multi-layer coatings may be described by the "rod" model in Fig.2.4-1 [2.76], The rods have different coefficient  of  thermal expansion, a\,a2, and (oc\>a-i>cc-i)  and different  Young's moduli, E\, Ej, and £3. Assume that the structure is initially in equilibrium and is then cooled. If  the overall strain induced in the structure by the cooling is e, it is clear that the rods will be subject to residual stresses, as rod 1 would prefer  to shrink more than rods 2 and 3. Thus, rod 1 will end up in uniaxial tension and rod 3 in uniaxial compression. The net value of  e will be determined by a force  balance. Assuming the materials are linear-elastic, the stresses in the rods are E\ (a\AT-s),  E2 (a 2AT-s),  and £3 (a3AT-e),  and by using a force  balance, the average coefficient  of  thermal expansion, e=aAT,  where [2.76]: E ] ex 1 4" JE cc 1 "t~ Ei gc  1 a = _J_J 1—1 (2.4.2) E,+E2+E3 Wi',  }wms> m ut a. Rod I i/ -rt'iwf; Mr  A W i+i f lIi M A fe.iJ a 2 Rod 2 •MM lif •Mm (a) TT a3 Rod 3 ' m ttfj f ^ i tfh  Jffl i> £ J rV f*£  j-j SIS f § & P-J kl  .ti (b) H o E? J £3± Fig.2.4-1 Model for  the expansion of  (a) three rods with different  coefficients  of  thermal expansion [2.74] and (b) alumina CB-CSG multi-coating layers with viscoelastic organic bond coat, as in. the present study. If  the Young's moduli values of  the rods (or coating layers) are the same, the average thermal expansion becomes a= (ai+a2+a3)/3. Microcracks lower the thermal expansion in non-cubic polycrystalline materials where there is the significant  thermal expansion anisotropy. For example, if  the tensile stresses in rod 1 become so high that failure  occurs, the stress in this rod is released. The equilibrium is now defined  by rods 2 and 3, so a= {a 2+ai)H.  Let's consider ai=4,a2=3, and a3=2 as assumed in the rod model. In the initial equilibrium average thermal expansion becomes 3. If  rod 1 fails  due to the stresses, new thermal expansion becomes 2.5. Thus, the overall coefficient  of  thermal expansion is reduced [2.76], Accordingly, microcracking in the multilayered ceramic coatings can reduce their thermal > expansion (contraction) and stiffness,  and relax the residual stresses. This rod model can be further  adapted to the current study by including a CA-derived viscoelastic bond coat, Fig.2.4-lb, represented by a spring-dashpot model [2.78]. If  the strain is maintained constant the stress changes with time [2.78] : E(t)  = ^ (2.4-3) where E(t)  is the relaxation modulus, a(l)  is stress, and so the constant strain. Newton's law, Eq.(2.4-4a), is the equation of  motion for  a model with a simple linear viscous behavior and integration, Eq.(2.4-4a) for  constant stress ao, yields Eq.(2.4-4b). de  „ „ N cr = 7]— (2.4-4a) dt s(t)  = ^ t  (2.4-4b) where 77 is the material viscosity. By using Maxwell model [2.79], viscoelastic behavior of  a material can be expressed through the combination of  Hooke's law (first  term on the right side) and Newton's law (second term on the right side): ^ = (2.4-5) dt  E dt  t] where 7= tE,  where x is the proportionality constant between E and 77, which is known as relaxation time of  element. For constant strain application, ds/dt  =0 and Eq. (2.4-5) can be transformed  as follows: d<J  ,, dt  ^ „ „ N — = d  In 0- = (2.4-6a) a r The expression is now integrated from  ctq at time 0 to <j(t)  at time t to give the following: In <r(f)  = In cr0 - - (2.4-6b) T Exponentiation and division by sq produces (2.4-6c) Where o(t)/so  = Eft)  and cr(>4b=£, so Eq.(2.4-6c) becomes E(t)  = Ee~"T  (2.4-6d) •From Eq.(2.4-6d), it is clear that stress relaxation, Eft),  is controlled by viscosity of  a material (through rj=rE). 2.4.1 Mitigation of  Residual Stresses in Coatings The strategies to reduce the residual stress in coatings include graded compositions and porosity [2.80-2.82] and inclusion of  bond coat at the interface  [2.83-2.85] between the ceramic layer and metal substrate. The above methods are widely used in the thermal barrier coatings (TBC) because the magnitude of  stresses in the coatings depends on the process temperature, on the coefficient  of  thermal expansion, and on the Young's modulus of  the deposited materials and metal substrates [2.86]. Portinha et al [2.81] reported an improvement in the thermal shock resistance of  Zr02-Y203 coating by grading the porosity (from  -10% in the first  layer to -17% in the top layer) and by using a bond coat (NiCoCrAlY); the gradient coating layers decrease thermal conductivity during processing and applications [2.80,2.81], Ilavsky et al [2.86] also investigated the effect  of  yttria-stabilized zirconia (YSZ) dispersed in metal (stainless-steel and NiCrAl) as a cermet bond coat, on thermal expansion, by two different  spraying methods such as atmospheric plasma spraying (APS) and high velocity oxygen fuel  (HVOF) spraying. They found  that the thermal expansion via both methods was decreased with the increasing YSZ in the metal (i.e., CTE of  16.4X10"6/°C for  100% NiCrAl and 12.9xlO-6/°C for  50% NiCrAl) as shown in Fig.2.4-2. Kozuka et al [2.85] reported use of  organic polymers as a bond coat, in the system of  sol-gel-derived titania coating on silica glass, with polyvinylpyrrolidone (PVP) bond coat. The critical thickness for  the coating was 0.2 (im without PVP, and increased up to 2 nm when PVP was added in the sol at a mole ratio of  PVP to titania of  0.5. 0 . 0 2 0 O J 0.015 J T3 S .2 o.oio « GD Thermal expansion of  stainless steel 1.APS 100% 2.APS 69.4% 3. APS 51% 4.HVOF 100% 5.HVOF NA (75%) 6.HVOF NA (50%) 5 0.005 0.000 200 400 600 Tempera ture [C] (a) 800 1000 0.020,-O fcJ  0.015 -J -a s •2 o.oio 03 OD S O g 0.005 Thermal expansion of  NiCrAl 1. APS 100% 2.APS 75% 3.APS 50% 4. HVOF 100% 5.HVOF 92.4% 6.HVOF 85% 0.000 200 400 600 800 Temperature [C] (b) 1000 Fig.2.4-2 Thermal expansion of  (a) stainless-steel-based and (b) NiCrAl-based materials [2.86] 2.4.2 Determination of  Residual Stress in Coating Systems There are the several methods to evaluate residual stresses in coatings, such as hole drilling, layer removal, and curvature measurement. Each technique has its own limitations depending on shape and coating material properties. In the current study, the alumina coated aluminum is subject to the residual stresses caused by thermal mismatch and it is therefore  of interest to compare the residual stresses measured by curvature and X-ray diffraction  (XRD). The effect  of  residual stresses on dielectric strength of  coatings was also investigated. 2.4.2.1 Determination of  Residual Stress by Curvature Method Residual stress determination by curvature is a common method and it is a non-destructive technique. This technique determines the average through-thickness stresses. Based on the knowledge of  the average Young's modulus, curvatures, stiffness,  bending moment, and geometry of  the coated metal, residual stresses can be calculated [2.87]. The curvature origin may be illustrated through the process illustrated in Fig.2.4-3. As a coating system, including metal substrate coated by ceramic, experiences differential  shrinkage (on cooling) or expansion in heating, the system bends (as well known for  the bi-metal switches) and residual stresses develop in the ceramic and metal, as represented schematically in Fig.2.4-3 [2.80,2.88], The bending curvature may be used to determine the level of  residual stresses. Consider a homogeneous metal substrate in the shape of  a beam with thickness ho, stiffness  /0, Young's modulus E0, and the point of  the neutral axis z0 where the strain is zero, Fig.2.4-4. For the simple case, z0= ho!2 and IQ=Eohol\2, A state of  equibiaxial stress exist in Ceramic Coating Layer (a) Metal a Substrate z / y V Cooling (b) Ceramic Coating Layer Metal Substrate T 2<TJ (c) Stress balance V i l Z cr — • < C7 — Neutral axis R Fig.2.4-3 Behavior of  a ceramic coated metal substrate in thermal cycling, (a) stress-free initial stage at elevated temperature,^, (b) ceramic coated substrate experiences differential shrinkage after  cooling to temperature, T2<T\. (c) Bending is allowed to balance the stresses in ceramic coating and a metal substrate [2.88] the plate when the length and width of  the beam are significantly  higher (i.e., >xl00) than its thickness which is considered in the current study. Thus, biaxial beam stiffness  can then be written as I  h3F J A; — -(1-v) 12(1-v) where v is Poisson's ratio. (2.4-9) Coatings i k 1 h\ f Substrate > Ah h0 x Fig.2.4-4 Profile  of  a coating structure [2.87] When a coating layer of  thickness Ah is deposited on the substrate of  thickness h0, new average Young's modulus E\ of  the coated substrate can be determined. The total thickness of  the beam becomes h\=hQ+Ah as shown in Fig.2.4-4. The stiffness  of  the new coated beam, I\ can be obtained experimentally by four-point  bending test using following  equation [2.87]; 7 , = ^ (2.4-10) k where M  is bending moment and k  curvature. From the difference  between the bending stiffness  A7=7i-7o, average Young's modulus E of  the added layer of  thickness Ah can be derived in the thin film,  (Ah«ho)  as follows  [2.87]: E = A/ Ah Ah ho~zo+ — V 2 (2.4-11) Once the Young's modulus of  the added layer is evaluated, the new average Young's modulus E and new position of  neutral axis zni of  the beam of  total thickness h\ can be obtained as follows  [2.87]: - AhE + h0E0 1 = 1 h (2.4-12) h0E0z0 + AhE hxEx K  + Ah (2.4-13) When the second layer of  thickness Ah is deposited on the first  coated layer, calculation of the new average Young's modulus E2 and neutral axis zN2 with thickness h2 is followed  as described above. In order to use the above procedure for  stress determination, the curvature of  the coated beam has to be evaluated before  and after  processing at a given temperature (i.e. 300°C). The variation of  curvature, Ak,  is enough to provide an internal stress distribution in the beam due to strain mismatch between the substrate and the coating. Knowing I\, ho, v, zni, E, and E(z),  the average stress in the added layer at the given temperature can be obtained as follows  [2.87]: A*(z)  = E(z)(z-z m)Ak-Ah*%)a< hE, 0 (2.4-14a) AAV' (2.4-14b) where E{z)  is the Young's modulus at position z. 2.4.2.2 Determination of  Residual Stress by X-ray Diffraction Determination of  residual stresses by XRD technique is a non-destructive method and may be used complementary to other methods (i.e., curvature measurement) [2.89-2.97], When materials are deformed  elastically, lattice spacing changes corresponding to the magnitude of  the stress applied, which causes the shift  of  the Bragg diffraction  peaks. By knowing the shift  of  diffraction  peaks and Young's modulus of  the material, stresses can be determined by X-ray diffraction  [2.89-2.96], There are two different  XRD techniques for  evaluation of  residual stresses in coatings. One is the ' 6 - 2 6 ' method and the other is the "Sin 2if/  'method. For better quantitative measurement, the Sin  ^/method is commonly used due to the uncertainty of  lattice spacing of unstressed material, d 0 [2.97], However, in the current study, we used the Ld-26'  due to the possibility of  observation of  strain change with coating deposition, wherein d 0 is observed by measuring ground gel-cast alumina reference  material. Consider a ceramic coated on a metal, with a difference  in coefficient  of  thermal expansion (CTE) between a ceramic (occ) and metal (o.m), where ac < <xm- When this coated metal is heat treated, biaxial or triaxial stress systems will develop [2.95], The structure of the ceramic coating system, as used in the current work, is subjected to a biaxial stress (i.e., t c « tM,  where tc and tM  is the thickness of  a coating and metal substrate, respectively) [2.93], The stress, <jz, in the normal direction (z  axis) at the free  surface  is always zero. Thus, biaxial stresses were considered in the XRD stress measurement. Consider a portion of  the surface  of  a stressed material, Fig.2.4-5. Stresses, a x and cry, are parallel to the surface  and <rz is zero. However, strain sz normal to the surface  has a finite value, as follows  [2.95]; where sx, sy, and sz is a strain in the direction of  x, y and z, respectively, and v is Poisson's ratio and E is a Young's modulus of  stressed body. The value of  SZ can be found  by measuring the lattice spacing d  under stress and do  at the stress-free  state of  planes parallel to the surface  and given by [2.95]: From the Eqs.(2.4-15) and (2.4-16) and considering that ax=Oy=crc, the value of  the residual stress parallel to the free  surface  can be obtained as follows  [2.95,2.97]: +ey) = ~(<r x +ay) E (2.4-15) (2.4-17a) cr = — E d  — d 0 2v d n (2.4-17b) Fig.2.4-5 Directions of  stress and strain (cx and a y parallel to the surface  and ctz normal to the surface)  at the surface  of  a stressed material. The magnitude of  the stresses determined by the XRD method may be different  from that measured by the curvature method because penetration depth of  X-ray is very shallow (i.e., < ~5 (im) and hence the stress measured by XRD is the surface  stress. However, the trend of  the stresses determined by XRD methods agrees well with the curvature method. Kesler et al [2.93] compared residual stresses of  Mo coatings on steel by XRD and curvature measurements, Fig.2.4-6. The stress values measured by XRD showed lower values than those measured by curvature measurement due to the nature of  the two measurement techniques. The values from  XRD can be further  reduced due to surface  roughness effects causing lower penetration of  X-rays, while the curvature measures the average stress in an added layer [2.93]. « £L it> tn £ w 160 120 eo 40 0.0 350 ~ 300 (o Ol 250 L^- 200 $ 150 100 m r a 0 2. 0,4 0.6 thickness (mm) (a) 0.S 1,0 0*2 0.4 T h i c k n e s s ( m m ) (b) 0.8 Fig.2.4-6 Comparison of  stresses of  Mo coatings on steel measured by (a) X-ray diffraction and (b) curvature measurement [2.93] CHAPTER  3 SCOPE  AND  OBJECTIVES 3.1 Scope of  the Investigation This study focuses  on innovative approaches to deposit > 200 fim  thick phosphate-bonded alumina coatings on aluminum substrates, for  the application as electrically insulating thick films.  The key feature  of  the work is development of  novel ceramic processing methodology for  thick ceramic films  on metallic substrates with substantially larger coefficient  of  thermal expansion. The following  four  processing approaches are undertaken to minimize the effects  of  the relatively large thermal expansion / contraction mismatch between alumina and aluminum alloy (AA5052) for  the process of  thick dielectric films  while avoiding coating failure:  (i) the use of  the composite sol-gel (CSG) for  the thick dielectric ceramic coatings (ii) the relatively low processing temperature of  300°C by using the novel chemically bonded composite sol-gel (CB-CSG) with chemical reaction between phosphoric acid and alumina; (ii) reduction of  thermal stresses through the use of  novel citric acid - derived organic (viscoelastic) bond coats, and (iii) multilayer (graded) coating deposition. The deposited films  are characterized for  the micro structural features,  residual stress (through bending and XRD methods) and dielectric strength. It was discovered that segregation of  CA/alumina sol nano-composite during the coating heat treatment process to the interface  between the coating and the substrate allows forming  in-situ nano-composite organo-ceramic bond coat (after  partial pyrolysis of  CA during the heat treatment). The effects  of  precursor preparation (alumina, sol, and CA) and coating thickness buildup on residual stress development and dielectric properties of  the coating were investigated in terms of  the coating thickness. 3.2 Objectives The principal objective is to study and develop novel ceramic coating processing techniques to decrease residual stresses in thick ceramic films  on metals. In particular, materials with relatively large difference  in coefficient  of  thermal expansion (CTE) between the two phases are of  interest, such as alumina and AA5052. In order to reach the above broad goal, the following  processing objectives and techniques were explored, with the specific  tasks defined  for  each objective: • Objective:  to reduce  shrinkage  of  ceramic films  on metallic  substrates;  to address this  objective,  the Composite  Sol-Gel  (CSG)  technique was used  for  coating  slurry preparation.  The objective was to overcome the basic disadvantage of  the conventional sol-gel technology, such as high shrinkage (up to -95 vol.%) causing micro-cracking and thick coating failure.  CSG can reduce the shrinkage to -75 vol.% • Objective:  to reduce  process temperature;  to achieve this,  CB-CSG  technique for ceramic bond formation  was developed.  The chemical reaction between phosphoric acid and alumina sol provides ceramic bond in the coating at process temperatures of 300°C, and the chemical reaction between phosphoric acid and metallic substrate provides adhesion of  the - coating to the substrate; the relatively low processing temperature of  300°C allows for  lowering thermal strain and reduction of  residual stresses. • Objective:  to reduce  stress buildup  during  deposition  of  thick  films:  addressed  by multi-layer  coating  deposition.  The objective to achieve a thick, low density of cracked coating, and to prevent delamination of  the ceramic coatings, requires that the deposition process is step-wise; in this work it was necessary to deposit multiple layers, -40 |_im thick each, including the heat treatment cycle after  each layer deposition, to obtain an overall coating thickness of  >200 jam. The system characteristics, in particular residual stress, are monitored after  depositing each consecutive layer of  coating. • Objective:  to relax  thermal  stresses in coating  through  provision for  viscoelastic flow  at interface;  to achieve this  objective,  organo-ceramic  bond coats were introduced.  In order to decrease the effect  of  the difference  in CTE between the alumina and AA5052, causing unacceptably large residual stresses in the >200 jam thick coatings, citric acid / alumina sol composite is utilized as a bond coat for  the mitigation of  the stresses by viscoelastic relaxation at the interface. The process development objectives listed above are augmented by characterization of  the coating-substrate systems processed at various conditions. Residual stresses were monitored using a curvature method and the results were compared to a X-ray diffraction  (XRD) method. As the applied aspect of  this work contemplates use of  the thick coatings as electrical insulators for  heating elements, the objective is also to characterize the coatings for electrical properties, in particular breakdown voltage and dielectric strength. CHAPTER  4 EXPERIMENTAL  METHODOLOGY 4.1 Processing of  Alumina Composite Sol-Gel 4.1.1 Alumina Composite Sols Boehmite (A100H, SOL-2PK, Condea, Germany) was used as a precursor for  the alumina sols and citric acid (CA, 99.7%, C 6H 60 7 , anhydrous, FisherScientific,  USA) was used as a dispersant and a bond coat precursor. Compositions of  the composite sol are given in Table 4.1-1. 0.1 wt.% of  CA in the total contents of  the coating solution was dissolved in 150ml of  distilled water, and alumina sol precursors (1.5M) were added into the distilled water and mixed at pH= 4. In order to obtain a homogeneous dispersion, the solution was ultrasonicated (with Ultrasonic Disruptor Probe, Horiba, USA) at room temperature for  10 min. The most suitable ratio of  the total amount of  alumina powders dispersed to the gel derived alumina sols was previously found  to be 86:14 [4.1]. Table 4.1-1 Preparation of  the alumina composite sol slurry by the concentration of  fine alumina particles, -0.5 fa,m,  in coarse alumina particles, ~3 fim Ratio of  Fine to Coarse Alumina Particles 30:70 35:65 40:60 45:55 50:50 H 2 0 (Distilled) (g) 150 150 150 150 150 Citric Acid (g) 0.234 0.234 0.234 0.234 0.234 AlOOH (g) 13.5 13.5 13.5 13.5 13.5 A1203 (g) 0.5 |j,m 21.15 24.68 28.2 31.73 35.25 3 |j,m 49.35 45.82 42.3 38.775 35.25 Description A30 A35 A40 A45 A50 In the current study, two different  particle sizes of  alumina powders (A16-SG, -0.5 (am, and P662B, -3 |am, Alcoa, USA) were dispersed in the alumina sol for  the fabrication  of  the composite sols with varying the ratio of  fine  to coarse particles (30:70, 35:65, 40:60, 45:55, 50:50; referred  to as mix A30, A35, A40, A45, and A50, respectively), Table 4.1-1. Ball-milling (-140 rpm) for  12h, 24h, and 48h was conducted for  optimizing the homogeneous alumina composite sols. Table 4.1-1 summarizes the sample preparations with the amount of each material. 4.1.2 Characterization of  Alumina Composite Sols Particle size distribution was analyzed by a particle size distribution analyzer using a light scattering method (Horiba CAPA-700, USA). The composite sols of  each concentration were diluted and put into a test tube provided along with the analyzer. The test tube was placed with a reference  (i.e. distilled water) in position and was centrifuged  at 960 rpm for -30 min. In addition to the particle size distribution, the specific  surface  area (m2/g) of particles in the composite sols of  each concentration was calculated by assuming spherical shape of  particles of  known density (i.e. 3.96 g/cc for  alumina). A water absorption method (generally called 'Archimedes Method') was used to measure the density of  the alumina composite sol-gel. Due to the difficulty  of  direct measurement of  coating porosity on aluminum alloy (i.e. due to coating separation in water during vacuuming), the density of  gel-cast CSG alumina was considered as representative of that of  the alumina coating, i.e. by neglecting kinetic effects  during spray coating. Alumina composite sols were gel-casted in a tray until they completely dried in air and were subsequently fired  at 300°C for  ~10min followed  by the phosphating (refer  to Section 4.2.3). Thereafter,  the samples were set in water and placed in a vacuum chamber. Water absorption was completed by the vacuuming, at pressure of  ~10 mmHg. Then all the samples were weighed for  the calculation of  the density as follows; where DA is the apparent density, W A the weight of  a sample in the air, W S the water-soaked weight of  the sample, and W, the weight of  the sample immersed in water. The weight of  the samples in air ("Wa") was pre-conducted prior to water absorption. A viscometer (Model DV-II, Brookfield,  USA) was used to measure the viscosity change of  the alumina composite sols with time. The alumina CSG slurry was placed in the viscometer right after  taking the slurry out of  the ball-milling device. Time interval between taking out and measuring the viscosity of  the slurry was approximately 2 min. Data were sequentially collected at 10 second intervals for  lh with a total of  360 data points. The minimum reading sensitivity is 20 cp for  the current equipment setting (i.e. 50 rpm, spindle #5). 4.1.3 Thermal Behavior of  CA Citric acid was used as a dispersant for  both alumina composite sol-gel and bond coat formation.  According to the literature [4.2], the melting point of  the CA is ~150°C and it decomposes at ~170°C. In order to verify  this thermal behavior of  CA, a Differential Thermal-Thermo Gravimetric Analyzer (DT-TGA, SETARAM TG96, France) was utilized. About 30 mg of  CA was placed in an alumina crucible and heated up to 500°C in a vacuum at heating rate of  10°C/min. In order to simulate the physical behavior of  CA in the alumina coating during heat treatment, porous sintered alumina with -60 vol % of  porosity (i.e. alumina slurry with a particle size of  3 (am was cast into a rectangle mold, dried, and then heat treated at 1200°C for  >10 hours) was impregnated with CA. In the model experiment, the sintered alumina was placed on CA (solid) in a Petri-dish, so the only bottom side of  the alumina contacted the CA. Then the Petri-dish was heat-treated up to 300°C to observe the pattern of  CA melting and wetting the alumina. Scanning electron microscopy (SEM, Hitachi S-3000N, Japan) and Energy Dispersive X-ray analysis (EDX, Hitachi S-3000N, Japan) were used to analyze the morphology and chemical composition of  such CA containing alumina. 4.2 Processing of  CB-CSG Coatings for  Thick Dielectric Films The CB-CSG coating process in the present study consists of  (i) preparation of composite sols (as described above) and surface  preparation (sand-blasting) of  AA5052 substrates; (ii) coating deposition (by spraying); (iii) specimen heat-treatment at 300°C; (iv) specimen impregnation with diluted phosphoric acid (by spraying) to achieve chemically bonded composite sol-gel, and (v) heat-treatment again at 300°C. These steps produced a single coating layer of  the thickness of  -40 (am. Therefore  to reach the total thickness of -200 (am, the process (steps ii-v) was repeated five  times. The steps of  heat-treatment and phosphating were conducted in sequence after  each coating layer deposition. The following sections explain the above steps in detail. 4.2.1 Surface  Preparation of  AA5052 Substrate Sand-blasting of  AA5052 substrates was conducted as a surface  preparation prior to the coating deposition, to promote interlocking between substrates and the ceramics deposited. AA5052 was cut into 55mmxl0mm as required for  the test of  Young's modulus and residual stress through curvature measurements. The substrate was sand-blasted (Cyclone, blasting system, USA) with a pressure of  -90 psi, using silica grit size of  60. Figure 4.2-1 compares the cross-sectional images of  the non-surface  treated and a sand blasted aluminum substrate. In the case of  the non-sand blasted substrate, the aluminum . . ' .... ' • • ' .•••••••••:•••:.• •:• ' . • . .. / • : . • • "">:' : ' • ' / ... . • . n t  '  ~  '  t^out., (a) (b) Fig.4.2-1 Cross sectional SEM images for  (a) non-sand blasted (b) sand blasted AA5052 substrate. substrate has an even surface,  resulting in the failure  of  a coating deposition due to the absence of  the interlocking between the coating and the substrate. On the other hand, a rough surface  is observed in the sand-blasted aluminum substrate, improving the initial bonding of  the coating through mechanical interlocking. The roughness of  a sand blasted substrate was measured by using a Form Talysurf  Series 2 (Talyor Hobson LTD., England), Fig.4.2-2. The method of  a center-line average Ra, was used and defined  as follows  [4.3]: 1 L R a = ~ l z \ d x (4.2-1) L 0 where z is measured from  the mean line and L the profile  sample length in the x-direction. Fig.4.2-2 Profilometer  of  the sand blasted aluminum substrate used in this work and schematic of  determination of  roughness value, Ra, using a mean line [4.3], The average roughness, Ra of  the substrate shown above is 5.87 |im. 4.2.2 Spray Coating Spray coating for  the composite sol-gel (CSG) coating was conducted on one side of the AA5052 substrates. The sand blasted AA5052 substrates were placed on a flat  board and sprayed (Delta Air Spray Gun, Graco Inc., USA) with alumina composite sols at a pressure of-25  psi, as shown in Fig.4.2-3. For the best coating surface  finish,  the slurry must reach proper viscosity range of  200cp-400cp. If  viscosity is too low, sprayed slurry flows  along the surface  of  the substrate. If  the viscosity is too high, the slurry is not sprayed properly and the spray gun is plugged. A maximum film  thickness of  40-50 (im without coating failure  (i.e. cracks causing coating separation during coating process) was achieved by single deposition. The coated substrate was then dried in air followed  by the heat-treatment at 300°C for  lOmin. Then, phosphating is required for  chemical bonding and hardening (details are in the following  section below). The processes can be repeated to achieve desired coating thickness. In the current study, 5 multi-layers were deposited in the same manner. Fig.4.2-3 Spray coating setup. Spraying has been repeated 5 times until the desired thickness of  coating (>200 i^m) is reached. 4.2.3 Chemical Bonding of  the Coatings by Phosphating 20 wt% diluted phosphoric acid (H3PO4, 85%, Fisher Scientific,  USA) was applied on top of  the CSG coated AA5052 via spray with a pressure of  -25 psi. The phosphoric acid penetrated into the coating pores during drying, and generating a chemical bonding via reaction between alumina and phosphoric acid during heat treatment, Fig.4.2-4. Alumina-hydrated alumina CSG layer Fig.4.2-4 Schematic of  the Chemically Bonded Composite Sol-Gel [2.33] Phosphating improves the interfacial  adhesion by forming  an aluminum phosphate between aluminum substrates and phosphoric acid according to following  reaction [4.4]: 2A1 + 6 H3PO4 2A1(H2P04)3 +3H2T (4.2-1) As shown in Eq.(4.2-1), hydrogen gas can be generated during phosphoric acid application onto the alumina coated surface.  It may cause coating failure  by delamination of  layers with excess amount of  phosphate during processing, so control of  the amount of  phosphate must be carefully  conducted. Thus, phosphating is one of  the most significant  steps of  the process and great deal of  care must be taken to execute it properly. If  phosphating is applied directly onto AA5052 prior to CSG coatings, thin aluminum phosphate film  formed  by the chemical reaction between phosphoric acid and AA5052 would be damaged by hydrogen, resulting in the decrease in the adhesion of  CSG to AA5052. This is the reason why post phosphating of Phosphoric Acid H3PO4, Phosphates Metal substrate: sand-blasted CSG coatings was executed in the current study. In order to determine the most suitable concentration of  phosphating in the current work, different  concentrations (i.e. 0.5, 1, 1.5, and 2 wt.%) of  diluted phosphoric acid were added into the pre-prepared composite alumina sols. Those composite sols were then poured ("gel-cast") in a circular aluminum tray and completely air-dried for  several days. The 2 wt.% of  phosphoric acid-contained alumina composite gels demonstrated no cracking even if it had a large shrinkage (i.e., -40 vol.%), while cracks occurred in all other concentrations. However, the problem of  controlling the amount of  phosphating during the spray processing of  the coatings still remains a challenge of  the current study. 4.2.4 Heat Treatment A heat treatment of  the CB-CSG alumina coatings is necessary to thermally activate the processes of  sol decomposition and chemical bonding (phosphating) reactions to cure the coating and to increase the coating adhesion. After  deposition of  CSG alumina and drying, the alumina coated AA5052 were heat-treated in air at 300 °C (at the heating rate of  -10 °C/min) for  lOmin and then slowly cooled (<17min) to room temperature. After phosphating (described in Section 4.2.3), the specimens were then heat-treated at the same conditions for  the chemical bonding. These steps were repeated until the desired coating thickness was obtained. In the current study, 5 layers were deposited and thus 10 heat treatments were executed. 4.3 Characterization of  CB-CSG Coatings for  Thick Dielectric Films 4.3.1 Dielectric Strength Measurements The Sentry 20 AC/DC Hipot Tester (QuadTech, Inc., USA) was used to measure the breakdown voltage, up to 5 kV-AC and 6 kV-DC. A copper flat-end  rod electrode with 3 mm diameter was utilized in contacting to a coating surface,  Figure 4.3-1. The voltage was applied for  at least 3 seconds according to safety  standard, at 5 points per coating sample (typically one point in the center and 4 points in the adjacent corners were selected). The smallest and the greatest values were discarded (the smallest value of  a breakdown voltage may significantly  affect  material's dielectric property in real cases. However, in the current study, no large differences  between the smallest and second-smallest values were observed. The values measured at each point are present in the appendix, Table A-4). Dielectric strength (DS, kV/mm) is defined  as DS=FB//, where V B is the breakdown voltage and t the thickness of  the sample. Fig.4.3-1 Setup of  the hipot tester for  the breakdown voltage measurement. 4.3.2 Curvature Measurements 'Talysurf  profilometer  and 3D surface  mapping (Taylor Hobson LTD, England) were utilized to measure the curvature before  and after  each heat treatment. This device has a stylus at the end of  an arm scanning across the surface  of  the sample, Fig.4.3-2. A side of sand-blasted AA5052 before  coating was faced  down on the station and the initial curvature of  the AA5052 was measured by probing the surface  three times with 0.5 mm span in each line and the averaged curvature for  each AA5052 was computed. After  each alumina coating deposition on such AA5052 and the heat treatment, the same side (opposite side or non-coated side) of  the coated samples as the surface  previously measured was being probed with the same condition, and then the change in the curvature before  and after  coating and heat treatment was observed. This process was repeated after  each deposition until the whole coating process was completed. Fig.4.3-2 Sample location in Talysurf  Profilomer. 4.3.3 Residual Stress Determination by X-ray Diffraction X-ray diffraction  (MultiFlex X-Ray Diffractometer,  Rigaku Co., Japan) was utilized to obtain the residual stresses of  the surface  of  the coatings for  validation of  residual stresses measured by curvature measurement. By knowing the strain changes of  the surface  of  the coatings as well as Young's modulus and Poisson's ratio of  a coating material, residual stresses of  the surface  of  each deposited coating layer can be calculated as described in Eqs.(2.4-15) through (2.4-17). In order to observe the strain changes of  the surface  of  the coated layers, it is necessary to measure the lattice spacing of  an unstressed material, which was the gel-cast alumina in the current study. This gel-cast alumina was heat-treated at 300°C followed  by phosphating by dipping into the 20 % diluted phosphoric acid and heat treated at 300 °C for  ~10 min, ground into powder, and scanned for  XRD reference  data. Strain for  each subsequent coating layer was measured and compared with the reference  data. After  completion of  the measurements, the software  (MDI Jade 7, Materials Data, Inc, USA) automatically calculates a lattice spacing of  each sample at a certain angle, 20. Intensity of the beam was 40 kV with 20 mA and 2-theta (20)  was set up from  20 to 80 degrees with a speed of  2 degrees/min. 4.3.4 Young's Modulus Measurement of  Alumina CB-CSG Coatings Average Young's modulus, E, of  the alumina coated AA5052 at room temperature and an elevated temperature up to 300 °C was measured by 'Dynamic Mechanical Thermal Analyzer (DMTA V, Rheometric Scientific,  USA), Fig.4.3-3, which is based on the 3-point bending theory. 'This unit examines viscoelastic or elastic materials according to temperature and frequency  dependent behavior. A small strain (deformation)  is imposed on the material by applying a stress. The amount of  strain resulting from  the applied stress leads to information  about the modulus of  the materials [4.5].' All measurements were repeated three times with different  coating samples and then average data were obtained. Specific  dimensions of  the specimen were required to fit  in the device (i.e., 55 mm in length and -6.5 mm in width for  a typical beam). The strain e (based on specimen geometry) of  the specimen was obtained by using Eq. 4.3-la and 4.3-lb. This strain was used as a Fig.4.3-3 Sample location in Dynamic Mechanical Thermal Analyzer, DMTA for  a stiffness and Young's modulus measurements for  beam samples. parameter in the operating software  and stresses exerted in the specimen were observed through Eqs. 4.3-2a and 4.3-2b [4.7], The average Young's modulus (<j/s) of  the specimen was then obtained using the formulas  (Eqs.4.3-1 and 4.3-2); e = K eD (4.3-la) K e=j  (4.3-lb) where s is the strain of  the sample, K s the strain constant for  a given geometry, and D the displacement of  the drive shaft  (=0.05 mm), t is the total sample thickness (substrate+coating) and I is the length of  span (= 40 mm). cr= K ax F  (4.3-2a) K a = ^ G c (4.3-2b) wt where a is the stress (MPa), K a the stress constant, w the width of  a specimen (mm), F  is the force  necessary to produce the sample displacement D (measured through the load cell), and Gc the gravitational constant (9.8 m/s2). In order to measure the Young's modulus of  the coated specimens at room temperature, the strain of  the sample was defined  (by the user based on the geometry of  the specimen using Eq. 4.3-1) and entered into a software  program as a test parameter; a load is automatically applied to the specimen to reach the specified  strain of  the specimen which was manually defined  in the beginning. Then this applied force  was used for  calculating stresses exerted to the coated specimen. In addition, average Young's modulus of  the specimens at elevated temperature up to 300 °C was also observed. For the observation of the change in the mechanical behavior of  the specimens during the heat treatment (including heating and cooling), a temperature of  the DMTA was held at 300°C for  30 min. Table 4.3-1 provides examples of  the data used in this measurement. A complete set of  data for  all samples is included in the Appendix Table A-1. For the verification  of  the measurement mentioned above, the change in Young's modulus of  coated samples at various temperatures (including heating up to 300 °C and cooling to room temperature) in terms of  time, DMA (Q800, TA Instruments, USA) has been used, Fig.4.3-4. DMA is the same as dynamic mechanical thermal analyzer (DMTA) but an alternative name. 3-point bending theory is also utilized throughout the test. Geometry of  all samples is required to be a beam (-55 mm X 10 mm). Those beam specimens are supported at both ends and a force  is applied in the middle. The amplitude (10 jam) of  the beams for the displacement is set up for  the constant strain of  the sample, and then the magnitude of  the force,-  IN, is applied to the beams. Table 4.3-1 Example of  sample geometries and strain used to obtain the Young's modulus by using Dynamic Mechanical Thermal Analyzer, DMTA Coating layers Total Thickness (mm, substrate + coating) Width (mm) Strain (f) AA5052 (Substrate) 1.56 6.32 0.000293 1st 1.6 6.45 0.000300 2 n d 1.65 6.48 0.000309 3 r d 1.69 6.5 0.000317 4 th <-th 1.75 6.5 0.000328 5® L8l 6^55 0.000339 Fig.4.3-4 DMA for  measuring average Young's modulus at various temperatures in 3-point bending. Temperature was ramped to 300°C at 10°C/min, and held for  30 min to simulate the coating heat treatment process. The change of  the Young's modulus on cooling was also measured. Then resultant stress/strain provides the average Young's modulus (<J/S)  of  the beams as follows. <7 = 3 • F  • L w-t2 £  = 6-S-t 2- L • 1 + — • (1 + v) • 10 \Lj (4.3-3a) (4.3-3b) where a is the stress (MPa), s the strain, F  the applied force  (N), 5 the amplitude of deformation  (fim),  L the half  sample length (span, mm), t the sample thickness (mm), w the sample width (mm), and v the Poison's ratio. The Average Young's modulus (coating + substrate), E, for  the coated AA5052 samples of  any given thickness at room temperature, was used to evaluate average residual stresses in coatings by combining Young's modulus of  added layer, E (-10 GPa, by micro-indentation method), curvature change before  and after  heat-treatment as well as geometry of coated aluminum. The calculation was previously described in Eqs.(2.4-10) through (2.4-14) in Chapter 3 and their results will be detailed in Chapter 6. CHAPTER  5 STUDY  OF ALUMINA  COMPOSITE  SOL-GEL  FOR THICK DIELECTRIC  FILMS 5.1 Characterization of  Alumina Composite Sol The coating process included two different  types of  alumina particles (-0.5 |am and ~3 |j,m average size) as fillers.  It was therefore  necessary to define  the most suitable composition of  the coatings. In this chapter, characterization of  composite sols and composite gel-cast samples are discussed. CSG viscosity, particle size distribution, and settling characteristics were investigated, and morphology and density were observed for alumina gel-cast. 5.1.1 Viscosity of  Alumina Composite Sols with Various Concentrations of  Fine Alumina Particles The viscosity of  the alumina composite sols with 5 different  concentrations (A30, A35, A40, A45, and A50) of  the fine  alumina particles, -0.5 (a,m, was compared for  various ball-milling times, and plotted as a function  of  aging times, for  up to 60min. Figure 5.1-1 shows the viscosity of  the alumina composite sols with various concentrations (A30-A50) after  12 hr ball-milling. There is significant  difference  between A50 and other samples. A50 exhibits initially higher viscosity than others and also shows rapid viscosity change up to -600 cp within 60 min, due to the overall higher surface  area of  the larger amount of  fine alumina particles in this sample. This large viscosity change over such a short period of  time indicates that this mixture will not be sprayable since the suitable range of  viscosity for spray-coating is 200-400 cp. Even though this range of  viscosity lasts -10 min for  A50, this fast  viscosity change may provide wide size distribution of  the as-sprayed CSG droplets (i.e., from  below ljam to tens of  micrometers), which results in uneven coating surface  during spraying. Therefore,  A50 with 12h ball-milling was not suitable for  the coating processing. Other concentrations of  A30-A40 show significantly  lower viscosity than the most suitable viscosity of  200-400cp for  spray coatings in the current work although the viscosity of  A45 has slightly increased after  50 min. 600.0 450.0 a o >> w 300.0 o o CO > 150.0 0.0 Time, min Fig.5.1-1 Viscosity of  alumina composite sols ball-milled for  12 hours. The concentration in the legend box stands for  the amount of  fine  alumina particles (~0.5p.rn) in the mixtures with coarse ones (~3|o,m). 0 10 20 30 40 50 60 The viscosity change observed for  the slurries is related to the settling rate of  alumina particles in the sol suspension. Settling velocity of  the particles depends on the size of particle and the viscosity of  a suspension. That is, unless the particles dispersed are small enough to be held in suspension by Brownian motion they will undergo settling. The particle settling velocity is given by [5.24], 2(p.  - p7)g a1 9 rj where a is the radius of  a particle, rj viscosity of  a medium, p\ and pi the densities of  the particle and fluid,  respectively, and ga is the gravitational constant. Eq. (5.1-1) is often  used to measure particle size by determining the rate of  particle settling [5.24]. The presence of  the large agglomerated particles leads to undesirable particle separation effects  caused by fast  sedimentation [5.14], As a result, a spindle measures viscosity of  the alumina sol suspension rather than whole composite slurry. Thus the slurry with high contents of  larger particles (i.e., < A50) shows lower viscosity than that with low contents of  larger particles (i.e. A50). Figure 5.1-2 compares the viscosity change for  all samples, versus time, after  24 hr ball-milling. According to the preliminary results, horizontal doted lines at the -200-400 cp represent viscosity of  the spray-able range for  the best coating result (of  the composite sols i.e., -10 |_im thickness in one direction pass without solution flow  along the edges of  the substrate as well as a fast  gelation of-10-20  min after  spray). The rapid viscosity change of A50 is observed until -1000 cp in 20min and it increases gradually. In other words, during the measurement of  the viscosity, the slurry becomes gelatinous and a gelatinous speed is dramatically increased at a certain time due to the hydrolysis and condensation mechanism of the sol-gel. This fast  gelation causes the dramatic increase in the shear stress (because of  a little liquid flow,  lowering a speed of  spindle) of  the surface  of  the spindle which results in an increase in the viscosity. After  this fast  gelation period, its gelation speed is slowly decreased causing gradual increase in the viscosity. The viscosity of  200-400 cp in A50 remains only for  ~5min. On the other hand, A40 and A45 show the slow increase of viscosity which is suitable for  spray coating. A30 and A35 exhibit viscosity less than 100 cp which is not sufficient  for  the spray coating due to the frequent  flow  of  solutions along the surface  during spraying. Regarding the most suitable viscosity of  200-400 cp and the time taken to reach the viscosity of  200 cp, A40 has been determined as the best candidate for spray coatings in the current work. Time, min Fig.5.1-2 Viscosity of  alumina composite sols ball-milled for  24 hours. Dot lines show the range of  the viscosity (-200-400 cp) suitable for  spray coatings. Figure 5.1-3 illustrates the viscosity change for  all samples as a function  of  time, after  48h ball-milling. The viscosities of  the A45 and A50 rapidly increase due to the result of  breaking down agglomerated alumina particles during the ball-milling. A40 demonstrates a gradual viscosity increase without abrupt change. Although A40 shows gradual change in viscosity, the spray-able time range is much shorter (e.g. ~10min) than that after  24 h ball-milling (e.g., ~30min). A30 and A35 still show a low viscosity until 50min aging and it shows the suitability for  spraying viscosity, 200-400 cp, after  50 min. 1800.0 1500.0 O.1200.0 o >> » 900.0 o o v> > 600.0 300.0 0.0 Time, min Fig.5.1-3 Viscosity of  alumina composite sol ball-milled for  48 hours. Dot lines show the range of  the viscosity (200-400 cp) suitable for  spray coatings. The concentration of  A40 offers  longest work time while A45 and A50 are beyond the workable range. A30 and A35 are below the viscosity range suitable for  spray coatings. 0 10 20 30 40 50 60 As a result of  these observations it is determined that higher concentration of  fine particles (from  A30 to A50) in the slurry and longer ball-milling time increases the slurry viscosity. Initially, ball-milling breaks down particles, and thus increases specific  surface area. Second (later) part of  ball milling is to mix the slurry homogeneously to achieve a stable suspension [5.27]. According to the above observations, supported by previously reported relationship between the viscosity and the surface  area [5.14,5.24,5.26], the slurry with 24h ball-milling has been selected for  further  experiments. Figure 5.1-4 demonstrates the change of  the specific  surface  area (determined by particle size analyzer (CAPA-700, Horiba, Ltd. Japan)) and the viscosity of  CSG (alumina particles dispersed in the alumina sols) by varying the concentration of  fine  alumina particles. Fig.5.1-4 Viscosity and specific  surface  area of  24h ball-milled mixed CSG with varying concentration of  the fine  particles. The higher surface  area correlates well with higher viscosity, for  all types of  samples. Based on these results, A40 with 24h ball-milling time has been selected for  further  research on CB-CSG spray coatings. 5.1.2 Particle Size Distribution in Alumina CSG Figure 5.1-5 shows the particle size distribution of  as received fine  (~0.5 (am) and coarse (-3 |am) alumina particles, as determined by Horiba Particle Size analyzer (refer  to Section 4.1.2). The bars indicate the frequency  (or wt%) of  each particle size range and the lines shows the cumulative frequency.  The majority of  particles (-80 %) are < 1 (am while the minority of  particles (-20 %) is >1 (am. Figures 5.1-6 show the particle size distribution for  the 24h ball-milled alumina composite sols with varying concentration of  the fine particles. It is observed that the particle size distributions are gradually wider with increasing concentration of  fine  particles. For example, the range of  the particle size distribution for  both A30 and A35 is from  0.2 (am to 3 |am, Fig.5.1-6a,b, from  0.2 jam to 5 (am for  A40 and A45, Fig.5.1-6c,d, and from  0.2 |am to 6fam  for  A50, Fig.5.1-6e, respectively. It is observed that the cumulative frequency  of  sub-micron sizes is < 40 % for  A30 and A3 5 while the frequency  of  the sub-micron sizes of  > 40 % occurs in > A40-A50. The maximum particle size of  alumina composite sols increases with the concentration of  fine  particles, likely due to agglomeration of  fine  particles during ball-milling. However, the agglomerated particles do not significantly  influence  the viscosity. Particle Size, jam Particle Size, |a.m Fig.5.1-5 Particle size distribution of  24h ball-milled alumina composite sols with single particle size fraction:  (a) fine  alumina particles and (b) coarse alumina particles (submicron range in the inset). 05 O 3 O a. 3 _ o e B > 3 3 era' Ul I o\ o <—K tr a> tfi E3 O) ^ I . O ?r w n> I-! to cr cr p 3 era © > LO O I . o rT N n> & w' r-K i"! a" c <—K o' E3 CD » cT 3 5 s» o 0 1 o on NJ O J o <D W c/> n" ro T 3 L»J y S S " > I O CT* S £ 3' y (ra cn o> Frequency, % IV) CD 00 O O O O O O O Frequency,% IV3 ^ cn 00 o o o o o o o Particle Size, |xm Particle Size, ^ im Fig.5.1-6 (continued) Particle Size, jam Fig.5.1-6 (continued) Table 5.1-1 summarizes the particle size distribution measurements for  the alumina composite sols as well as the specific  surface  area. The specific  surface  area slightly increases for  A30 and A3 5, and significantly  increases (by about 7 %) from  A3 5 to A40. Table 5.1-1 Summary of  particle size distribution of  the alumina composite sols after  24h ball-milling. Concentration of  Fine Alumina, wt% Specific Surface  Area, m2/g Standard Deviation, (j.m Diameter in Median, ^m Minimum/Maximum Particle Size, jam P662B alone 0.785 1.57 2.17 2/4 A30 1.914 0.54 1.23 0.2/3 A35 1.916 0.59 1.23 0.2/3 A40 2.056 0.81 1.13 0.2/5 A45 2.066 1.01 1.14 0.2/5 A50 2.076 1.24 1.17 0.2/6 A16 alone 3.782 0.64 0.44 0.2/3 No significant  effect  of  the concentration of  fine  particles on the viscosity was observed for A30 and A3 5 whilst the remarkable increase in the specific  surface  area for  A40 appears to leading to viscosity change (e.g., from  <40 cps to ~ 200 cps in the first  20 min after  ball milling) leading to the best performance  of  this suspension in the current spray coating. 5.1.3 Settling of  Alumina Particles in the Composite Sol Settling rate of  particles dispersed in a suspension is a simple method for  evaluating the degree of  dispersion, or stabilization of  suspension [5.23]. The agglomerated and coarser particles settle faster  while fine  particles in stable slurries settle slower and produce densely packed cake. According to Eq.(5.1-1), settling depends on the size of  particle and the viscosity of  a medium. Figure 5.1-7 shows the calculated settling velocity of  alumina particles dispersed in the alumina sol, according to Eq.(5.1-1). For the evaluation of  settling Radius of Particles, |xm Fig.5.1-7 Prediction of  the velocity of  settling alumina particles in the alumina sol suspension in terms of  the radius of  the particles. particles in the current study, the viscosity of  the fluid  (i.e., alumina sol suspension) used was 15 cp [4.1], density of  the particles and fluid,  3.96 g/cc and 1 g/cc, and gravity constant 9.81 m/s . According to Eq.(5.1-1), the velocity increases with the radius of  the particles. It is observed that the sedimentation velocity of  the fine  particles, -0.5 |am, is approximately 35 times slower than that of  the coarse ones at ~3 |am. Thus, higher concentration of  fine particles leads to higher viscosity of  CSG due to slower particle settling. To confirm  these predictions, the settling patterns for  particles in alumina composite sol were observed. Figure 5.1-8 shows the settlement pattern of  the as-received fine  and coarse alumina particles after  ball-milling for  12h and 24h. No settling of  fine  particles was observed (left  2 samples in all Fig.5.1-8) while the coarse particles (right 2 samples) substantially settled, i.e. the settling height increased up to 6 times within 12 hrs, in agreement with Eq.(5.1-1). The rate of  the sedimentation for  the fine  and coarse alumina samples is plotted as a function of  time, for  ball-milling time of  12 h and 24 h, in Fig.5.1-9. It is observed that, regardless of the ball-milling time, settling only happens for  the coarse particles. The settling rate of coarser particles decreases continuously, i.e., from  0.18 to 0.12 mm/min or from  3 to 2 (am/sec, in the first  3 hours of  test, possibly due to increased concentration and interaction between the particles. Assuming that the initial settling rate of  about 3 (am/sec is indicative of  the actual size of  settling agglomerate, Figure 5.1-7 may be used to estimate the average radius of  the settling particles. It is read that the radius is close to 2 (am, i.e. diameter close to 4 (am, indicating low degree of  agglomeration of  the nominally 3 (am large particles in the suspension. (c) (d) Fig.5.1-8 Settling patterns for  fine  (-0.5 (im) and coarse (-3 (am) alumina particles in alumina sol suspension with a ball-milling of  (a) 0 min (b) 30 min (c) 5h, and (d) 12h. Left  2 cylinders represent fine  alumina particles (with 12h and 24h ball-milling, respectively) while right 2cylinders represent coarse alumina particles (with 12h and 24h ball-milling, respectively) in all pictures (a-d). 0 120 240 360 480 600 720 840 Time,min Fig.5.1-9 The rate of  settlement of  two different  alumina particles, -0.5 (j,m and ~3 jam in alumina suspension of  viscosity of  10-20 cp. Slurries were ball-milled for  12 h and 24 h. Figure 5.1-10 exhibits the settlement pattern of  the alumina composite sols with varying concentrations of  fine  particles (i.e. samples A30 to A50) after  12 h ball-milling, for up to 24 hrs of  sedimentation time. The sample A50 shows settlement distance <1% of  the original height. Although the sedimentation is also observed for  A40 and A45, those samples show less settling distance and lower settling rate than A30 and A35. However, in order to have stable slurries for  spray, the slurries with no sedimentation are required. Thus, the settling results suggest that more extensive ball-milling is needed for  better dispersion of  the alumina particles in the sol suspension. Figure 5.1-11 plots the settling rate as a function  of (a) time and (b) concentration of  fine  alumina particles, based on Fig.5.1-10. In Fig.5.1-1 la, (c) (d) Fig.5.1-10 Settlement of  alumina composite sol after  12h ball-milling. Pictures were taken (a) right after  ball-milling, (b) after  lhr, (c) 12hrs, and (d) 24hrs after  ball milling stop. The samples are A30, A35, A40, A45, and A50 from  left  to right in each picture. Vertical arrows in (d) indicate the magnitude of  separation of  the alumina sol and alumina particles. fast  settling is observed during the first  30 min and then slower thereafter.  Figure 5.1-llb shows the settling speed in terms of  the concentrations of  fine  particles added. It is shown that the higher concentration of  coarse particles the higher the settling rate. In particular, A30 and A3 5 whose median particle size is > 3 |am demonstrate faster  settling than all others whose particle sizes is < 3 (am. It is also shown that the difference  of  the settling rate is up to 6 times depending on concentration of  fine  particles. It seems that ball-milling for  12h is not sufficient  for  the spray coatings. Thus, 24 h ball-milling was undertaken for  all the samples. Settling of  particles dispersed in alumina sol after  24 h ball-milling is illustrated in Fig.5.1-12. No significant  settling is observed for  all concentrations except for  minor settling observed in < A40 at 24 h monitoring. By comparing the previous results in Fig.5.1-10, the current result shows that the longer ball-milling time improves stabilization of  the suspensions, and thus less sedimentation of  particles. It is believed that this slower rate of  sedimentation was related to a degree of  homogeneity of  the fine  particles dispersed as a result of  ball-milling. For the case of  A30 and A3 5, the particle settling is still observed. This result suggests that longer ball-milling is required for  a homogeneous dispersion and particle breakdown. Based on Fig.5.1-12, the rate of  settling of  mixed particles was plotted in Fig.5.1-13. Settling happened only for  A30 and A3 5 and their settling rates decreased by as much as ~2 times as compared to those 12 hr ball-milled, Fig.5.1-12a. Faster settlement was observed in the first  -90 min. in both concentrations of  A30 and A35, and slowed down afterwards.  The settling tests agree with Stokes' Law, Eq. (5.1-1), i.e. fine  particles-containing-slurry settle 15 30 45 Time, min (a) 60 50 45 40 35 30 Concentration of Fine Alumina Particles, wt% (b) Fig.5.1-11 Rate of  settlement of  mixed alumina particles dispersed in alumina sol after  12 h ball-milling as a function  of  (a) time and (b) concentrations of  fine  alumina particles. (c) (d) Fig.5.1-12 Settlement pattern of  mixed alumina particles after  24h ball-milling. Pictures were taken (a) right after  ball-milling, (b) after  lh, (c) after  12h and (d) after  24h from  the ball milling stop. The samples are A30, A35, A40, A45, and A50 from  the left  in each picture. Vertical arrows in (d) indicate the separation of  the alumina sols and particles. Measurable settling was observed in A30 and A35 only. slower. It is concluded that higher viscosity and slower settling rate has been observed for higher concentration of  fine  particles with longer ball-milling time due to the better homogeneity and higher surface  area of  solids in suspension. Thus, the most suitable mixing ratio of  the fine  particles to coarse particles used for  the current study was determined to be 40 to 60 and appropriate ball-milling time is 24 hrs. Accordingly such alumina composite sols have been gel-cast and characterized for  morphology and density. 15 45 75 105 135 165 Time, min Fig.5.1-13 Rate of  settlement of  alumina CSG suspensions after  24 h ball-milling. 5.2 Characterization of  Alumina Composite Sol-Gel 5.2.1 Morphology of  Alumina Composite Sol-Gel with Various Concentration of  Fine Size Alumina Particles The objective of  this part of  the work is to determine how different  size alumina particles affect  CSG morphology, in particular the ability to decrease the porosity of  the final CB-CSG coating for  the thick dielectric films.  Subsequently, the best microstructure, consisting of  the smallest pores and the least amount of  pores, was chosen as a candidate for further  investigation of  spray coating process, including CA-derived interfacial  bond coat formation. Figure 5.2-1 shows the microstructures of  a fracture  surface  of  dry-pressed pellets of as-received (a) 100 % coarse alumina particles (~3 (am) and (b) 100 % fine  alumina particles (-0.5 fam),  pelletized by a pressure of  ~ 63 MPa (1 ton load on 1.4 cm diameter pellet). This test is a simple demonstration of  how such different  particles alone pack under the same conditions. Plate-like particles are observed in Fig.5.2-la, which randomly contact each other, through edge-edge, edge-surface,  and surface-surface  contact points. Thus, many large voids (i.e., resulting in overall porosity of  -55 %) are observed. The fine  particles, Fig.5.2-lb, pack much better, resulting in an overall porosity of  -45 % for  the pellet. Figure 5.2-2 shows the fracture  surfaces  of  gel-cast of  alumina composite sols (CSG) with various concentrations of  the fine  and coarse alumina particles (A30 to A50) and heat treated at 300 °C for  20 min. Sol-gel derived alumina is observed in voids between the crystalline alumina particles. The presence of  this sol-gel derived alumina is due to the low temperature heat-treatment of  alumina sol at 300 °C. It is well known that the first (a) (b) Fig.5.2-1 Fracture surfaces  of  pelletized alumina samples, (a) 100% coarse alumina particles, ~3 (im and (b) 100% fine  alumina particles, -0.5 [am. (e) Fig.5.2-2 Microstructures of  alumina CSG with the concentrations of  (a) A30, (b) A3 5, (c) A40, (d) A45, and (e) A50 of  fine  alumina particles (~0.5fim)  in gel-cast samples (all in the same magnification  of  3k). crystalline phase appearing during heat treatment of  the sol is a rhombohedral alumina (i.e., y-alumina), at -450 °C [5.11-5.13]. XRD data for  the currently used alumina sol with varying temperature from  200-500 °C are shown in Fig.5.2-3. It is observed that all peaks are broadened after  heat treatment at 300 °C. Also, no significant  phase change is observed up to 300 °C while the phase change (formation  of  y-alumina) is observed after  400 °C. Sol-gel derived alumina is well distributed throughout the alumina gel cast, Fig.5.2-2. It is therefore  expected that the sol-gel derived alumina acts as a binder for  the crystalline a-alumina particles and it is the first  to react with the phosphoric acid during the chemical bonding process [2.33], However, large fraction  of  voids are still observed in A30 and A35 as well as agglomerations; less voids and agglomerations are found  for  A40, and in particular for  A45 and A50 (refer  to Section 5.2.2 below for  detailed data of  density measurement for all these samples). 2-Theta Fig. 5.2-3 XRD spectra of  alumina sol after  heat treatment at 200-500°C 5.2.2 Effect  of  the Specific  Surface  Area on Density and Viscosity of Alumina Composite Sol-Gel The density of  gel-cast CSG samples is plotted in Fig.5.2-4. For the CSG containing only the coarse alumina particles (~3 |_im), the re-separation due to particles settling down during gel-casting is too fast  and therefore,  the density was not measured. However, the density of  CSG with only fine  (-0.5 |_im) alumina particles is 2.25 g/cc (i.e. porosity of -36%) which is close to that of  A40 and A45. This density gradually decreases with the increase of  coarse particles content in CSG, except for  A50 which shows higher density than 2.3 o 2.2 o o> & (A C d) o 2.1 2 -I 8 1 1 i 1 1 8 0 30 35 40 45 50 100 Concentration of Fine Alumina Particles, % Fig.5.2-4 Density of  gel cast CSG including a mixture of  two different  sizes of  alumina powders vs. concentration of  the fine  alumina particles. The size of  the fine  and coarse particles is -0.5 and ~3|iim, respectively. • - -f i * i 8 1 1 i  - a CSG containing fine  particles only. This result somewhat deviates from  literature data [2.30] where the highest density in binary system appears for  mixture ratio of  35-40 % fine  particles to 65-60 % coarse particles. However, this ratio has been obtained for  macroscopic systems, i.e. by vibrating steels spheres to fill  the voids between larger particles with smaller particles. Gel-casting is very different  process, where no external factors  (i.e., press or vibration) were applied, and at the same time electrostatic interactions between the fine  particles may affect the packing density. The gel-cast CSG density change with the concentrations of  fine particles correlates well with the specific  surface  area (m2/g), as shown in Fig.5.2-5. 30 35 40 45 50 Concentration of Fine Particles, % Fig.5.2-5 Comparison of  density, and specific  surface  area of  CSG as a function  of concentration of  fine  particles. 5.3 Summary Alumina composite sol for  spray coatings were investigated in terms of  viscosity, particle size distribution, settling rate of  particles, density, and morphology. The mixture of two different  particle sizes of  the alumina composite sol exhibits better properties than single size particles alone, i.e. more stable viscosity and higher density after  gel-casting. 12 hour ball-milled composite sols exhibit low viscosity of  < 40cp for  A30 and A45 while increase in viscosity was observed for  A50. 24 hour and 48hr ball-milled composite sols showed higher viscosity than 12 hour ball-milled slurry for  all concentrations due to the homogeneity of particle size distributions. A40 after  24 hour ball-milling shows the most suitable viscosity range of  200-400 cps for  -30 min, which is suitable for  spray coating. The ratio of  coarse to fine  particles at this concentration (A40) is close to 60:40 which is declared as the most suitable mixture in this study. Analysis of  the particle size distributions indicate that the more fine  particles in the suspension the greater is the maximum particle size. This is due to agglomeration of  fine  particles dispersed in the alumina sols. The presence of  submicron particles gradually increases from  37 % to 47 % for  the compositions A30, A35, A40, and then it is constant as the concentration of  the fine  particles increases in A45 and A50. Longer ball-milling time (i.e. 24 hours) provided less settling than a shorter time (i.e. 12 hours), due to increase of  the specific  surface  area of  particles in the slurry. The composite sols with higher concentration of  fine  particles demonstrated relatively higher density than ones with lower concentration of  fine  particles after  gel-casting. The most suitable ratio of  the coarse to fine  particles in the composite sols has been determined to be 60:40. Accordingly, the most suitable concentration of  A40 (40 wt.% of  the fine  particles) with 24 hours ball-milling has been selected for  spray coating to fabricate  thick alumina dielectric films  on AA5052. CHAPTER  6 CB-CSG  COATINGS  WITH  CITRIC  ACID-DERIVED BOND  COAT 6.1 Cross Sectional Morphology of  CB-CSG Coatings Figures 6.1-1 through Fig.6.1-3 show the cross sectional microstructures of  the CB-CSG alumina coated AA5052 using the slurry mixture of  A40 (composed of  40% of  0.5 (am fine  alumina particles and 60% of  3 (am coarse alumina particles, in alumina sol, refer  to Table 4.1-1 in Section 4.1.1 for  further  details). The first  coating layer, Fig.6.1-1, with the thickness of  -40 (j,m was homogeneously deposited on AA5052 sandblasted substrate without cracks. It is observed that the "amorphous phase", which was determined to be a composite of  alumina sol and the citric acid (CA) residue, appears uniformly  throughout the coatings and fills  the pores. This phase was also observed along the interface  as a ~l(j,m thin layer. This interfacial  film  is the key difference  between the samples free  of  CA and samples with CA, and forms  a unique organo-ceramic bond coat (BC) in this system. The effect  of CA on coatings will be discussed in more detail in section 6.2.2. The 2 n d layer does not introduce any new features  into this system. After  the 3 r d layer is deposited, resulting in -120 jam total coating thickness, Fig.6.1-2, vertical cracks near the surface  and a delaminated-ceramic-metal interface  are observed; these structural discontinuities are filled with the amorphous phase (which is later determined to be sol / CA derivative composite, refer  to Section 6.2.2). The interfacial  BC film  thickens to the range of  1-2 (am. The last (5th) coating segment deposition, Fig.6.1-3, results in a total coating thickness of-200  |am. The BC film  has further  widened up to -10 |o,m, and more vertical cracks form,  but again they are filled  with the amorphous phase. (c) Fig.6.1-1 SEM images of  the 1st layer of  the A40 coating, (a) overall view, (b) the interface, and (c) the coating volume, respectively. Fig.6.1-2 SEM images of  the A40 coating after  deposition of  the 3 r d layer, (a) overall view, (b) the interface,  and (c) the coating volume, respectively. (c) Fig.6.1-3 SEM images of  the coating after  deposition of  the 5 th layer of  the mixture of  A40. (a) overall view, (b) the interface,  and (c) the coating volume, respectively. Figures 6.1-1 through Fig.6.1-3 show the development of  the morphology of  CA-derived amorphous phase within the multilayered coatings. The thickness of  the amorphous phase at the interface  gradually increases with the additional coating layers, reaching up to ~5 % of  the total coating thickness. It is clear therefore  that the amorphous sol/CA derivative composite phase continuously migrates to the interface  during the consecutive steps of  the coating processing. Once at the interface,  the phase acts as a unique viscoelastic bond coat, helping to relax the residual stresses and thus to indirectly improve the coating micro structure (i.e. preventing thermal cycle cracking, and thus providing higher bonding strength and higher dielectric strength). CA-derived bond coat provided bond strength of  up to -15 MPa (observed with the first  coating layer deposited) while only 5MPa bond strength was observed for  the coating without CA. The unique ability of  the CB-CSG system containing CA is the segregation of  CA+sol nanocomposite towards micro structural discontinuities, e.g. the interfaces  but also cracks which may be forming  during heat treatment (thus thermal cycling) of  the coating. Such segregation, filling  the cracks, in essence heals the cracks and thus strengthens the coating. This hypothesis for  the presence of  CA derived amorphous composite phase (including alumina sol nanoparticles) is supported by studying the morphology of  the alumina bulk gel-cast CB-CSG impregnated with CA, as illustrated later in Fig.6.2-4. 6.2 Effects  of  Citric Acid on CB-CSG Coatings The studies of  CA-derived organic bond coats for  TDFs were prompted by the observation that the addition of  CA to the alumina-based CB-CSG slurries increased the coating adhesion on metals to -15 MPa for  coatings up to 200 |am thick from  -5 MPa (without CA and a thickness of  <50 jam). It was subsequently discovered that CA, during heat treatment of  the coating, is rejected from  the coating volume and thus forms  an organic-inorganic composite (i.e. organic phase derived from  partial decomposition of  CA, and mixed with alumina sol) layer at the interface  between the coating and the metal. The structural and chemical characteristics of  such "modified  CA" remain largely unknown in this work, and therefore  will be referred  to as CA-Derivative, CAD in short. An additional key observation was that no delamination of  the coatings from  the metal substrate was observed during processing, even for  relatively thick coatings, for  the systems including CA. The effect  of  CA on coatings will be discussed later in the section 6.2.2. A hypothesis was formulated  in the current study that the products of  the partial thermal decomposition of  CA de-wet the alumina and aluminum phosphates in the coatings, and thus migrate to the interface  (and to the top surface  of  the coating). At the interface,  such viscolastically deforming  films  substantially relax the residual stresses due to thermal expansion/contraction mismatch between the coating and the substrate. The results of research to verify  this hypothesis are presented in this chapter. 6.2.1 Behavior of  Citric Acid at Elevated Temperatures A Differential  Thermal-Thermo Gravimetric Analyzer (DT-TGA, SETARAM TG96, France) was used to investigate the thermal properties of  as-received CA and CA admixed to the alumina coatings. Gel-cast alumina, containing 0.1 wt.% citric acid, which was fabricated  under the same conditions as the coating slurry A40, was examined instead of using alumina coated aluminum due to the size limitation of  the equipment. For example, a crucible for  the DT-TGA is designated only for  a small amount (~0.5g) of  specimen material. DT-TGA results for  the as-received CA, Fig.6.2-la, are compared with the TGA results for  the gel-cast alumina with/without CA, Fig.6.2-lb, and the DTA results for  the gel-cast alumina with/without CA, Fig.6.2-lc. In Fig.6.2-la, an endothermic peak without the reduction of  mass is observed at 150 °C, indicating melting of  CA [2.47]. Upon further heating, mass reduction is observed between -200 and -250 °C due to an endothermic reaction of  dehydration. Two endothermic peaks are also observed at -330 °C where the mass reduction is almost completed due to the decomposition of  hydroxyl group (-OH) and carboxyl group (-COOH) [2.51], The products formed  by this decomposition at -330 °C are cis or trans aconitic acid or aconitic anhydride [2.51], Until this temperature, there is still a small amount (-10 %) of  CA present. Thus, the CA-Derivative, CAD, is a complex mix of the products of  partial decomposition of  CA, and not-decomposed CA. Figure 6.2-lb indicates a continuous weight loss through the entire thermal cycling due to the dehydration of  alumina CSG and CA decomposition. A slight difference  in the weight loss between CA-containing and non CA-containing alumina CSG is observed due to the dehydration and decomposition of  CA. The heat flow  curves of  both specimens in Fig.6.2-lc overlap, i.e. Temperature, °C Temperature, °C 1 \ j \ - \ \ V /  \ \ / \ / \ / o 100 200 300 400 500 Temperature, °C Fig.6.2-1 DTA-TGA data for  (a) as-received CA; (b) alumina CSG with/without CA (TGA only) and (c) alumina CSG with/without CA (DTA only). Heat flow  for  alumina CSG with/without CA overlaps, resulting in a single curve. thermal behavior of  the alumina CSG with/without CA is essentially the same (or beyond resolution capability of  the instrument). For the investigation of  phase changes of  CA at elevated temperatures (up to 300°C) with varying hold times at 300°C, X-ray diffraction  (XRD) spectra and Nuclear Magnetic Resonance (NMR, Avance 300 Spectrometer, Bruker Biospin Co., USA) have been utilized. In the XRD spectra, Fig.6.2-2a,b, it is seen that as- received CA is crystalline and it becomes amorphous at 170-300°C, although broad peaks at 20 = -20 and ~35deg are still present at 170°C and 200°C. No peak at 20= -35 is observed after  heat treatment at 300°C for  10-30 min while the peak at 20= -20 with a low intensity is still present. Figure 6.2-3 shows the result of  the NMR test for  citric acid heat-treated at 300°C for -10 min in air, resulting in CAD, and they are compared with the spectra of  as-received CA. In Fig.6.2-3a, the main peaks appear at 70-80 and 170-180 ppm. Each peak indicates methyl group, C-C, and >C=0, respectively. After  heat treatment at 300°C for  -10 min, new peaks are formed  at 10-20 ppm and 120-140 ppm indicative of  alkyl groups, i.e., methyl group, -CH3 or C-H, and an olefinic  carbon species, C=C, respectively [2.51,2.55] while the peak at 75 ppm disappeared. The peak at 170-180 ppm is still present, indicative of  a functional group such as >C=0< group [2.51,2.55]. This result agrees well with a study of  the NMR of CA [2.51], and indicates that CA is to some extent still present in CAD after  a heat treatment at 300°C up to 30 minutes, although the decomposition of  CA has been suggested to start at ~170°C [2.47]. Heat-treated CA (white as-received) appeared light brown and viscous at ~150°C. It turned into dark brown above 170°C and later gradually black CAD up to 300°C, 2 Theta (a) 300°C, 30min > 300°C, 20min 300°C, 10min 200°C, 10min 170°C, 10 min 10 20 30 40 50 60 2-Theta 70 80 90 (b) Fig.6.2-2 XRD spectra of  CA at elevated temperatures; (a) as-received and (b) after  holding at elevated temperature up to 300°C with varying hold time at 300°C from  10 to 30 min. 2 CO 1S0 100 50 i:0 •' PPm (a) '••••--,- I I '••••' - I • ' ' .'.'. I ..'•'.'..'. I .' ' • [.. •. • • • . 250: 200. _ 150 -ICO - .•• ; .0' •'..' ":-50:- • pprn (b) (c) Fig.6.2-3 13C NMR spectra of  CA (a) as-received, (b) heat-treated at 300 °C and (c) the reference  from  the literature [2.51], and was of  a solid consistency. It is anticipated that this amorphous black residue, termed in this work "CA Derivative" CAD plays a significant  role in the formation  of  the interfacial bond-coat. The color change indicates the gradual (or partial) decomposition of  CA, with the remains including complex organic material CAD with multiple functional  groups such as -CH3, C=0, and C=C, as seen in the NMR spectra. 6.2.2 Role of  Citric Acid in CB-CSG Coatings For better understanding of  the effect  of  CA on the coatings during the heat treatment up to 300°C and CAD formation,  porous (-50%) model sintered alumina have been studied before  and after  impregnation with a large excess of  CA (approximately -5 wt% of  CA in the impregnated samples). Figure 6.2-4 shows the comparison between fracture  surfaces  of  as-received (a,b) and CA-impregnated (c,d) porous alumina after  the heat treatment at 300°C. It is clearly observed that alumina particles are aggregated into relatively large (-3 |_im) clusters and pores between particles, Fig.6.2-4b. However, in Fig.6. l-4d which is also a magnification  of  Fig.6.2-4c, a continuous amorphous film  between particles as well as pores is observed, which is the CAD residue (after  heat treatment). (c) (d) Fig.6.2-4 Comparison of  the fracture  morphologies of  as-received (a,b) and (c,d) CA-impregnated porous alumina specimens (~3(im, -50 % porosity) followed  by heat treatment at 300°C. EDX analysis, Fig.6.2-5, supports the hypothesis of  the presence of  the CAD residue after 300°C heat treatment. Peaks for  100 % of  alumina are observed in Fig.6.2-5a while there is a new peak representing 28 at% carbon in Fig.6.2-5b (reliability of  determination of  the amount of  carbon by EDX is however low since carbon is one of  the lightest materials EDX can detect. However, the ratio of  other elements remains constant). This EDX result agrees well with DT-TGA and NMR confirming  the presence of  the residue CAD in the heat treated alumina samples. H kcV (a) (b) Fig.6.2-5 EDX analysis of  (a) the as-received and (b) CA-impregnated model porous alumina specimens. The key phenomenon controlling the in-situ formation  of  the organic CAD-based bond coat is the transfer  of  the CA from  within the coating volume, to the coating/metal interface  (and coating surface),  with simultaneous decomposition of  CA during the heat treatment. It is believed that the migration of  molten CA through the pores of  the specimen during heat treatment above ~200°C (wherein the migrating viscous CAD liquid carries also fine  alumina sol particles) may be caused by the change (increase) of  the wetting angle of CA-derived organic residue on alumina, due to increase of  the interfacial  energy T (at CA-alumina interface). As the wetting angle <9 is controlled through the well known equation, i.e. cos(0) = {r(sv>- r(si)}/ r(iv) (6.2-1) where 6 is the wetting angle, r (sv) the interfacial  energy between solid and vapor, r (si) between solid and liquid, and .T(lv) between liquid and vapor, /"(sv) is expected to remain constant during the heat treatment, whereas decomposition of  CA is expected to increase the /"(si) and r (lv), thus decrease cos(0)  and increase of  the wetting angle 0. The resulting dynamic de-wetting process aids in repelling of  the organic phase from  within the fine  pores of  coarse alumina particles network, towards the coarser pores, cracks, interface  and free surface  of  the coating (similar phenomenon is sometimes observed in experiments with mixed metal powders, wherein poor wettability of  the liquid component on the solid powder results in repulsion of  the liquid from  the mix [6.1]). The organic phase exuded from  the volume of  the coating may also carry with it the finest  fraction  of  the coating particles, i.e. the alumina-sol (AlOOH) nano-particles. The wetting of  alumina sol by molten CA is expected to be better than alumina due to the hydroxylated surface  of  the sol particles (this is confirmed  later in this Chapter), and thus de-wetting driven separation of  CA and AlOOH is more difficult  than separation of  CA and A1203. Effectively,  the interfacial  film  of  a nano-composite of  CAD mixed with alumina sol particles forms  the interfacial  bond coat. This hypothesis is further  verified  below. Figure 6.2-6 compares the cross sections of  CA-free  and CA containing CB-CSG alumina coated aluminum. Figure 6.2-6a illustrates the typical coating morphology of  the CB-CSG alumina, with voids between coarse alumina particles (~ 40±5 vol%, obtained by water absorption), and the alumina sol-derived phase filling  the pores and binding the particles together [6.2,6.3]. The -0.1 wt% CA containing CB-CSG alumina coating, Fig.6.2-6b, shows slightly less pores (~35±5 vol%) and amorphous phase at the interface  between the coating and the AA5052 substrate. EDX analysis of  this interfacial  film,  Fig.6.2-7, indicates the presence of  C, O, Al, and small amount of  P (hydrogen is not "seen" by EDX), suggesting that the film  is composed of  alumina (sol-derived), organic residue (CAD), and trace amount of  phosphate. Atomic ratio of  aluminum to oxygen is 1:2=0.5 in alumina sol (AlOOH) and 1:1.5=0.67 in calcined alumina (A1203). The Al/O ratio determined for  the Bond Coat (BC) layer, refer  to Fig.6.2-6, is about 1/2.6=0.38, suggesting that part of  the oxygen resides in CAD. Although hydrogen is not detectable by EDX, its amount in CAD is expected to be rather small, compare Fig.6.2-1, 2. As BC micro structure does not indicate presence of  any calcined alumina (these >0.5pm particles would be visible under SEM), it is reasonable to assume that BC includes AlOOH, with atomic ratio of  aluminum to oxygen of  0.5, and CAD. Neglecting the content of  hydrogen and phosphorous, the BC is therefore  composed of  approximately 10 at% of  Al, 20 at% of  O (in AlOOH), and 6at% of  O and 64 at% of  C in CAD. This would translate into approximate composition of  the BC of  40 wt% of  boehmite sol (perhaps partially decomposed sol) and 60 wt% of  CAD, both phases having poor crystallinity. • • i (b) Fig.6.2-6 Comparison of  alumina coated aluminum (a) without CA and (b) with CA. -40 vol% porosity is observed in (a) while - 3 5 vol% porosity is observed in (b) due to filling effect  of  CA. Porosity was measured by water absorption. kCts keV (a) 80 s ? 6 0 -' ro c o 2 40 +-> c o o c o O 20 C 0 Al P Element (b) Fig.6.2-7 EDX analysis, by (a) peaks and (b) quantification,  of  the Bond Coat layer between the CB-CSG coating and aluminum (the system includes additionally hydrogen, which is not detected by EDX). To further  investigate the above hypothesis, wetting angle of  molten CA at 150°C on the surface  of  sintered alumina was measured to be -135°, Fig.6.2-8a while the wetting angle of  the CA on gel-cast alumina sol sample was -70°, Fig.6.2-8b. The wettability of  CA on alumina sols and a sintered alumina has also been investigated at elevated temperatures up to ~300°C. The initial CA (i.e. as dissolved in water in 1:1 by wt.%) shows the wetting angle of -10°. As the temperature increases to 150°C, molten CA contracts (wetting angle increases) and it is pushed out from  the surface  of  the alumina while it bubbles due to the dehydration. Above 170°C CA color progressively changes to dark brown, indicating the decomposition of  CA as previously discussed in Section 6.2.1, while the wetting angle increases towards 135°. (a) (b) Fig.6.2-8 Investigation of  wettibility of  CA melted at ~150°C on (a) a sintered alumina and (b) sol-gel derived alumina. 6.2.3 Stress Mitigation for  CB-CSG Coated Aluminum It has been assumed that the bond coat (BC) consisting of  alumina sol and a CAD increased thermal shock resistance of  the coatings on AA5052 by relaxing stresses during the process, in terms of  successfully  made thick alumina coatings on AA5052. In order to verify this assumption, it was hypothesized that the sol+CAD bond coat may behave viscoelastically, and thus provides means to relax the differential  thermal expansion/contraction stress. In order to evaluate this hypothesis, three different  types of coated beam specimens were tested for  the pattern of  strain recovery in three point bending of  1) alumina coatings with and 2) without CA on aluminum alloy substrate, as well as 3) an as-received Al alloy substrate alone as a reference  sample. The procedure was as follows: constant force  of  18N was applied for  lh to each specimen and then removed; strain recovery was then observed and compared for  each beam, Fig.6.2-9 Both the AA5052 alloy substrate and the ceramic coating (Al203-based CB-CSG) can be considered as elastic materials at room temperature, while CAD/alumina sol is expected to behave as a viscoelastic material. If  this is true, the AA5052 substrate only and the substrate with CA-free  coating should recover to zero-strain immediately after  stress removal, whereas the CA-containing coated sample should exhibit non-zero recovery strain, decreasing gradually with time. Figure 6.2-9a illustrates essentially identical behavior for  the substrate only, and the substrate with the ceramic coating free  of  CA. Strain recovery for  AA5052 substrate only and CA-free  CB-CSG coated AA5052 seemed to be time dependant. The strain of  the 0.1 (a) -•-AI5052 -«- Coated Al with CA (%) -•-Coated Al without CA ("/ ~ 0.05 (0 V) Recovery zone 100 Time, min 200 70 80 90 100 110 120 Time,min 1.8 2 2 .2 (c) Coated AI5052 with CA y = -0.4876x- 1.2223 Coated AI5052 without CA — y = -1.0166X - 0.5139 • — A I 5 0 5 2 = ~ 1 - 6 0 6 6 x + 0 3 7 2 7 Log (Time),min Fig.6.2-9 Profiles  of  strain recovery after  removal of  the constant force  of  18N, applied to AA5052 substrate, and the substrate with alumina coatings with/without citric acid. The force  was removed after  holding for  60 min. CA in the legend refers  to citric acid, (a) overview, (b) magnified  between 70 and 120 min, and (c) log scale of  (b) AA5052 substrate at the applied force  removal point was about 0.0027 |am and it gradually decreased to the constant strain of  about 0.0018 |am. The difference  between these strains is however within the strain resolution of  DMA, so it can not be considered as significant.  The CA-free  CB-CSG coated AA5052 showed the strain of  about 0.0045 fim  at the applied force removal point and it decreased to about 0.0025 fim. The difference  between the strains was big enough to be considered as significant,  presumably due to slow crack growth in the coating. The behavior of  the CA-containing CB-CSG coated AA5052 also showed time dependent strain recovery. Its strain at the applied force  removal point was about 0.0083 (am and it reached the constant strain of  about 0.0060 (am after  120 min. However, this CA-containing CB-CSG showed the highest strain at the end of  the constant force  application period, and it also showed the highest residual strain after  the force  removal. This is presumably due to the combination of  progressive cracking in the coating, and deformation of  CAD interface.  While the AA5052 substrate recovered essentially all strain upon removal of  the applied stress, and the sample CB-SCG coated without CA also showed the same trend (98% recovery), CA-containing CB-CSG coated specimen recovers only 92 % of  its original deformation.  These observations are not sufficient  to claim that CAD+sol behaves viscoelastically at the interface  between CB-CSG coatings and AA5052 substrate. Observation of  the coating's retention on various samples, Fig. A-5 in the Appendix, suggests a strong effect  of  CA on mitigation of  residual stress. Further research on viscoelastic properties for  CAD+sol is suggested for  the future  work. The log-log plots in Fig.6.2-9c indicate strain recovery patterns for  the three types of specimens by assuming the following  equation [2.79]; £(t)  = £0e r (6.2-2) From the above equation, it can be understood that the higher relaxation time, x, the lower the strain changes. For example, each slope of  the sample indicates the value of  the 'logeh (or 0.434/T)' by taking the log function  in Eq.(6.2-2). For each slope, one can obtain the relaxation time,x, for  each sample as follows;  x=0.891 for  the CA containing sample, x= 0.427 for  the CA free  sample, and x=0.270 for  AA5052 alone. This result shows that the CA containing sample has the highest value of  x, showing the longest stress relaxation time. It is also observed that rather higher x is obtained in the CA free  sample than AA5052 alone. It is presumably due to the defects  in the coatings (i.e., porosity and cracks). Therefore,  CA containing sample in Fig.6.2-9c describes sufficiently  well viscoelastic behavior of  these samples [2.79] Viscoelastic materials can be characterized in terms of  relaxation modulus, defined  by the following  equation [2.79]. t E(t)  = E-e~ (6.2-3) The comparison of  the modulus of  alumina coated AA5052 (with/without CA) and AA5052 substrate alone was studied at sinusoidal force  of  40 MPa applied for  lh in order to verify  the above demonstrated viscoelastic effect  of  CAD on coatings (refer  to Section 4.3.4 for  the details of  procedure and calculation of  the modulus). Fig.6.2-10a summarizes the results of this test in linear axis system and Fig.6.2-10b in log-log scale. It is observed that the relaxation modulus of  as-received AA5052 substrate decreases during the first  period of loading time, -30 min and then it remains approximately constant. The decrease in modulus of  AA5052 at constant stress loading may be related to relaxation of  residual stresses introduced during forming  and machining of  the alloy (refer  to the Appendix Fig.A-1). The relaxation modulus of  CB-CSG alumina coated AA5052 without the CA also decreases during the test with a similar rate and time constant as un-coated specimen. The effects  of the CA are weak as behavior of  the AA5052 substrate dominates behavior of  the whole coating systems as it is much thicker than coating (-1600 |_im vs -80 jam). At t=0, the moduli of  the three specimens are 64 GPa, 59 GPa and 58 GPa respectively. The lower value of  relaxation modulus of  CB-CSG alumina-only-coated AA5052 at t=0 is due to lower Young's modulus of  the alumina coating (10 GPa, obtained by indentation method, refer  to Section 4.3.4). Similarly, relaxation modulus for  the CA-containing coated AA5052 at times is even lower than one with no CA containing coating, due to the stress relaxation by CAD in the alumina/Al interface.  In terms of  the relaxation time, T, the lowest x (=31.000) is obtained in the CA containing sample while the relaxation times for  the AA5052 only and the CA free  sample are 60.278 and 43.838, respectively. This result agrees well with that obtained from  Fig.9.2-9c which showed the highest T in the CA containing sample. According to the above result, it can be concluded that the CA plays an important role as a stress relaxation during coating processing. These results may also be compared with calculations of  the moduli, assuming the lower value of  Young's modulus for  additional layers, and of  the CAD bond coat. According to Eq.(2.4-12), if  the modulus of  the substrate of  thickness ho is Eo, and a layer of  coating thickness, Ah, is added (with modulus E), the modulus of  composite specimen is now re CL O § 60 "5 T5 O S C o m 50 x m Q) a: 40 10 20 30 40 Time, min 50 60 70 4.9 re CL (!) ST D) o (a) As received Al y = -0.0072X +4 .8133 CA-containing alumina coated Al y = -0.014x + 4.7692 4.7 1 Log(time), min (b) Fig.6.2-10 Relaxation moduli of  CA containing CB-CSG alumina coated AA5052, compared with as received AA5052 and conventional (CA-free)  CB-CSG alumina coatings on AA5052. Thickness of  the coatings was -100 |_im. (a) linear axis scale and (b) log-log scale. Ei=(Ah-E+ho-Eo)/hi,  where hi=ho+Ah. Using the specific  geometry of  the test samples (i.e. Ah=40 (am and ho= 1600 jam) and average Young's modulus of  alumina coated AA5052 (i.e. Ei=58 GPa), one can calculate the Young's modulus of  the added coating layer, E. Using the procedure outlined above, the calculation of  E of  the added layer of  CB-CSG coating produces a negative value of  about -180 GPa, indicating Ei < Eo (and considering h/=Ah+ho  ~ ho due to ho »Ah). It means that the CB-CSG alumina coating with CAD bond coat does not contribute to the increase in the stiffness  of  the system. This is believed to be yet another evidence that the CAD intermediate layer (bond coat) relaxes the interfacial  stress during specimen deformation  (this effect  will be discussed in more details in Section 6.3.2). 6.3 Determination of  Residual Stresses in CB-CSG Alumina Coated Aluminum alloy (AA5052) Residual stress in the coating/substrate system produces curvature changes of  the coated substrate. In the current research, it was expected that residual stresses in thick multi-layered CB-CSG alumina films  on aluminum alloy is due to the relatively large thermal expansion mismatch (i.e. alumina: 8xlO"6/°C, believed unchanged due to inclusion of  about ~2 wt% of  phosphates in the chemical bonding process, and AA5052, 24xlO"6/°C, believed to be unmodified  by the minor alloying elements in AA5052). Monitoring of  residual stress as a function  of  process parameters, in particular coating thickness and presence of  CAD, allows optimizing the deposition process. In order to determine the residual stress value, the curvature change of  the coated substrate was measured before  and after  coating deposition and heat treatment. 6.3.1 Curvature of  CB-CSG Alumina Coated AA5052 Alloy System Figure 6.3-1 shows the radius of  curvature of  CB-CSG alumina coated AA5052 for each additional coating layer, as the overall coating thickness is increased. The initial curvature of  AA5052 substrate before  coating deposition has been measured to be 2.2 ± 0.5 m, presumably due to machining and sand-blasting induced stress (the Appendix Fig.Al). 4.5 3.5 2.5 40 80 120 Coating Thickness, n.m 160 200 Fig.6.3-1 Radius of  the curvature of  AA5052 aluminum substrate before  (i.e. at zero coating thickness) and after  deposition of  coatings of  increasing thickness. Subsequently, the measurements were repeated after  each layer of  coating deposition followed  by heat treatment at 300°C. It is observed that each additional coating layer alters the bending of  substrate/coating system, i.e. increase in radius of  curvature is observed. The radius of  the curvature of  the substrate with the first  coating layer deposited (-40 fj,m) increases from  2.2 m to 3.6 m, i.e. the system becomes more flat  after  the first  coating layer deposition. This radius of  curvature (3.6m average) remains approximately constant after deposition of  the subsequent layers of  the coating, although the scatter of  results is relatively large. Considering just the average values of  the radius of  curvature, deposition of  the second coating layer (-80 p.m in total) decreases the radius, and then it gradually increases again with each additional layer deposition up to 200 |im of  the final  coating thickness. It appears therefore  that cracks, affecting  the state of  stresses and thus beam curvature, formed  in the coatings until the thickness of  -80 jam. The cracks appear to form  in the third layer of  the coating, i.e. exceeding 120 jam thickness. The maximum stresses caused by thermal mismatch between the coatings and aluminum substrate might be in the first  coating layer. The stresses subsequently relax during follow-up  heat treatment after  deposition of  the second coating layer. The observed change of  sample curvature agrees well with the literature, i.e. it has been observed that the curvature can shift  from  convex to concave during repeated heat treatment due to the balance of  residual stresses [6.4], 6.3.2 Young's Modulus of  CB-CSG Alumina Coated AA5052 According to the Eq.(2.4-14), average Young's modulus of  the coated specimen is important factor  in determination of  the residual stress developing in the additional coating layers. Thus, the change in average Young's modulus of  the coated sample, with each layer of  coating deposition, was determined in Section 2.4.2.1. Figure 6.3-2 shows the comparison of  the average Young's modulus for  CB-CSG alumina coated AA5052 calculated using Eq.(2.4-12) [2.60] (using the modulus of  the coated film  determined through indentation to be ~10 GPa) and the modulus experimentally obtained by Dynamic Mechanical Thermal Analyzer (DMTA), as a function  of  the coating thickness. Coating Thickness, jam Fig.6.3-2 Average Young's modulus of  coated AA5052 aluminum, as determined by calculation and experiment. The calculated average Young's modulus of  the coated AA5052 agrees well with the experimental average Young's modulus until the accumulated coating thickness of  120 |j,m, and subsequently the measured values deviate (decrease) from  the calculated ones, reaching difference  of-9  GPa (or -14 %) for  the coating thickness of  -200 jam. This difference  is attributed to accumulation of  defects  (i.e., cracking), with coating thickness [6.5,6.6], It appears therefore  that -80 |am is the critical thickness for  the coating to start significant cracking due to accumulation of  residual stresses. Figure 6.3-3 compiles the results of  measurements of  the change of  average Young's modulus of  as-received AA5052 substrate and coated samples with temperature, when the consecutive layers of  coatings are added to the sample (Dynamic Mechanical Analysis was used, as described in Section 4.3.4). As expected, Young's modulus of  all samples decreases during heat treatment at up to 300°C, and increases again during cooling. At constant temperature of  300°C, held for  30 min (Fig 6.3-4b-f)  an increase in the modulus of  coated samples is observed while the modulus of  un-coated AA5052 remains constant. This effect  is attributed to the progress of  chemical bonding process in the coating layers (refer  to Section 2.1.4). However, as determined above, the parallel process of  coating damage accumulation, i.e. cracking, in coatings thicker than 80 (im counter-balances the effect  of  the chemical bonding on sample stiffness.  This complex interplay of  stiffness  increase due to chemical bonding, and decrease due to accumulation of  microfracture  in the coatings cooled to near-room temperature, determines the final  modulus of  the coatings. For example, only the first and the second layer coated substrate return to their initial Young's modulus through the entire heating cycle, as elasticity of  the AA5052 dominates the system Fig.6.3-3 Young's modulus of  (a) as-received AA5052 and (b)-(f)  CB-CSG alumina coated AA5052 aluminum measured during heating and cooling cycles: (b) 1s t layer deposited (c) 2 n d layer added (d) 3 r d layer added (e) 4 th layer added and (f)  5 t h layer added. behavior and there are no defects  in the coatings. Significant  decrease in the modulus during the cycle is found  between the 3 r d and the 4 th layer, indicating damage to the coatings. This conclusion is supported by analyzing cross-sectional morphology of  the coatings in Fig.6.1-1 through Fig.6.1-3. Heat treatment of  the coating/substrate system results in complex and interactive phenomena including chemical bonding, differential  thermal expansion/contraction and damage accumulation. During the heat treatment, both the coating and the substrate are free to expand without constraint as the coating (and the interface)  has no integrity due to lack of chemical bond within the coating. The chemical bonding process at 300°C provides, however, a bond within the coating, and between the coating and the interface,  thus increasing coating stiffness.  Upon cooling, differential  thermal shrinkage (i.e. larger in the AA5052 substrate than alumina coating) introduces stress which may lead to coating cracking, thus decreasing its stiffness.  The process repeats for  the consecutive layers of  coating deposition. For example, the loss of  the Young's modulus for  the third layer is ~ 3.6 %, and -8.9 % for  the 4 th layer, and 8.3 % for  the 5 t h layer. 6.3.3 Strain Measurement by X-ray Diffraction  (XRD) Figure 6.3-4 shows XRD spectra of  the chemically bonded CSG alumina coatings, under the stress-free  state (refer  to Fig.A-4), Fig.6.2-4a, which is a reference  ground gel-cast material, and for  the multilayer CB-CSG coated aluminum in the order of  1s t to 5 th layer, Fig.6.2-4b-f. Fig.6.3-4 XRD peaks of  alpha-alumina at (a) stress-free  state and (b)~(f)  under the stressed state in the order of  1st to 5 t h layer. The aluminum substrate peaks are observed for  the coated samples until the 3 r d layer due to the coating penetration by the incident beam of  40 kV, and by the diffracted  signal. It is observed that the main peak at 29 = 35.24 has slightly been shifted  toward the smaller 29 angle until the 3 r d layer and shifted  again toward the larger 29 angle thereafter.  The shift  is indicative of  lattice spacing change due to residual stress. Tensile stress increases lattice spacing, so the corresponding x-ray peaks shift  to smaller angles [6.7]. This phenomenon is observed in Fig.6.3-5a-b. A different  direction in the angle shifts  is observed from  between the first  and second layers and at 3 rd layer (-120 |am thick), in agreement with the curvature data in Fig.6.3-1, Section 6.3.1. Figure 6.3-5 shows the strain in terms of  relative lattice spacing shift,  Ad/d,  calculated through 29 (degree) angle shifts  for  the major peaks, as a function  of  coating thickness, using formula  Eq.2.4-16, section 2.4.2.2. The diffraction  angle shift  reflects  the state of  stress within the coating, i.e., tension with an angle shift  toward smaller values and compression towards larger [6.7]. The angle shift  toward smaller values occurs between stress-free  state and 1s t coating layer and it remains until the 2 n d layer. The angle shifts  towards larger values between 2n d and 3 r d layer. The associated strain essentially returns to near-zero values, if  the scatter of  the results is taken into the account. This indicates stress relaxation, likely through coating cracking, as determined previously. These all explanations can be supported by cross sectional morphologies of  coatings in Figs. 6.1-1 through 6.1-3. •o < w 0.004 0 . 0 0 2 - 0 . 0 0 2 -0.004 35.24 35.20 a> a> o •g 35.16 CN^ T 35.12 G> c < 35.08 35.04 Coating Thickness,|im (a) 40 80 120 160 Coating Thickness,nm (b) 200 Fig.6.3-5 Comparison of  (a) strain and (b) angle shift  by XRD vs. coating thickness. 6.3.4 Average Residual Stresses in the Alumina CB-CSG Coating Layer on Aluminum alloy (AA5052) Based on the lattice spacing change, the state of  stresses can be determined at a certain coating thickness, using Eqs. (2.4-17a,b) in Section 2.4.2.2 The surface  of  the 1st coating layer, -40 |_im thick, was in a state of  tension. The magnitude of  tensile stress decreased with additional coating layer deposition as the previously deposited (underlying) coating layers decrease the difference  in CTE between the additional coating and the underlying substrate (of  aluminum alloy and the already deposited CB-CSG coatings). The stresses cause bending of  the substrate/coating system [6.4]. By knowing the curvature, Young's modulus, and geometry of  coated substrates, the residual stress can be obtained according to the bending theory [2.60]. The respective formulas  and method of  calculation are included in Section 2.4.2.1. Figure 6.3-6 illustrates the result of  the calculations of  residual stresses in CB-CSG alumina coated aluminum through the thickness of  alumina coating layers. The empirical values of  average Young's moduli of  AA5052 substrate and coated samples were used as discussed in Section 6.3.2, as well as the calculated average Young's moduli (refer  to Section 2.4-12 for  the formula).  No significant  difference  in the residual stresses calculated using the above two values of  Young's modulus was found  throughout the thickness even at the thickness of  200 (j.m, where the difference  of  10 GPa of  Young's modulus was found,  as shown in Fig.6.3-3. Tensile stresses (100-170MPa) were dominant up to the 2n d layer of thickness of  80 |im (i.e. the radius of  the curvature decreases), then they gradually change to approximately constant (about -50MPa) compressive stresses (the radius of  the curvature increases). This was previously predicted by the result from  the curvature change: the direction change in the curvature of  the alumina coated aluminum happens between the 2 n d (80 |im) and 3 rd (120 (am) layer. At the thickness of  120 (am and above, the coating has the lowest, constant compressive stress. It is expected that during the heat treatment after  the 3 rd layer of  deposition, cracks form  and therefore  the stresses are related. Since the CB-CSG alumina coating has a relatively high level of  porosity, -45% and a low Young's modulus (i.e. 10 GPa), it is also reasonable to expect a low value of  the residual stresses. The maximum residual stress measured is -180 MPa in tension in the first  layer of  thickness -40 (am, and the minimum stress is -80 MPa in compression in the 5 th layer, 200 fim  thick. Fig.6.3-6 Comparison of  residual stress obtained from  bending theory, using calculated Young's modulus and experimentally obtained Young's modulus by 3-point bending test for alumina CB-CSG coatings on AA5052. The residual stresses calculated based on sample bending are compared with those obtained by XRD measurement in Fig.6.3-7. The residual stresses obtained by XRD measurement are indicative of  the surface  stresses due to the shallow penetration of  X-ray. As discussed in Section 6.3.3, strain changes in each coating layer were used to determine the residual stresses. It is observed that the tensile stresses are dominant up to the 2n d layer, 80 (am, and change to compressive stresses at the 3 r d layer, and then the state of  the stress shifts to the tensile stress again. The XRD derived trend is supported by the results of  the curvature measurement. The differences  may be partially attributed to the specifics  of  each method: the values obtained through the curvature measurements refer  to the average residual stress through the coating layers, while the values by XRD refer  to the surface  of  the coating material and potential errors may also arise from  the measurements of  the shift  angle (20) due to the relatively high scanning speed. The resulting scatter is relatively large, which may lead to a general conclusion that significant  stresses are present in coatings thinner than 120 |am, and are subsequently relaxed through cracking to near-zero value for  thicker coatings. (0 CL </> tn a> (/) "ro 3 T3 'w 0 01 200 150 f 100 50 0 -50 - 1 0 0 -150 40 • By Curvature method • By XRD method 80 1110 i 160 2{j0 Coating Thickness, ^m Fig.6.3-7 Comparison of  the residual stresses determined by XRD and curvature methods 6.4 Electrical Properties of  Alumina CB-CSG Coatings Thick dielectric films  (-200 |im), consisting of  5 consecutively deposited layers of CB-CSG alumina, each ~40|am thick in a single deposit, with CA-derived bond coat, have been successfully  deposited on AA5052. The microstructures and the thickness of  the alumina coatings affect  the dielectric strength of  coatings, particularly crack and pores [6.8]. Those defects  degrade the electrical properties: the more pores or cracks the lower the breakdown voltage. The fabrication  of  the CB-CSG alumina multi-layered coatings with the dielectric strength of  >2000 Vac is the applied target for  the current study. To determine the dielectric strength, the breakdown voltage of  the coatings was measured as a function  of coating thickness. However, as shown above, the thickness of  the coatings is also a parameter determining the level of  residual stresses, and eventually cracking of  thicker coatings. As a result, it is also essential to study the relationship between residual stresses and dielectric strength in terms of  coating thickness. 6.4.1 Dielectric Strength of  CB-CSG Alumina Coatings 6.4.1.1 Effect  of  the shape of  an electrode In order to apply the CB-CSG technology for  the manufacture  of  low cost and high efficiency  heating elements, a dielectric breakdown voltage of  over 2000VAc is necessary for the ceramic coatings. However, the measured value of  the breakdown voltage depends not only on the properties of  coating materials but also the geometry of  an electrode which directly contacts the surface  of  the insulating material [6.8]. A sharp-ended electrode generally reads a higher value than a flat-ended  one. The tip of  the sharp-ended electrode has less contact area to a coating surface,  leading to the low probability of  discharging by gases (i.e., air) when an electric field  is applied. Since it is difficult  to fabricate  a perfectly  even and smooth coating surface,  flaws  such as cracks and roughness always exist throughout the surface  of  the coatings. These flaws  may introduce unexpected discharge leading to low breakdown voltages. Discharge is the phenomenon of  ionization of  electrons in an electrode [6.8]. Therefore,  the breakdown voltage values will be different  depending upon both the shape of  the electrode and a part of  the coating surface  where the electrode is placed. In this study, the effect  of  the shape of  the electrode on the dielectric breakdown voltage of  the alumina coated aluminum has been investigated, Fig.6.4-1. As expected, a higher breakdown voltage is obtained for  the sharp-ended electrode, especially for  the coating thickness above 80 jam where more defects  (i.e., pores or cracks) are present in the coating. Therefore,  it is reasonable to use a flat-ended  probe in this study for  more accurate (averaged-out over the electrode end) data collection. 3000 oi:nn • r ai-e ueu e euuuue 2500 " % > T g, 2000 " " ro a O T II > 1500 - n 1 * i " o 1 2 1000 - I as f Q> T m * 500 -0 1 1 i 1 — 0 40 80 120 160 200 Coating Thickness, ^m Fig.6.4-1 Comparison of  the effect  of  the shape of  the electrode (sharp end vs. flat  round end) on the dielectric breakdown voltage of  CB-CSG alumina coatings deposited on aluminum. • Sharp-ended electrode • Flat-ended electrode } t 6.4.1.2 Dielectric Strength of  CB-CSG Alumina Coated AA5052 The breakdown voltage and dielectric strength are related through the micro structure and thickness of  coating material. The dielectric strength is inversely proportional to the thickness of  coatings after  certain thickness is reached, even though the breakdown voltage increases with the thickness. This is because the probability of  flaws  occurring in the coating increases with the coating thickness. We believe that only thermal breakdown and ionization breakdown components operate in the current coatings due to the local heat generation by the electrode and pores or cracks present in the coatings. Since the dielectric strength is related to the microstructures of  material, it is worthwhile plotting the porosity of  the coating material to better understand the relation between the dielectric strength and the porosity. Figure 6.4-2 shows open porosity (vol%) of the consecutive CB-CSG alumina coating layers deposited on aluminum, as determined through water absorption. The porosity displays a linear increase with each added layer. The increase in the porosity is mainly caused by the cracks, affecting  dielectric strength as coating thickness increases. Figures 6.4-3 and Fig.6.4-4 show the result of  the measurements of  breakdown voltage and the dielectric strength of  the CB-CSG alumina coated aluminum, respectively. Complete data on the breakdown voltage of  the alumina coated AA5052 is included in the Appendix, Table A-4. The breakdown voltage for  the coatings of  the total thickness of  -200 |-im is -2100 Vac and the highest dielectric strength of  ~15kV/mm is observed at the first layer. The breakdown voltage for  the 1st layer was higher than that of  the two-layered Coating Thickness, jam Fig.6.4-2 Open porosity of  the consecutive alumina coating layers. coating, and the trend continues through layer #5 although at smaller rate. The 2 n d layer of thickness of  -80 |im can be considered as a critical thickness regarding dielectric strength. This thickness agrees well with the study by Olding et.al  [2.67] of  a similar system based on alumina-silica dielectric coatings. The dielectric strength, Fig.6.4-4 obtained from  the result of  the breakdown voltage in Fig.6.4-3, is inversely proportional to the thickness due to the increase in the density of  flaws  (pores and cracks) with thickness of  the coating. For the 1s t layer of  40 (am, the dielectric strength is -15 kV/mm. However, at the 2 n d layer of  80 jam, the dielectric strength decreases with thickness and this trend continues in each subsequent layer. The relation between residual stresses (as determined through sample curvature measurements) and the dielectric strength as a function  of  the coating thickness is plotted in Fig.6.4-5. Residual stresses decrease until the 4 th layer is deposited, at overall coating thickness of  160 jam, and slightly increase after  this thickness. Similarly, the dielectric 3000 80 120 Thickness, nm 160 200 Fig.6.4-3 Breakdown voltage of  the CB-CSG alumina coated aluminum as a function  of thickness of  the coating. The average thickness of  one single deposit is ~ 40 ^m and the coating is made of  5 layers. 120 Coating Thickness, nm 200 Fig.6.4-4 Dielectric strength of  the CB-CSG alumina coated aluminum with the thickness of the coating. strength decreases between the thicknesses of  40 and 80 p.m, and then gradually decreases thereafter.  For the 1s t layer, high values of  the residual stresses and dielectric strength are obtained due to no significant  density of  defects  present in the coatings. 200 150 (0 9- 100 • Residual Stress Measured by Curvature • Dielectric Strength </> CD CO « 3 35 o 50 0 i -50 -100 -150 40 80 1210 160 Coating Thickness, nm 200, 18 15 12 6 0 E E > o> £ <D k. U) o 'C o 03 O Q Fig.6.4-5 Relationships between residual stress (A) and dielectric strength ( • ) with respect to coating thickness. However, after  the 2 n d layer, the residual stresses decrease along with the dielectric strength due to the increment of  the density of  cracks and pores. After  this thickness, the state of  the stresses changes to compression without the significant  difference  in the dielectric strength. Although the presence of  defects  is one of  the determining factors  for  the dielectric strength, accumulation of  coating layers partially compensate for  the loss of  dielectric strength due to increasing density of  defects. 6.5 Summary The properties of  CB-CSG alumina coated AA5052 alloy have been studied in terms of  the process parameters affecting  the mechanical and dielectric properties. The citric acid (CA) dispersant combines with the alumina sol component of  CSG, partially decomposes during the heat treatment into an amorphous phase starting at ~150°C, and fills  the cracks and pores within CB-CSG alumina. It also migrates to the interface  between the coating and the substrate. It is hypothesized that the sol/CA derivative containing CB-CSG alumina coatings shows viscoelastic behavior and thus aids in stress relaxation especially in the regions of  high density of  sol/CA derivative composite, such as an interface. With the knowledge of  geometry, average Young's modulus E of  CB-CSG alumina coated AA5052, and the Young's modulus E of  the consecutive CB-CSG alumina layers, and the coated sample curvature before  and after  heat treatment at a given temperature, the residual stress of  coated layers was calculated. The residual stresses obtained by the curvature method were compared with those obtained by monitoring lattice parameter change by X-ray diffraction.  The values of  stresses obtained by each technique were different  but similar trend was obtained. The difference  may be attributed to the specifics  of the measurement methods: the X-ray method determined only the stresses at the surface  of the coatings while the curvature method determined the average stresses through the entire thicknesses of  the coating. The stress in coating layers was predominantly tensile (especially for  thinner coatings) while the substrate is in compression. The maximum stress determined by both methods was found  in the 1st layer due to the strong effect  of  thermal mismatch in the coating layer near the substrate. After  the 1st layer has been deposited and heat treated, this stress continuously decreased. Eventually cracks formed  in the thicker coatings and thus the stresses decreased. The dielectric strength of  the CB-CSG alumina coated aluminum decreased with the increase in the coating thickness. The maximum dielectric strength of  ~15 kV/mm was observed for  the 1s t coating segment deposition and this value decreased for  each subsequent layer. However, the scale of  decrease between the 1s t and 2 n d layer is different  from  other layers due to the increase in the density of  pores. After  this initial thickness, in addition to pores, cracks also form  due to the thermal mismatch during processing, further  decreasing the dielectric strength. The relationship between the dielectric strength and the residual stresses in the coating as a function  of  the thickness of  the coatings has also been investigated. The trend of the residual stresses agrees well with the dielectric strength of  the CB-CSG alumina coated AA5052 in terms of  the coating thickness due to the increase of  defect  density in the microstructure. The maximum value in the 1st layer for  both the residual stress and dielectric strength means that there were few  defects  in the coatings. From the 2 n d layer deposition, the residual stress decreases, which means that the defects  present release the stress, but also the presence of  cracks lowers the dielectric strength. Therefore,  it is observed that the dielectric strength and the residual stresses in the coatings are related to the microstructures of  the coatings. CHAPTER  7 CONCLUSIONS This thesis concerns novel processes for  thick films  of  ceramics on metal substrates with substantially larger coefficient  of  thermal expansion. The general objective was to propose and analyze the new processing methods to decrease residual stress in the films  and thus to prevent large-scale cracking in the films.  The newly developed process was then used to achieve relatively thick (~0.2mm) dielectric films  of  alumina-based ceramic on aluminum alloy AA5052, by Chemically Bonded Composite Sol-Gel (CB-CSG) process. This study of the CB-CSG alumina coated AA5052 aluminum system proposed and evaluated novel processing methods to overcome the current limits of  deposition of  thick ceramic coatings on metallic substrates with large difference  in CTE. The principal strategies undertaken in the material process engineering involved: (i) multi-layer film  deposition, (ii) composite sol-gel slurries with bi-modal particle size distribution, (iii) chemical bonding of  the film  through phosphating, and (iv) introduction of  organic-phase (citric acid) derived bond coats at the interface  between the AA5052 substrate and the ceramic coating. As a result of  development and investigation of  the above novel CB-CSG processing techniques thick alumina-based coatings have been successfully  deposited on AA5052 substrate and characterized. The thickness of  the coatings was approximately 200|im, and they were processed through spray-deposition of  consecutive layers, approximately 40 jam thick each, heat-treated at 300°C for  -10 min after  each deposition. Two size fractions  of alumina powders ("fine"  of  average size 0.5 jam and "coarse" of  3 jam) were used in formulation  of  the Composite Sol-Gel (CSG) slurry. The mixing ratio of  the fine/coarse particles content was investigated based on slurry viscosity and stability, as well as properties of  the final  coating. The coatings were characterized for  microstructure, residual stresses and dielectric strength, as a function  of  the process parameters. The most important finding  of  this work is that it is possible to deposit thick ceramic films  on aluminum alloy substrates, if  all four  processing strategies listed above are implemented simultaneously. In particular, the citric acid - derived organo-ceramic bond coats seem to play important role in relaxing residual stresses resulting from  differential thermal contraction and expansion. Based on the results of  this work, the following  specific conclusions are drawn in reference  to the four  process development categories listed above: (i) composite sol-gel slurries with bi-modal particle size distribution This was introduced to decrease the overall process temperature, and to decrease film  strain during thermal treatment as well as to increase the density and stability of composite sols. The sol-gel based slurry allows to decrease the process temperature to 500°C level, but large shrinkage (up to 90 vol.%) is a significant  disadvantage, leading to failure  of coatings thicker than about 1 [j,m. A composite sol-gel (CSG) is introduced to decrease such a high shrinkage by adding ceramic fillers  into the sol, providing the composition of  a solid body of  86 vol.% of  the filler  and 14vol% of  the sol-derived phase. It was observed that the longer the ball-milling time and the larger the amount of  the fine  particles the higher the viscosity of  the solution. A40 composition with 24 h ball-milling demonstrated the appropriate viscosity required for  the spray-deposition of  the coatings, which is 200-400 cp. The investigation of  CSG slurry stability through observation of  the settling patterns supports the importance of  ball-milling time in achieving slurry stability. For the same concentration of  fine  particles, the longer ball-milling improved CSG slurry stability, as evidenced through the slower settling time. The 12 h ball-milled A40 CSG showed the settling rate of  0.16 mm/min for  the first  15min while no sedimentation was observed in the 24 h ball-milled A40. Based on these results, A40 has been selected as an appropriate CSG concentration for  the spray coatings of  thick dielectric alumina films  on AA5052. (ii) chemical bonding of  the CSG films Chemically bonded composite sol-gel (CB-CSG) technique provides decreasing process temperature to 300°C. CB-CSG also increases the adhesion (to the level of  lOMPa) between AA5052 and the alumina coatings and as such increases the system resistance to thermal mismatch stresses. The aluminum phosphate bonds the alumina filler  particles, and thus provides denser and harder ceramic than that of  ordinary sol-gel process. (iii) multi-layer coating deposition The thickness limit of  a single layer deposition without coating failure  was experimentally determined to be -40 |_im, due to excessive residual stress accumulation in thicker coatings caused by the relatively large thermal expansion mismatch between the coating and the substrate. Therefore,  multi-layer deposition reduces the stresses in total alumina coating thickness of  -200 ^m is deposited with 5 multi-layers of  -40 |am thick each without failure  of  coatings. (iv) citric acid derived bond coats Citric acid (CA) melts and partially decomposes during the 300 °C heat treatment in coating processing, providing poorly defined  CAD ("citric acid derivative") amorphous phase, forming  a CAD+sol derived alumina amorphous bond coat at the interface  with AA5052. Three point bending test was applied to quantify  the effect  of  CAD-based bond coat on stress mitigation in the system, with constant force  of  18N to the coated sample for 60 min. After  the force  removal, it was observed that alumina coatings on AA5052 without CAD bond coat recovered in 120 minutes -99 % of  the deformation,  while only -92 % of  the recovery was observed for  the same time for  the CAD bond coat containing samples. This was presumably related to the viscoelastic behavior of  the CAD bond coat, and it allows deposition of  films  of  up to 200 |_im thickness. (v) residual stress in the films Residual stresses in the coating, after  accumulation of  consecutive coating layers, were evaluated by curvature measurement and XRD method. The maximum residual tensile stress of  -150+30 MPa was observed after  first  layer deposition (at -40 p.m) while the residual compressive stress of  -60+20 MPa was reached at the fourth  layer, the thickness of -160 |_un. XRD method demonstrated the same trend of  the stresses. The maximum residual tensile stress of  100+60 MPa has been determined for  the first  layer, -40 |im, and the residual compressive stress of  -40+60 MPa for  the fourth  layer, at -160 (am. (vi) dielectric strength of  the films Dielectric strength of  the CB-CSG alumina coated AA5052 aluminum was determined from  the measurements of  the breakdown voltage (in AC). As expected, the breakdown voltage (kV) increased proportionally to the coating thickness. The dielectric strength reached a maximum of  about 15 kV/mm for  the first  layer, and subsequently decreased to reach minimum of  about 10 kV/mm for  the fifth  coating layer. The most rapid decrease of  the dielectric strength was found  between the first  layer of  -40 |_im and the second layer (total coating thickness -80 ^m), and it gradually decreased thereafter.  It is believed that this decrease in the dielectric strength after  the first  layer is caused by increased density of  cracks in the coating, as also evidenced by decrease of  the residual stress in the coatings. CHAPTER  8 RECOMMENDATIONS  FOR FUTURE  WORK Although it has been shown that the combination of  the CB-CSG technique and CAD organo-ceramic bond coats substantially expanded the possibilities of  deposition of thick dielectric ceramic films  on light metals, the effects  of  CAD on the coating behavior and properties have not been fully  clarified.  This issue is important if  the technology is to be transferred  to other coating systems. Therefore,  further  investigations including theoretical and experimental approaches are required to verify  the influence  of  CAD bond coats on the coatings and system characteristics. The composition of  CAD organo-ceramic bond coats was determined by EDX and it is observed that such bond coats are made of  partially decomposed CA and alumina gel-derived sol. No exact chemical composition was determined. Therefore,  it is recommended to investigate the bond coat composition by using XPS, XRD and other methods such as DTA/TGA. The hypothesis regarding viscoelastic behaviour of  the CAD organo-ceramic bond coats has not been fully  confirmed  in this study. Therefore,  it is recommended that detailed nano-indentation studies are performed  on the interlayer, to support (or reject) the hypothesis on the viscoelastic behavior of  CAD organo-ceramic bond coats. Adhesion of  the CB-CSG coatings, even with use of  CAD bond coats, remains relatively low at -10 MPa. This is believed to be due to the large difference  in CTE between alumina and AA5052. It is recommended to evaluate further  methods to decrease the effect of  the large difference  in CTE on bonding strength of  the coatings. Particle size-graded multi-layers and controlled porosity (low stiffness)  layers may be deposited and evaluated. The use of  composite films  may also be promising to increase the adhesive strength of  the films.  Alumina/silica composites may be introduced by dispersing alumina in silica sol. It is anticipated that better coating adhesion would result due to the presence of  amorphous phase of  silica. In addition, the dielectric strength of  the coating would be increased due to high dielectric strength of  the silica (i.e., 25-40 kV/mm). REFERENCES [1.1] M. Nakamichi and H. Kawamura, "Out-of-Pile  Characterization of  AI2O3 Coating as Electrical Insulator," Fusion Engineering and Design, 59-58, 719-713 (2001). [1.2] L.N.Semenyuk, "Electrical Properties of  Alumina Films," Inorganic Materials, 32 [12], 1310-1311 (1996). [1.3] K.S. Shamala, L.C.S. Murthy, and K.N. Rao, "Studies on Optical and Dielectric Properties of  AI2O3 Thin Films Prepared by Electron Beam Evaporation and Spray Pyrolysis Method," Mater. Sci. Eng. B, 106, 269-274 (2004). [1.4] Hyungkeun Kim and Tom Troczynski, "Thick Alumina Dielectric Films on Aluminum through Chemically Bonded Composite Sol-Gel," Ceramics International, In Press, Corrected Proof,  Available online 18 January 2006. [1.5] J.D. Sibold, "Wear Applications," pp.973-977, Ceramics  and  Glasses, Engineered Materials  Handbook,  Vol.4, ASM International, USA, 1991 [1.6] Kim H, Chen A, Yang Q, and Troczynski T, "Sol-Gel Alumina Environmental Barrier Coatings for  SiC Grit" Materials Science and Engineering A-structural materials properties microstructure and processing, 420 (1-2): 150-154 March 25 2006. [1.7] M. Liu and A.V. Joshi, "Oxygen Sensors," pp.1131-1139, Ceramics  and  Glasses, Engineered  Materials  Handbook,  Vol.4, ASM International, USA, 1991 [1.8] J.X.Ahang, R.S. Chandel, Y.Z. Chen, and H.P.Seow, "Effect  of  Residual Stress on the Strength of  an Alumina-Steel Joint by Partial Transient Liquid Phase (PTLP) Brazing," J. of  Mater. Process. Technol., 122, 220-225 (2002). [1.9] C.H. Hsueh and Evans, "Residual Stresses in Metal/Ceramic Bonded Strips," J. Am. Ceram. Soc., 68 [5] 241-248 (1985). [1.10] Y.Chiang, D BBirnie III, and W.D Kingery, "Special Topic 5.3, Thermal Shock Resistance," Physical Ceramics, John Wiley & Sons, Inc., pp.464-466 (1997). [1.11] T. Olding, M. Sayer, and D. Barrow, "Ceramic Sol-Gel Composite Coatings for Electrical Insulation," Thin Solid Films 398-399, 581 (2001). [1.12] H.Kozuka, S. Takenaka, H. Tokita, T. Hirano, Y. Higashi, and T. Hatani, "Stress and Cracks in Gel-Derived Ceramic Coatings and Thick Film Formation," J. Sol-Gel Sci. Techn., 26, 681-686 (2003). [1.13] L.F. Francis, A.V. McCormick, and D.M. Vaessen, "Development and Measurement of  Stress in Polymer Coatings," J. Mater. Sci., 37, 4717-4731 (2002). [1.14] David J. Green, "9.8 Thermal Stresses and Thermal Shock" pp.298-3.1, An Introduction  to the Mechanical  Properties  of  Ceramics,  Cambridge University Press, 1998. [1.15] P.C. Hidber, T.J. Graule, and L.J. Gauckler, "Competitive Adsorption of  Citric Acid and Poly(vinyl alcohol) onto Alumina and its Influence  on the Binder Migration during Drying," J.Am. Ceram. Soc., 78[7] 1775-80 (1995) [1.16] P.C. Hidber, T.J. Graule, and L.J. Gauckler, "Citric Acid-a Dispersant for  Aqueous Alumina Suspensions," J. Am. Ceram. Soc., 79[7] 1857-67 (1996). [1.17] L.Zhang, S. Raghavan, S. Meikle, and G. Hudson, "Inhibition of  Alumina Deposition during Tungsten Chemical Mechanical Planarization through the Use of Citric Acid," J. Elecrochem. Soc., 146[4], 1442-1447, (1999). [1.18] K. Vishista, F.D. Gnanam, "Role of  Deflocculants  on the Rheological Properties of Boehmite Sol," Materials Letters, 58, 1576-1581 (2004). 1.19] J.W. Fischer, L.H. Merwin, R. A. Nissan, "NMR Investigation of  the Thermo-analysis of  Citric Acid," Applied Spectroscopy, 49 [1],120-126 (1995). 1.20] T. Troczynski and Q.Yang, "Process for  Making Chemically Bonded Sol-Gel Ceramics," US patent No.6284682 Sept.4, 2001. 1.21] Werner Rausch, The  Phosphating of  Metals,  SAM International, Finishing Publication LTD, 1990. 1.22] M. Miyayama et al., "Engineering Properties of  Single Oxides," Vol. 4, Ceramics and  Glasses, Engineered  Materials  Handbooks,  ASM International, 748-757. 1.23] R.C Buchanan, Ceramic  Materials  for  Electronics.  Processing,  Properties,  and Applications,  Marcel Dekker, Inc., pp.2-77, 1986. 1.24] ASM  Handbooks  online," available: http://products.asminternational.org/hbk/index.jsp, Dec. 08, 2006 2.1] R.W.Cahn, P.Haasen, and E.J.Kramer, Materials Science and Technology, Processing of  Ceramics part 1,17A, pp.192-207 (1996). 2.2] L.C. Klein, Sol-Gel  Coatings  in Thin  Film  Process II,  pp 501-522, Academic Press, 1991 2.3] D.R. Uhlmann, "Sol-Gel Science and Technology. Current State and Future Prospects," J. Sol-Gel Sci. Technol. 13, 153 (1998). 2.4] M. Guglielmi, "Sol-Gel Coatings on Metals," J. Sol-Gel Sci. Technol. 8, 443 (1997). 2.5] J.D. Mackenzie and E.P.Bescher, "Physical Properties of  Sol-Gel Coatings," J. Sol-Gel Sci. Technol. 19, 23 (2000). 2.6] D.R Uhlmann, T.Suratwala, K.Davidson, J.M. Boulton, and G.Teowee, "Sol-Gel Derived Coatings on Glass," J.Non-Cryst Solids 218, 113 (1997). 2.7] Y.Liu and L.Gao, "Dispersion of  Aqueous Alumina Suspensions using Copolymers with Synergistic Functional Groups," Mater. Chem. Phys. 82, 362 (2003). 2.8] Y.Jia, Y.Kanno, and Z.Xie, "New Gel-Casting Process for  Alumina Ceramics based on Gelation of  Alginate," J. Euro. Ceram. Soc. 22, 1911 (2002). 2.9] H.Scholze, "Glasses and Ceramics from  Gels," Proceedings of  Second International Workshop on Glasses and Glass Ceramic from  Gels, J.Non-Cryst. Solids.63, 1984 2.10] C.W.Turner, "Sol-Gel Process-Principles and Applications", Ceramic Bulletin, 70,9, 1478-1490(1991). 2.11] L.L. Murrell, N.C. Dispenziere Jr, and K.S. Kim, American Ceramic Society Symp. Series, 437, 97(1990) 2.12] B. C. Jeffrey,  and GW.Scherer, "2.1 Solution chemistry of  inorganic precursors", Sol-gel science, the physics and chemistry of  sol-gel processing, 60-67, Academic press, Boston, 1990. 2.13] T. Troczynski, Lecture note. MMAT 382, The Department of  Metals and Materials Engineering, The University of  British Columbia, 2002. 2.14] Jones,R.W, "Fundamental principles of  sol-gel technology", The Institute of Metals, U.S.A., 1989. 2.15] T. Troczynski, Lecture note. MMAT 382, The Department of  Metals and Materials Engineering, The University of  British Columbia, 2002. 2.16] Jones,R.W, "Chapter 5. Formation of  Sols", Fundamental principles of  sol-gel technology, 40-48, The Institute of  Metals, U.S.A., 1989. [2.17] R.K. Iier, The Chemistry of  Silica, Wiley Interscience, 1979. [2.18] Blinker, D.E.Clark, and D.R.Ulrich, "Better Ceramics through Chemistry," Materials Research Society, 1984. [2.19] C.J. Blinker, D.E.Clark, and D.R.Ulrich, "Better Ceramics through Chemistry II," Materials Research Society, 1986 [2.20] C.J. Blinker, D.E.Clark, and D.R.Ulrich, "Better Ceramics through Chemistry III, "Materials Research Society, 1988. [2.21] George John, "Post Deposition Treatment of  Thermal Sprayed Coatings," Masters Thesis, The University of  British Columbia, 1996. [2.22] Robert D. Shoup, "Sol-Gel Processes" Engineered  Materials  Handbook,  Vol.4, Ceramics and Glasses, Edited by Materials Information  Society, pp 445, 1991. [2.23] David A. Barrow, T.Edward Petroff,  and Michael Sayer. "Method for  Producing Thick Ceramic Films by a Sol Gel Coating Process, US patent number: 5585136, Dec. 17, 1996. [2.24] S.Wilson, H.M. Hawthorne, Q.Yang, and T.Troczynski, "Scale Effects  in Abrasive Wear of  Composite Sol-Gel Alumina Coated Light Alloys," Wear 251,1042 (2001). [2.25] D.A. Barrow, T.E.Petroff,  and M.Sayer, "Thick Ceramic Coatings using a Sol-Gel based Ceramic-Ceramic 0-3 Composite," Surface  and Coatings Technology, 76-77, 113-118 (1995). [2.26] L.L.Murrell, "Sols and Mixtures of  Sols as Precursors of  Unique Oxides," Catalysis Todays 35 225-245 (1997). [2.27] Chih-Peng, Shaw-Bing Wen, and Ting-Tai Lee, "Preparation of  Nanometer-Sized a- Alumina Powders by Calcining an Emulsion of  Boehmite and Oleic Acid," J.Am.Ceram.Soc.,85[l], 129-33 (2002). [2.28] Chih-Peng and Shaw-Bing Wen, "Variations in a Boehmite Gel and Oleic Acid Emulsion under Calcination," J.Am.Ceram.Soc., 85[6] 1467-72 (2002). [2.29] Rex W. Grimshaw, "Chapter 7. Physical and Structure Properties of  Raw Ceramic Materials," The  Chemistry  and  Physics of  Clays,  4 t h edition, 395-411, Wiley-Interscience Division, John Wiley and Sons Inc. U.S.A,, 1971. [2.30] R.K. McGeary, "Mechanical Packing of  Spherical Particles," J. Am. Ceram. Soc., 44 [10], 513-522 (1961). [2.31] James S.Reed, "13.2 Packing in Interstices among Coarser Particles" in Introduction  to the Principles  of  ceramic Processing,  pp. 188-194, John Wiley & sons, Inc., 1988. [2.32] D.D.L. Chung, "Acid Aluminum Phosphate for  the Binding and Coating of Materials," J..Mater.Sci. 38, 2785-2791 (2003). [2.33] Carolyn Moorlag, "Chemically Bonded Composite Sol-Gel Ceramics. A Study of Alumina Phosphate Reaction Products," Master's thesis, Department of  Metals and Materials Engineering, The University of  British Columbia, Aug. 2000. [2.34] J.V. Bothe, Jr, And P.W. Brown, "Low-Temperature Formation of  Aluminum Orthophosphate," J.Am Ceram.Soc. 76 [2] 362-68 (1993). [2.35] J.V. Bothe, Jr, And P.W. Brown, "Reactivity of  Alumina toward Phosphoric Acid," J.Am Ceram.Soc. 76 [9] 2353-58 (1993). [2.36] M.Vippola, S. Ahmaniemi, J. Keranen, P. Vuoristo, T. Lepisto, T. Mantyla, and E.Olsson, "Aluminum Phosphate Sealed Alumina Coating Characterization of Microstructure," Mater. Sci. Eng. A 323 1 -8 (2002). [2.37] M.Vippola, J.Vuorinen, P.Vuoristo, T.Lepisto, and T. Mantyla, "Thermal Analysis of  Plasma Sprayed Oxide Coatings Sealed With Aluminum Phosphate," J. Euro.Ceram.Soc. 22, 1987-1946 (2002) [2.38] M. Tsuhako, K. Hasegawa, T. Matsuo, I. Motooka, and M. Kobayashi, "Studies of the Preparation and Physical Properties of  Multivalent Metal Condensed Phosphates. Part Xi. The Preparation of  Various Aluminum Phosphates and Their Thermal Transformation,"  Bull.Chem.Soc. Japan 48[6] 1830-1835 (1975). [2.39] D.M. Roy, "New Strong Cement Materials. Chemically Bonded Ceramics," Science, 235 [4789] 651-658 (1987). [2.40] J.S. Reed," "11.9 Reaction Bonds" In Introduction  to the Principles  of  Ceramic Processing,  pp. 166-168, John Wiley & Sons, Inc., 1988 [2.41] R.W.Cahn, P.Haasen, and E.J.Kramer, "Materials Science and Technology," Processing of  Ceramics Part 1,17a, Pp.226-242 (1996) [2.42] B.C.Jeffrey,  and G.W. Scherer, "Chapter 4.1 Aqueous Metal Salts," Sol-gel Science, the Physics and Chemistry of  Sol-Gel Processing, 237-250, Academic Press, Boston, 1990 [2.43] PC. Hidber, T.J. Graule, and L.J. Gauckler, "Competitive Adsorption of  Citric Acid and Poly(vinyl alcohol) onto Alumina and its Influence  on the Binder Migration during Drying," J. Am. Ceram. Soc., 78[7] 1775-80 (1995) [2.44] P.C. Hidber, T.J. Graule, and Ludwig J. Gauckler, "Citric Acid-a Dispersant for Aqueous Alumina Suspensions," J. Am. Ceram. Soc., 79[7] 1857-67 (1996). [2.45] Y. Liu, L. Gao, and J. Guo, "Comparative Study on the Stabilizing Effect  of  2-Phosphonobutane-l,2,4-Tricarboxylic Acid and Citric Acid for  Alumina Suspensions," Colloid and Surfaces  A. Physicochemical and Engineering Aspects, 193, 187-195 (2000). [2.46] P. Tomaski , C.H. Schilling, R. Jankowiak, and J.C. Kim, "The Role of  Organic Dispersants in Aqueous Alumina Suspensions," J. Eur. Ceram. Soc., 23, 913-919, (2003). [2.47] CRC  Handbook  of  Chemistry  and  Physics, c251 -252, 59th edition, CRC press Inc., USA, 1978-1979. [2.48] L.C. Pathak, T.B. Singh, S.Das, A.K. Verma, and P. Ramachandrarao, "Effect  of pH on the Combustion Synthesis of  Nano-Crystalline Alumina Powder," Materials Letters 57, 380-385 (2002). [2.49] J. Cesarano III and I.A. Aksay, "Processing of  Highly Concentrated Aqueous a-Alumina Suspensions Stabilized with Polyelectrolytes," J. Am. Ceram. Soc., 71[12] 1062-67 (1988). [2.50] K.S. Chou and L.J. Lee, "Effect  of  Dispersant on the Rheological Properties and Slip Casting of  Concentrated Alumina Slurry," J. Am. Ceram. Soc., 72[9] 1622-27 (1989). [2.51] J.W. Fischer, L.H. Merwin, and R.A. Nissan, "NMR Investigation of  the Thermolysis of  Citric Acid," Appl. Spectroscopy, 49[1],120-126 (1995) [2.52] R.Kummert and W.Stumm,"The Surface  Complexation of  Organic Acids on Hydrous y-Al203," J. Colloid Interface  Sci., 75[2] 373-85 (1980). [2.53] W.Stumm, R.Kummert, and S.Sigg,"A ligand Exchange Model for  the Adsorption of  Inorganic and Organic Ligands at Hydrous Oxide Interfaces,"  Corat.Chem. Acta, 53,291-312(1980). , . ' • [2.54] Volume 10, Materials Characterization and Volume 17, Nondestructive Evaluation and Quality Control, ASM  Handbooks  Online, http://products.asminternational.org/hbk/index,jsp. available Dec. 16,2006. [2.55] Section 9: Molecular Structure and Spectroscopy, The  CRC  Handbook  of Chemistry  & Physics Online, http://www.hbcpnetbase.com, available Dec. 16, 2006. [2.56] D.C. Cranmer, "Overview of  Technical, Engineering, and Advanced Ceramics," Volume 4, Ceramics  and  Glasses, Engineered  Materials  Handbooks,  ASM International, 16-20. [2.57] M.K ferber  and V.J Tennery, "Structural Applications for  Technical, Engineering, and Advanced Ceramics," Volume 4, Ceramics  and  Glasses, Engineered  Materials Handbooks,  ASM International, 959-960. [2.58] R.C. Buchanan, "Electrical/Electronic Applications for  Advanced Ceramics," Volume 4, Ceramics  and  Glasses, Engineered  Materials  Handbooks,  ASM International, 1105-1106 [2.59] Masaru Miyayama et al., "Engineering Properties of  Single Oxides," Volume 4 , Ceramics  and  Glasses, Engineered  Materials  Handbooks,  ASM International, 748 -757. [2.60] R.W.Grimshaw, "Thermal Shock Resistance" in The  Chemistry  and  Physics of Clays,  Wiley-Inter  science, 4 th edition, 948-955, NY, USA, 1971 [2.61] David J. Green, An Introduction  to the Mechanical  Properties  of  Ceramics, Cambridge University Press, pp.301-303, 1998. [2.62] E.D. Kolb, A.M.Glass, R.L. Rosenberg, J.C. Grenier, and R.A. Laudise, "Dielectric and Piezoelectric Properties of  Aluminum Phosphate," IEEE Transactions on Sonics and Ultrasonics 29 [3]: 167-168 (1982). [2.63] R.D.Shannon, A.J. Vega, B.H.T. Chai, and G.R. Rossman, "Effect  of  H 2 0 on Dielectric Properties of  Berlinite:I. Dielectric Constant," J.Phys. D: Appl. Phys. 26,93-100 (1993). [2.64] L.A Garcia-Cerda et al. "Dielectric Properties of  Si02 Thin Films Prepared by the Sol-Gel Technique," J.Vac. Sci.Technol. B 18[1], Jan/Feb (2000). [2.65] H. Kim, S. Odoul, C. Lee, ad Y. Kweon, "The Electrical Insulation Behavior and Sealing Effects  of  Plasma-Sprayed Alumina-Titania Coatings," Surf.  Coat. Technol. 140, 293 (2001). [2.66] J. Molla, R. Moreno, and A. Ibarra, "Effect  of  Mg Doping on Dielectric Properties of  Alumina," J. Appl. Phys. 80[2], 1028 (1996). [2.67] T. Olding, M. Sayer, and D. Barrow, "Ceramic Sol-Gel Composite Coatings for Electrical Insulation," Thin Solid Films 398-399, 581 (2001). [2.68] J.M. Heintz, P.Letullier, J.L. Miane, and J. Ravez, "Dielectric Properties of  alumina Lamellar Ceramics," Mater. Sci. Eng. B 52, 84 (1998). [2.69] X.H. Wang and Y. Hirata, "Electrical and Rheological Properties of  Colloidal Alumina Suspensions," Ceram. Proc. Res. 1[1], 64 (2000). [2.70] P.Miranzo, L. Tabernero, J.S. Moya, and J.R. Jurado, "Effect  of  Sintering Atmosphere on the Densification  and Electrical Properties of  Alumina," J. Am. Ceram. Soc. 73[7], 2119 (1990). [2.71] K.S. Shamala, L.C.S. Murthy, and K.N. Rao, "Studies on Optical and Dielectric Properties of  AI2O3 Thin Films Prepared by Electron Beam Evaporation and Spray Pyrolysis Method," Mater. Soc. Eng. B 106, 269 (2004). [2.72] J.J. O'dwyer, The  Theory  of  Dielectric  Breakdown  of  Solids,  Oxford  at The Clarendon Press, 1964. [2.73] David J. Green, "2.3 Engineering Elastic Constants," pp. 18-24, An Introduction  to the Mechanical  Properties  of  Ceramics,  Cambridge University Press, 1998. [2.74] David J. Green, "9.8 Thermal Stresses and Thermal Shock" pp.298-3.1, An Introduction  to the Mechanical  Properties  of  Ceramics,  Cambridge University Press, 1998. [2.75] David J. Green, "2.9 Coefficient  of  Thermal Expansion," pp.36-40, An Introduction to the Mechanical  Properties  of  Ceramics,  Cambridge University Press, 1998. [2.76] David J. Green, "3.7 Thermal Expansion Behavior of  Polycrystalline Ceramics" pp.94-98, An Introduction  to the Mechanical  Properties  of  Ceramics,  Cambridge University Press, 1998. [2.77] David J. Green, "4.3 Bending of  Beams," pp.108-113, An Introduction  to the Mechanical  Properties  of  Ceramics,  Cambridge University Press, 1998 [2.78] John J. Aklonis and William J. Macknight, "2 Phenomenological Treatment of Viscosity, pp.7-35, Introduction  to Polymer Viscoelasticity,  2 n d edition, John Wiley&Sons, Inc., USA [2.79] John J. Aklonis and William J. Macknight, "7. Viscoelastic Models," pp. 139-156, Introduction  to Polymer Viscoelasticity,  2n d edition, John Wiley&Sons, Inc., USA [2.80] C.H. Hsueh and Mattison K. Ferber, "Apparent Coefficient  of  Thermal Expansion and Residual Stresses in Multilayer Capacitors," Composites part A. Applied Science and Manufacturing,  33, 1115-1121 (2002) [2.81] A.Portinha, V.Teixeira, J.Carneiro, J.Martins, M.F. Costa, R.Vassen, and D.Stoever, "Characterization of  Thermal Barrier Coatings with a Gradient in Porosity," Surf. Coat. Technol., 195,245-251 (2005). [2.82] V. Teixeira, "Numerical Analysis of  the Influence  of  Coatings Porosity and Substrate Elastic Properties on the Residual Stresses in High Temperature Graded Coatings," Surf.  Coat. Technol., 146-147, 79-84 (2001). [2.83] I.T.Spitsberg, D.R.Mumm, and A.G. Evans, "On the Failure Mechanisms of Thermal Barrier Coatings with Diffusion  Aluminide Bond Coatings," Mater. Sci. Eng. A, 394, 176-191 (2005). [2.84] D.Pan, M.W. Chen, PK. Wright, and K.J. Hemker, "Evolution of  a Diffusion Aluminide Bond Coat for  Thermal Barrier Coatings during Thermal cycling," Acta Materialia, 51, 2205-2217 (2003). [2.85] H.Kozuka, S.Takenaka, H.Tokita, T.Hirano, Y.Higashi, and T.Hamatani, "Stress and Cracks in Gel-Derived Ceramic Coatings and Thick Film Formation," J.Sol-Gel Sci.Techn. 26, 681-686 (2003). [2.86] J.Ilavsky and C.C. Berndt, "Thermal Expansion Properties of  Metallic and Cermet Coatings," Surf.  Coat. Technol., 102, 19-24, (1998). [2.87] O.Kesler, M. Finot, S.Suresh, and S.Sampath, "Determination of  Processing-Induced Stresses and Properties of  Layered and Graded Coatings. Experimental Method and Results for  Plasma-Sprayed Ni-Al203," Acta Mater. Vol.45 No.8, pp.3123-3134, 1997. [2.88] C.H. Hsueh and A.G. Evans, "Residual Stresses in Metal/Ceramic Bonded Strips," J. Am. Ceramic. Soc., 68 [5] 241-48 (1985). [2.89] I.C. Niyan and J.B. Cohen, "2. Fundamental Concepts in Stress Analysis," ppl3-46, Residual  Stress,  Measurement  by Diffraction  and  Interpretation,  MRE Materials Research and Engineering, Springer-Verlag New York Inc. 1987. [2.90] S. Sathish, T.J. Moran, R.W. Martin, and R. Reibel, "Residual Stress Measurement with Focused Acoustic Waves and Direct Comparison with X-ray Diffraction  Stress Measurement," Mater. Sci. Eng.A, in press, (2005). [2.91] E.Atar, C. Sarioglu, H.Cimenoglu, and E.S. Kayali, "Residual Stresses in (Zr,Hf)N Films (up to 11.9 at. % Hf)  Measured by X-ray Diffraction  using Experimentally Calculated XECs," Surf.  Coat. Technol., 191, 188-194 (2005). [2.92] C.Sariogle, Ugur Demirler, M.Kursat Kazmanli, and Mustafa  Urgen, "Measurement of  Residual Stresses by X-ray Diffraction  Techniques in MoN and Mo2N Coatings Deposited by Arc PVD on High-Speed Steel Substrate," Surf.  Coat. Technol., 190, 238-243, (2005). [2.93] O. Kesler, J.Matejicek, S.Sampath, S.Suresh, T.Gnaeupel-Herold, P.C. Brand, and H.J. Prask, "Measurement of  Residual Stress in Plasma-Sprayed Metallic Ceramic and Composite Coatings," Mater. Sci. Eng. A, 257,215-224 (1998). [2.94] T.C. Totemeier and J.K. Wright, "Residual Stress Determination in Thermally Sprayed Coatings-A Comparison of  Curvature Models and X-ray Technique," Surf. Coat. Technol., in press, 2005. [2.95] B.D. Cullity, "16 Measurement of  Residual Stress," pp.447-478, Elements  of  X-ray diffraction,  2 n d edition, Addison-wesley publishing Company, Inc. [2.96] J.Pina, A. Dias, and J.L. Lebrun, "Study by X-ray Diffraction  and Mechanical Analysis of  the Residual Stress Generation during Thermal Spraying," Mater. Sci. Eng. A, 347,21-31 (2003). [2.97] A. Portinha, V. Teixeira, J. Carneiro, M.G.. Beghi, C.E. Bottani, N. Franco, R.Vassen, D. Stoever, and A.D. Sequeira, "Residual Stresses and Elastic Modulus of  Thermal Barrier Coatings Graded in Porosity," Surf.  Coat. & Tech., 188-189, 120-128, (2004). [4.1] Q.Yang, Composite  Sol-Gel,  Ph.D. Thesis, Department of  Metals and Materials Engineering, The University of  British Columbia, May, 1999. [4.2] J.W. Fischer, L.H. Merwin, and R.A. Nissan, "NMR Investigation of  the Thermolysis of  Citric Acid," Appl. Spectroscopy, 49,1,120-126 (1995) [4.3] T.R. Thomas, "Chapter 4. Profile  Description," Rough Surfaces,  72-90, Longman Group Limited, U.K., 1982. [4.4] B.Cheng, S. Ramamurthy and N.S. Mclntyre, "Characterization of  Phosphate Films on Aluminum Surfaces,"  Journal of  Materials Engineering and Performance, 6 [4], 405-412 (1997). [4.5] User  Guide  902-50015, A Brochure of  Rheometire Scientific,  Inc., April 2000. [5.1] W.H. Gitzen, "Chap. 16 Sintering: Alumina as a Ceramic Material," Special Publication No.4, The American Ceramic Society, Inc., 127-134 (1970). [5.2] A.J Brown, J. Bultitude, J.M. Lawson, H.D. Winbow, and S. Witek, "Raw Materials," Ceramics  and  Glasses, Engineered  Materials  Handbooks,  ASM International, Volume 4, 43-51. [5.3] Jerry Weinstein, "Mineral Processing," Ceramics  and  Glasses, Engineered Materials  Handbooks,  ASM International, Volume 4, 961-965. [5.4] Rex W. Grimshaw, "Chapter 7: Physical and Structure Properties of  Raw Ceramic Materials," The  Chemistry  and  Physics of  Clays,  4 t h edition, 407-411, Wiley-Interscience Division, John Wiley and Sons Inc. U.S.A,, 1971. [5.5] Rex W. Grimshaw, "Chapter 7: Physical and Structure Properties of  Raw Ceramic Materials," The  Chemistry  and  Physics of  Clays,,  4 t h edition, 395-403, Wiley-Interscience Division, John Wiley and Sons Inc. U.S.A., 1971. [5.6] Relva C Buchanan, Ceramic  Materials  for  Electronics:  Processing,  Properties,  and Applications,  Marcel Dekker, Inc., 2-77, 1986. [5.7] R.W.Cahn, P.Haasen, and E.J.Kramer, "Section 7.3: Materials for  Casting Ceramic Tapes in Materials Science and Technology," Processing of  Ceramics  part I, Volume 17A "pp.207-216,1996. [5.8] Lawrench H. Van Vlack, Elements  of  Materials  Science: An Introductory  Text  for Engineering  Students,  Addison-wesley publishing Company, Inc., Massachusetts, USA, 1959. [5.9] James S. Reed, "Particle Packing Characteristics," Introduction  to the Principles  of Ceramic  Processing,  185-199, A Wiley & Sons, Inc., USA, 1988 [5.10] James S. Reed, "8. Density, Pore Structure, and Specific  Surface  Area" in Introduction  to the Principles  of  Ceramic  Processing,  John Wiley& sons, Inc., USA, pp.105-119,1988 [5.11] X. Krokidis, P. Raybaud, A.E. Gobichon, B. Rebours, P. Euzen, and H. Toulhoat, "Theoretical Study of  the Dehydration Process of  Boehmite to y-Alumina," J.Phys. Chem.B 105, 5121-5130 (2001). [5.12] John W. halloran, " Calcination," Volume 4, Ceramics  and  Glasses, Engineered Materials  Handbooks,  ASM International, 109-114. [5.13] Robert D. Shoup, "Sol-Gel Processes," Volume 4, Ceramics  and  Glasses, Engineered  Materials  Handbooks,  ASM International, 445-452 [5.14] R. Griffiths  and C. Radfore,  "Chapter 3:Calculations in Ceramics," Chapter 4: Density and Specific  Gravity," Porous Solid,  Livesey Ltd., Shrewsbury, England, 28-42, 1965. [5.15] R.W.Cahn, P.Haasen, and E.J.Kramer, "Chapter 2 Process Control in the Manufacture  of  Ceramics: 2.3.1 Grain Size and Specific  Surface,"  Materials Science and  Technology,  Volume 17A, Germany, 36-38, 1996. [5.16] P.C. Hidber, T.J. Graule, and L.J. Gauckler, "Competitive Adsorption of  Citric Acid and Poly(vinyl Alcohol) onto Alumina and its Influence  on the Binder Migration during Drying," J. Am. Ceram. Soc., 78[7] 1775-80 (1995) [5.17] P.C. Hidber, T.J. Graule, and L.J. Gauckler, "Citric Acid-a Dispersant for  Aqueous Alumina Suspensions," J. Am. Ceram. Soc., 79[7] 1857-67 (1996). [5.18] J. Cesarano III and I.A. Aksay, "Stability of  Aqueous 01-AI2O3 Suspensions with Poly(methacrylic acid) Polyelectrolyte," J. Am Ceram. Soc., 71[4], 250-255 (1988). [5.19] J.A. Lewis, "Colloidal Processing of  Ceramics," J. Am Ceram. Soc., 83[10], 2341-2359 (2000). [5.20] Y. Liu and L. Gao, "Dispersion of  Aqueous Alumina Suspensions using Copolymers with Synergistic Functional Groups," Materials Chemistry and Physics, 82, 362-369, 2003. [5.21] Rex W. Grimshaw, "Chapter 8. Influence  of  Water in Ceramic System," The Chemistry  and  Physics of  Clays,  4 th edition, 483-491, Wiley- Interscience Division, John Wiley and Sons Inc. U.S.A,, 1971. [5.22] Rex W. Grimshaw, "Chapter 7. Physical and Structure Properties of  Raw Ceramic Materials," The  Chemistry  and  Physics of  Clays,  4 t h edition, 395-403, Wiley-Interscience Division, John Wiley and Sons Inc. U.S.A,, 1971. [5.23] R.W.Cahn, P.Haasen, E.J.Kramer, "7.3. Assessment of  properties," Materials Science and  Technology,  Processing of  Ceramics  part I,  Edited, Volume 17A,238-242, VCH, 1996 [5.24] David J. Green, An Introduction  to the Mechanical  Properties  of  Ceramics, Cambridge University Press, pp. 146-148, 1998. [5.25] Brinker, C.Jeffrey,  and George W.Scherer, "Chapter 5.1 Phenomenology," Sol-Gel Science: the Physics and  Chemistry  of  Sol-Gel  Processing,  304-314, Academic Press, Boston, 1990. [5.26] Brinker, C.Jeffrey,  and George W.Scherer, "Chapter 3.2 Hydorlysis and Condensation of  Silicon Alkoxides," Sol-gel  Science: the Physics and  Chemistry  of Sol-Gel  Processing,  203-212, Academic Press, Boston, 1990 [5.27] R.W.Cahn, P.Haasen, E.J.Kramer, "7.3.2 Slurries for  Tape Casting," in Materials Science and  Technology,  Processing of  Ceramics  part I,  Edited, Volume 17A, 216-238, VCH, 1996. [5.28] Brinker, C.Jeffrey,  and George W.Scherer, "Chapter 4.1 Aqueous Metal Salts," Sol-Gel Science: the Physics and  Chemistry  of  Sol-Gel  Processing,  237-250, Academic Press, Boston, 1990. [5.29] W.C Ormsby and J.H. Marcus, "Flow Properties of  Aqueous Suspensions Containing Kaolins of  Varying Degrees of  Crystallinity," J. Am Ceram. Soc., 50, 190-195, 1967. [6.1] James S. Reed, "2. Surface  Chemistry" in Introduction  to the Principles  of Ceramic  Processing,  Wiley-Interscience Publication, John Wiley & Sons, Inc., USA, pp.17- 26,1988. [6.2] C. Peng, S.B. Wen, and T.T. Lee, "Preparation of  Nanometer-Sized a- Alumina Powders by Calcining an Emulsion of  Boehmite and Oleic Acid," J.Am.Ceram.Soc.,85[l], 129-33 (2002). [6.3] C. Peng and S.B. Wen, "Variations in a Boehmite Gel and Oleic Acid Emulsion under Calcination," J.Am.Ceram.Soc., 85[6] 1467-72 (2002). [6.4] C.H. Hsueh, and A.G. Evans, "Residual Stresses in Metals/Ceramic Bonded Strips," J.Am.Ceram. Soc., 68[5], 241-248 (1985). [6.5] David J. Green, "4.3 Bending of  Beams," An Introduction  to the Mechanical Properties  of  Ceramics,  108-113, 1998, Cambridge University Press. [6.6] David J. Green, "3.7 Thermal Expansion Behavior of  Polycrystalline Ceramics, 9.9 Thermal Shock Resistance Parameters," An Introduction  to the Mechanical Properties  of  Ceramics,  94-98, 301-305, Cambridge University Press, 1998. [6.7] B.D. Cullity, "9.4 Crystal perfection,"  Elements  ofX-Ray  Diffraction,  pp.263-269, Third printing, Addison-Wesley Publishing company Inc., 1967." [6.8] J.J. O'dwyer, The  Theory  of  Dielectric  Breakdown  of  Solids,  Oxford  at the Clarendon Press, 1964. APPENDIX Fig.A-1 Effect  of  the Sand-blasting on the shape of  AA5052 Veeco X Profile Rq 8.49 um Sa- ... B ® um il 9 97 um Rv •»,85 um .•u'cii 0.01 i q i.-*M,w>,., Ka »,- ,im Rt WM um_ Jt« -4.11 um . 11 M mm mk > us .a ' - Curve 0.00  nm Dist ......................... :...SB Mgle - ° ; AvttHt <S 76 uin Title: 30.150753 4 7 Ann O.Mmua Note: 30.150753 3-Dimensional Interactive Display Date: 4/26/2005 Time: 11:36:46 Surface  Stats: Ra: 10.92 um Rq: 13.60 um Rt: 58.71 um Measurement Info: Magnification:  5.00 Measurement Mode: VSI Sampling: 99.50 um Array Size: 401 X 3 Title: 30.150753 Note: 30.150753 18.7 15.0 - 10.0 5.0 - 0.0 - -5.0 10.0 15.0 20.0 25.0 30.0 35.0 -40.0 (a) As-received AA5052 after  cutting in a beam shape Fig.A-1 Effect  of  the Sand-blasting on the shape of  AA5052 (Cont') Veeco X Profile 20 2S » 3S M Title: 30 .150753 Note: 30 .150753 Y Profile X 11.04 * 3.00 HI . a s i - • tun - mm jSgK . » *:* —j . . . . . . . Area -0.11 3-Dimensional Interactive Display Date: 4/26/2005 Time: 11:45:39 Surface  Stats: Ra: 34.31 um Rq: 40.43 um Rt: 162.76 um Measurement Info: Magnification:  5.00 Measurement Mode: VSI Sampling: 99.50 um Array Size: 401 X 3 Title: 30.150753 Note: 30.150753 (b) After  sand-blasting with silica grit (size of  60) Fig.A-2 Comparison of  Stress/Strain Curve between As-received AA5052 and Alumina coated AA5052 (-100 p,m coating thickness) with/without CA Strain,% Fig.A-3 Surface  images of  alumina coated AA5052 alloy (a) 1st layer (b) 2 n d layer • H (c) 3 rd layer (d) 4 th layer (e) 5 th layer Fig.A-4 Comparison of  XRD spectra of  as-received alumina powders with phosphated alumina CB-CSG under the stress-free  state w c o •+ •> c The peaks in (a) are for  as-received fine  alumina powders (0.5 jam), the peaks in (b) are for as received coarse alumina powders (5 jam), and the peaks in (c) are for  the phosphated alumina CB-CSG (mixed with 40/60 of  fine/coarse  particles) under the stress free  state Fig.A-5 Comparison of  the effect  of  chemical bonding and CAD+sol on alumina coatings on AA5052 (a) Alumina CSG coating on AA5052 without chemical bonding by using phosphoric acid. The complete coating failure  (i.e., delamination) occurred at the thickness of  the coating about 100 jam (b) CA-free  alumina CB-CSG coatings on AA5052. The partial coating failure (i.e., delamination) occurred at the thickness of  about 200 |im (c) CA-containing alumina CB-CSG coatings on AA5052. The coating thickness is about 200 |im and no coating failure  (i.e., delamination) occurred during the process. Table A-l Parameters for  obtaining average Young's Modulus of  alumina coated AA5052 using 3-Point Bending Strain Stress Young's #of Thickness, Width, constant, Strain, constant, modulus, layer mm mm K8 s K* E (Gpa) substrate 1.62 6.4 0.006075 0.000304 35258.06 66.7 1 1.67 6.45 0.006263 0.000313 32921.21 63.52 1 1.674 6.45 0.006278 0.000314 32764.07 . 64.16 2 1.704 6.43 0.00639 0.00032 31718.91 61.24 2 1.704 6.43 0.00639 0.00032 31718.91 60.87 3 1.771 6.5 0.006641 0.000332 29048.11 61.15 3 1.771 6.51 0.006641 0.000332 29003.49 61.12 4 1.809 6.56 0.006784 0.000339 27585.92 56.64 4 1.781 6.5 0.006679 0.000334 28722.83 55.87 5 1.883 6.57 0.007061 0.000353 25421.57 52.03 5 1.868 6.57 0.007005 0.00035 25831.48 52.12 Table A-2 Specifications  of  Dynamic Mechanical Analyzer (Q800, TA Instruments, USA) Maximum Force 18N Minimum Force 0.0001N Force Resolution 0.00001N Strain Resolution 1 nm Modulus Range 10j to 3 X 10 u Pa Modulus Precision ±1% Tan5 Sensitivity 0.0001 Tan8 Resolution 0.00001 Frequency Range 0.01 to 200 Hz Dynamic Sample Deformation  Range ±0.5to 10,000 [im Temperature Range -150 to 600 °C Heating Range 0.1 to 20 °C/min Cooling Rate 0.1 to 10 °C/min Isothermal Stability +0.1 °C Time/Temperature Superposition Yes 13 Table A-3 C-NMR Absorptions of  major functional  groups [2.55] 5 (ppm) Group Family Example (5 of  italicized carbon) 220-165 >C=0 (CH3)2CO (206.0) (CH3)2CHCOCH3 (212.1) Aldehydes CH3CHO (199.7) a,(3-Unsaturated CH3CH=CHCHO (192.4) carbonyls CH2=CHCOCH3 (169.9) Carboxylic acids HCO2H (166.0) CH3CO2H (178.1) Amides HCONH2 (165.0) CH3CONH2 (172.7) Esters CH3CO2CH2CH3 (170.3) CH2=CHCO2CH3 (165.5) 140-120 >C=C< Aromatic C^ HG (128.5) Alkenes CH2=CH2 (123.2) CH2=CHCH3 (115.9,136.2) CH2=CHCH2CI (117.5, 133.7) CH3CH=CHCH2CH3 (132.7) 125-115 -CN Nitriles CH3-CN (117.7) 80-70 -cc- Alkynes HCCH (71.9) CH3CCH3 (73.9) 70-45 -C-0 Esters CH3OOCH2CH3 (57.6, 67.9) Alcohols HOCH3 (49.0) HOCH2CH3 (57.0) 40-20 -C-NH2 Amines CH3NH2 (26.9) CH3CH2NH2 (35.9) 30-15 -S-CH3 Sulfides  (thioethers) C6H5-S-CH3 (15.6) 30-(-2.3) -C-H Alkanes, cycloalkanes CH4 (-2.3) CH3CH3 (5.7) CH3CH2CH3 (15.8, 16.3) CH3CH2CH2CH3 (13.4, 25.2) CH3CH2CH2CH2CH3 (13.9, 22.8,34.7) Cyclohexane (26.9) Table A-4 Full Data set for  Dielectric Breakdown Voltages of  the Alumina Coated AA5052 Thickness (mm) Breakdown Voltage (V) Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 Point 7 40 570 570 572 617 619 668 680 80 812 819 918 918 967 1118 1118 120 1088 1116 1216 1317 1366 1466 1504 160 1440 1468 1517 1665 1865 2015 2166 200 1744 1767 1916 2066 2365 2516 2674 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0078497/manifest

Comment

Related Items