Bond Coat SAT PhD Thesis

UNIVERSITY OF CAMBRIDGE Department of Materials

Science and Metallurgy

Thermophysical Properties of Plasma Sprayed Thermal Barrier

Coatings

Sofia A. Tsipas

St. John’s College

June 2005

Dissertation Submitted for the Degree of Doctor of Philosophy

-i-

Preface

This dissertation is submitted for the degree of Doctor of Philosophy at the University of

Cambridge. The work described in this document was carried out between September 2001 and

June 2005 under the supervision of Prof. T.W. Clyne in the Department of Materials Science and

Metallurgy at the University of Cambridge. This dissertation is the result of my own original work

and includes nothing which is the outcome of work done in collaboration except where specifically

indicated in the text and has not been submitted, either in part or in its entity, for a degree at any

other university. This dissertation is less 60,000 words in length.

Sofia Tsipas

St John’s College

Cambridge

June 2005

-ii-

Abstract Since advanced aerospace and power generation gas turbine engines are now designed such that nickel

superalloy components operate at temperatures very close to their melting points, current strategies for performance

improvements are centred on thermal barrier coatings (TBCs). State-of-the-art TBC systems are composed of a

zirconia ceramic top coat (300-600m thick), deposited either by air plasma spraying (APS) or electron beam

assisted physical vapour deposition (EB-PVD) over a metallic bond coat (100m thick) deposited by vacuum

plasma spraying (VPS). The top coat acts as a thermal barrier, generating a temperature drop of up to ~300˚C,

while the bond coat provides corrosion and oxidation protection for the substrate and creates a rough surface which

may promote adhesion of the top coat. Failure of TBCs usually occurs by buckling and spalling of the coating.

APS top coats, which are considerably cheaper to produce, are composed of overlapping splats oriented parallel to

the plane of the coating, with poor inter-splat bonding, many fine microcracks and pores. Therefore, they have very

low stiffness. This prevents large stresses, and hence large driving forces for spallation, from being generated in

the top coat. There is current interest in pushing the top coat surface temperature up to about 1300-1400˚C in

aeroengines, while the aim for power generation is around 1200˚C. During the course of this work several aspects

of the behaviour of plasma sprayed coatings were investigated.

The effect of composition, impurity level, microstructural anisotropy and thermal cycling on the sintering

behaviour was investigated. Sintering involves healing of microcracks and strengthening of inter-splat bonding

during prolonged heat treatment. Healing of microcracks and improved inter-splat bonding due to sintering raise

the coating stiffness. This stiffening, will raise the strain energy release rate associated with a given misfit strain

and hence make spallation more likely.

TBCs operate in service under high thermal gradients. Therefore, it is crucially important to know how

such conditions will affect the top coat properties. The sintering characteristics, stress distribution and phase

constitution in the top coat were found to be strongly dependant on the presence of a through-thickness thermal

gradient.

The effect of heat treatment on the thermal conductivity of the top coat was also investigated. A key

objective for TBCs is to maximise the temperature drop across the thickness of the top coat. This requires in turn

that the thermal conductivity of the top coat should be minimised and remain low during exposure to service

conditions. Healing of microcracks and improved inter-splat bonding, due to sintering in service conditions, will

increase the thermal conductivity. A novel steady state technique was used to monitor the changes in thermal

conductivity with heat treatment. The factors that play the dominant roles in reducing the thermal conductivity of

top coat were identified. The importance of thermal conduction through the pores in the top coat was highlighted.

The use of an analytical heat flow model showed that it should be possible to predict the conductivity from

measurable microstructural parameters.

There is an increasing interest in developing new top coat materials that will be more sinter resistant and

have lower thermal conductivity. Novel material compositions have been investigated with respect to their sintering

behaviour, microstructural and phase evolution. This work will make it possible to pinpoint to the key issues for

better top coat performance and give new insight for the development of new materials for their use in TBCs.

-iii-

Acknowledgements

I gratefully acknowledge the provision of funding for the research described in this dissertation by

Sulzer Inc., Cambridge European Trust and Isaac Newton Awards. Thanks to Prof. D.J. Fray for

the provison of office, laboratory and workshop facilities at the Department of Materials Science

and Metallurgy, University of Cambridge.

There are many people who have helped me in the production of this work, but I am most indebted

to my supervisor, Prof. Bill Clyne, from whom I have learned so much. I want to express my

sincere gratitude and appreciation for his guidance and expert advice over the past three years. His

commitment and great knowledge have been an inspiration to me throughout my PhD. I would also

like to thank him for making the lab such a peasant place to work.

Thanks are also due to Dr. Jason Doesburg, Keith Harrison, Clive Britton and Andrew Nicoll from

Sulzer Inc. for providing the spraying powders, and for many useful discussions as well as their

encouragement regarding this project.

Many thanks also go to the technical staff at the Department of Materials Science and Metallurgy

for their assistance, especially Paul Stawkes and Mike Brand in the workshop, for being the people I

could rely on and for always being so happy to see me. Special thanks must go to Kevin Roberts,

for his technical and practical expertise, which has been of great value to this work, as well as for

his good company. I must also thank Terry Mosdal, for his help with spraying, Robert Stearn for

his help with experimental set ups and Mary Vickers for her help with X-ray diffraction.

It is the people that are the soul of the Gordon Laboratory, and I would like to thank them all for

making it such a pleasant place to work (and for putting up with me in general…): Thank you Dr.

Igor Golosnoy for overlooking my work, for the “prroper” serious science, for the vodka nights, for

hanging up and worrying about my well-being in general; Athina, thank you for always being there

for me and for your invaluable friendship; Katerina, thank you for being so sweet, for your

friendship, for all your ideas that made so many days in the lab unforgettable and for the Greek

flags; Jin for being unfortunate enough to sit next to me for almost three years, for all your help with

technical issues and learning to dance Greek dances so well, Debu, the guru, thank you for all your

stories, for cooking proper curry and demonstrating how to take care of drunk people; Shiladitya

(who’s your daddy?) for finally resigning to the fact of being a “lost case” and for finding the best

way to tolerate Greek noise: use ear protectors (keep up the David Hasselhoff look!); Amaia-

empanada, for being the best person to carry on and take my place and desk (keep me updated on

-iv-

the Cambridge gossip) and for sharing the same addiction with me: S&C; James C (go home). for

his unparallel ability to give me useless presents and for being so charitable (it wouldn’t have

worked anyway); Andrew for reminding me that whatever I do it is meant to go wrong and for his

lack of fashion sense; James D for being the man; Martin for being the other person, other than me,

working late in the lab; Chris for the healthy food option on Fridays; TomI for being so “cool” and

Jamie for his “Greeks, Geeks, Freaks and…Brits” comment. I wish you all the best of luck. Also

some past members deserve a mention, Russell-Barri for the samosas and pub-quiz expertise, Katy

for the regular lab pub visits, Dave for all the plants and Alex, yeah yeah yeah, who spent a shorter

time in the lab but became a big part of it.

A thankyou is also due to many of my friends in Cambridge, Renne, Bernard, Aldo, Milja, Ivana,

Petchya, Tony, Napoleon, Maro, George, Marianna and Kostas for their much-valued friendship,

for making my time here so enjoyable and keeping me sane. A special thank you to Carlos for his

never-failing support, encouragement, understanding and patience; I couldn’t have finished this

without you.

Last, but above all, I would like to thank my loving family, Mami, Papa, and Marili, for their

constant unconditional support, which has been incredibly important to me, and for their love and

encouragement throughout my whole time in Cambridge. It is to my father, Prof. Dimitris Tsipas,

that I wish to dedicate this thesis.

-v-

TABLE OF CONTENTS

Preface .............................................................................................................................................. i

Abstract ............................................................................................................................................ ii

Acknowledgements .......................................................................................................................... iii

TABLE OF CONTENTS ................................................................................................................. v

Nomenclature ................................................................................................................................... x

1 Introduction ................................................................................................................................ 1

2 Thermal Barrier Coatings for Gas Turbine Applications ..................................................... 3

2.1 The gas turbine engine ..................................................................................................... 3

2.1.1 Materials for Gas Turbines .......................................................................................... 4

2.2 Thermal Barrier Coating Systems .................................................................................... 6

2.2.1 Nickel Based Superalloys ............................................................................................ 6

2.2.2 Bond Coat .................................................................................................................... 9

2.2.3 Top Coat ..................................................................................................................... 10

2.3 Coating Systems ............................................................................................................. 11

2.3.1 EB-PVD Coatings ...................................................................................................... 11

2.3.2 Plasma Sprayed Coatings ........................................................................................... 11

2.4 The Plasma spraying Process ......................................................................................... 13

2.4.1 The Plasma Jet ........................................................................................................... 13

2.4.2 Coatings Microstructure and Properties ..................................................................... 14

2.4.3 Parameters Affecting the Process .............................................................................. 16

3 Review of Thermomechanical Behaviour of Thermal Barrier Coatings ............................ 18

3.1 Stresses in Plasma Sprayed Coatings ............................................................................. 18

3.1.1 Mechanisms of Stress Generation .............................................................................. 18

3.1.1.1 Quenching Stress ............................................................................................................................. 18

3.1.1.2 Differential Thermal Contraction. ................................................................................................... 20

3.1.1.3 Solid State Transformations ............................................................................................................ 21

3.1.2 Measurement of Residual Stresses ............................................................................. 21

3.1.2.1 Curvature Methods .......................................................................................................................... 21

-vi-

3.1.2.2 Diffraction Methods ........................................................................................................................ 23

3.1.3 Modelling of Residual Stresses .................................................................................. 24

3.2 Mechanical Properties of TBCs ..................................................................................... 28

3.2.1 Mechanical Behaviour of PS TBCs ........................................................................... 28

3.2.2 Prediction of Mechanical Properties of PS Materials ................................................ 32

3.3 The Thermally Grown Oxide ......................................................................................... 34

3.4 Creep of Bond Coat and Thermally Grown Oxide ........................................................ 35

3.4.1 Creep in the Bond coat ............................................................................................... 36

3.4.2 Creep of the TGO ....................................................................................................... 36

3.5 Sintering of the Top Coat ............................................................................................... 37

3.5.1 Sintering mechanisms in ceramics ............................................................................. 37

3.5.1.1 Solid State Sintering ........................................................................................................................ 37

3.5.1.2 Liquid Phase Sintering .................................................................................................................... 42

3.5.2 Sintering of Zirconia based top coats ......................................................................... 43

3.6 Phase changes in Top Coat ............................................................................................ 45

4 Review of Thermal Conduction in TBCs ............................................................................... 49

4.1 Thermal Conduction in Ceramics .................................................................................. 49

4.1.1 Phonon Conduction in Ceramics................................................................................ 49

4.1.2 Radiative Heat Transfer in Ceramics ......................................................................... 51

4.2 Methods for Measuring the Thermal Conductivity of Solids ........................................ 52

4.2.1 Steady State Methods ................................................................................................. 52

4.2.2 Dynamic Methods ...................................................................................................... 54

4.2.2.1 Transient Hot Strip .......................................................................................................................... 55

4.2.2.2 Transient Plane Strip (Hot Disk) ..................................................................................................... 56

4.2.2.3 Laser Flash Method ......................................................................................................................... 58

4.3 Thermal Conductivity of TBCs ...................................................................................... 61

4.4 Modelling of Heat Flow in Plasma-Sprayed TBCs ....................................................... 65

4.4.1 Morphological representation of PS TBCs. ............................................................... 66

4.4.2 Physical Assumptions ................................................................................................ 67

4.4.3 Numerical Model ....................................................................................................... 69

-vii-

4.4.4 Two Flux Region Analytical Model .......................................................................... 70

4.4.5 Predicted Conductivity ............................................................................................... 73

5 Experimental Procedures ........................................................................................................ 74

5.1 Characterization of Starting Materials ........................................................................... 74

5.1.1 Chemical Composition ............................................................................................... 74

5.1.2 Particle Size Distribution ........................................................................................... 74

5.1.3 Powder Morphology .................................................................................................. 76

5.1.4 Phase Constitution of Starting Powders ..................................................................... 78

5.2 Coating Production ........................................................................................................ 79

5.2.1 Substrate Preparation ................................................................................................. 79

5.2.2 Plasma Spraying ......................................................................................................... 80

5.3 Sample Preparation Procedure ....................................................................................... 82

5.3.1 Detaching the Top Coat ............................................................................................. 82

5.3.2 Metallographic preparation ........................................................................................ 83

5.4 Coating Characterisation ................................................................................................ 84

5.4.1 Surface roughness ...................................................................................................... 84

5.4.2 Scanning Electron Microscopy .................................................................................. 84

5.4.3 Dilatometry ................................................................................................................ 84

5.4.4 X-Ray Diffraction ...................................................................................................... 85

5.4.5 Porosity Measurements .............................................................................................. 87

5.4.6 Stiffness Measurement ............................................................................................... 88

5.4.6.1 Cantilever Bending .......................................................................................................................... 88

5.4.6.2 Nanoindentation .............................................................................................................................. 89

5.5 Heat Treatment of Thermal Barrier Coatings ................................................................ 90

5.5.1 Isothermal Heat Treatment......................................................................................... 90

5.5.2 Heat Treatment with a High Thermal Gradient ......................................................... 90

5.6 Measurement of Thermal conductivity .......................................................................... 91

5.6.1 Steady State Rig ......................................................................................................... 91

5.6.1.1 Experimental Set Up ....................................................................................................................... 91

5.6.1.2 Data Analysis .................................................................................................................................. 94

-viii-

5.6.1.3 Validation of Technique .................................................................................................................. 95

5.6.2 Hot Disk ..................................................................................................................... 97

5.6.3 Laser Flash ................................................................................................................. 99

6 Thermal Stability of Plasma-Sprayed Top Coats ............................................................... 101

6.1 Changes in Phase Constitution of Plasma-Sprayed Top Coats due to Heat Treatment

101

6.1.1 As-sprayed ............................................................................................................... 101

6.1.1.1 Yttria stabilized zirconia top coats ................................................................................................ 101

6.1.1.2 Dysprosia stabilized zirconia top coats ......................................................................................... 104

6.1.1.3 Yttria-lanthana stabilized zirconia top coats.................................................................................. 105

6.1.1.4 Ceria stabilized zirconia top coats ................................................................................................. 106

6.1.2 Isothermal Heat Treatment....................................................................................... 108

6.1.2.1 Yttria stabilized zirconia top coats ................................................................................................ 108

6.1.2.2 Dysprosia stabilized zirconia top coats ......................................................................................... 113

6.1.2.3 Yttria-lanthana stabilized zirconia top coats.................................................................................. 114

6.1.3 Heat Treatment Under a Thermal Gradient ............................................................. 115

6.2 Microstructural development under service conditions ............................................... 117

6.2.1 Isothermal Heat Treatment....................................................................................... 117

6.2.2 Heat Treatment Under a Thermal Gradient ............................................................. 120

6.3 Effect of Heat Treatment of Detached Top Coats on sintering behaviour ................... 122

6.3.1 Effect of Composition and Morphology on Sintering Characteristics ..................... 122

6.3.2 Effect of Phase Transformation on Measured Volume Changes ............................. 129

6.3.3 Effect of Thermal Cycling on Sintering Behaviour ................................................. 131

6.3.4 Effect of Sintering on Coating Porosity ................................................................... 132

6.4 Conclusions .................................................................................................................. 135

7 Thermomechanical Behaviour of Plasma Sprayed TBCs .................................................. 137

7.1 Effect of Top Coat Sintering on Mechanical Properties .............................................. 137

7.1.1 Cantilever Bending .................................................................................................. 137

7.1.2 Nanoindentation ....................................................................................................... 138

7.1.2.1 As-sprayed and After Isothermal Heat Treatment ......................................................................... 138

7.1.2.2 Heat Treatment Under a Thermal gradient .................................................................................... 139

-ix-

7.2 Microstructure and Properties of Thick As-sprayed TBCs .......................................... 139

7.2.1 The Effect of Substrate Temperature on Microstructure and Stiffness ................... 140

7.2.2 The Effect of Substrate Temperature on Residual Stresses ..................................... 145

7.3 Stresses in TBCs after heat treatment .......................................................................... 145

7.4 Conclusions .................................................................................................................. 147

8 Thermal Conduction in Plasma-Sprayed TBCs .................................................................. 148

8.1 Thermal Conductivity of Plasma sprayed TBCs .......................................................... 148

8.1.1 Steady state Rig ........................................................................................................ 148

8.1.2 Laser Flash Measurements ....................................................................................... 150

8.1.3 Hot Disk ................................................................................................................... 152

8.2 Effect of Powder Composition on Thermal Conduction ............................................. 153

8.3 The Effect of Heat Treatment on the Thermal conductivity of TBCs ......................... 155

8.4 Effect of Pore Conductivity on Overall Thermal Conductivity ................................... 157

8.5 Experimental and Predicted Thermal Conductivity ..................................................... 158

8.6 Conclusions .................................................................................................................. 160

9 Conclusions and Future Work .............................................................................................. 162

9.1 General Conclusions .................................................................................................... 162

9.2 Future work .................................................................................................................. 164

References ....................................................................................................................................... 166

x

Nomenclature

Symbol Units Parameter

(i) Roman Symbols

A m2 cross-sectional area

a m lattice parameter

B Pa m K-1

a constant, with a value of 2.5 10-5

for air

c m lattice parameter

C mol concentration

Cp J kg-1

K-1

specific heat capacity at constant pressure

Cv J kg-1

K-1

specific heat at constant volume

D m2

s-1

diffusion coefficient

d m inter-bridge (or interpore) distance.

dv m the thickness of the pore

E Pa Young modulus.

h W m-2

K-1

interfacial thermal conductance,

h m thickness (subscripts s and d refer to the thickness of the substrate and the

deposit respectively).

I m4 second moment of area.

J kg s-1

mass flux

kb J K-1

Boltzmann’s constant, 1.38 × 10-23

J K-1

k W m-1

K-1

thermal conductivity

kr W m-1

K-1

thermal conductivity by photon conduction

kph W m-1

K-1

thermal conductivity from phonon conduction

kt W m-1

K-1

total thermal conductivity

k0 W m-1

K-1

total conductivity of the lamellae

kp W m-1

K-1

thermal conductivity of the pore

kg0 W m

-1 K

-1 normal conductivity of the gas

kres W m-1

K-1

effective thermal conductivity of the “contact resistance” region in TFR

model

k1, k2 W m-1

K-1

thermal conductivities of the corresponding regions in TFR model

keff W m-1

K-1

effective thermal conductivity,

ktrue W m-1

K-1

actual thermal conductivity,

L m length of the specimen.

Lo m initial sample length

Lh m thickness of the unit cell in numerical and TFR model

Lv m height of unit cell in numerical and TFR model

lp m mean free path for scattering of phonons

lpd m mean free path for scattering of phonons by point defects

lr m mean free path for photon scattering

xi

m kg mass (subscripts air, liq and laq refer to measurements made in air,

measurements made when immersed in a liquid and the sample coated with

lacquer; subscripts g and total refer to the mass of the glass container, and the

total mass of the powder and the glass container respectively; superscripts uc

and c refer to uncoated and coated sample respectively)

M N m bending moment.

n - refractive index

n - integer.

P atm pressure

P N force (describing edge loading or applied loading)

p % porosity

pf - porosity fraction (by volume)

q W input/output power

Q W m-2

heat flux (subscripts upper and lower refer to the heat flux in the upper and

lower substrate respectively)

r m neck radius of sintering particles

R ohms resistance (subscript 0 refers to resistance at time t = 0)

R m radius of sintering particles

S m2

area of the “contact” region in TFR model

Sbr m2 cross section area of the bridge.

Stot m2 the total cross section area of the unit cell

T K or °C temperature (subscripts s and m refer to the substrate temperature and the

melting temperature respectively).

t s time (subscript max refers to total time of transient experiment)

V m3 volume

x m distance

y m displacement, relative to the interface, through the thickness of the beam

(ii) Greek Symbols

α m2 s

-1 thermal diffusivity

K-1

coefficient of thermal expansion.

β K-1

temperature coefficient for the electrical resistivity

γ m-1

scattering coefficient

Δ m probing depth (in Hot Disk technique)

Δε - linear misfit strain.

ΔL m the length change of a material

m-1 change in curvature.

T K temperature drop/difference (subscripts c and i refer to temperature drop

across the coating and interfaces respectively)

x m sample thickness/ length

- average axial component of residual strain (subscripts s and d refer those of

the substrate and the deposit respectively).

xii

m-1 curvature.

κ m-1

absorption coefficient

λ m-1

extinction coefficient

μ m4 J

-1s

-1 grain boundary mobility

ν m s-1

the mean phonon velocity

- Poisson ratio.

kg m-3 density (subscripts d and th refer to the experimentally determined density of

the deposit material and the corresponding theoretical (bulk) density

respectively).

W m-2

K-4 Stephen-Boltzmann’s constant, 5.668610

-6

Pa stress (subscripts s and d refer to the residual stresses in the substrate and

deposit respectively).

0 Pa quenching stress predicted by linear elastic behaviour.

q Pa quenching stress.

y Pa yield stress.

ω s-1

frequency of phonon lattice waves

Ω m3 volume per lattice site

(iii) Acronyms and Abbreviations

APS atmospheric plasma spraying

BC bond coat

CSZ ceria stabilized zirconia

CTE coefficient of thermal expansion

DC direct current

DS directional solidification

DSZ dysprosia stabilized zirconia

EB-PVD electron-beam physical vapour deposition

EDX energy dispersive x-rays

LPPS low pressure plasma spraying

PS plasma sprayed

PSZ partially stabilized zirconia

SANS small angle neutron scattering

SEM scanning electron microscope

TBC thermal barrier coating

TC top coat

TET turbine entry temperature

TGO thermally grown oxide

VPS vacuum plasma spraying

XRD x-ray diffraction

YLaSZ yttria-lanthana stabilized zirconia

YSZ yttria stabilized zirconia

Chapter 1: Introduction

1

1 Introduction

Gas turbine engines operate under severe thermal, mechanical, and chemical conditions. Turbine

technology strongly depends on development of suitable material systems that can operate in such

harsh environment. Successive modification of alloy compositions lead to the introduction of

nickel based superalloys in jet design. Further efficiency improvements of aero engines as well as

land-based gas turbines require increased operating temperatures. However, limits on the operating

temperature are imposed by the onset of melting of the superalloys. The use of thermal barrier

coatings allows increases in the operating temperature without any increase in the temperature to

which the metal is exposed, it can reduce the amount of cooling required whilst maintaining the

operating temperature of the turbine. Alternatively, improved durability and reliability will result

by reducing the component temperatures. Coatings are thus commonly applied to these components.

Thermal barrier coatings (TBCs) for use on turbine blades consist of a two-layer system. A

relatively thin “bond coat” is first deposited and a second layer (“top coat”) is then sprayed onto the

first. The bond coat is typically a MCrAlY metal alloy (e.g. Ni-22wt%Cr-10wt%Al-1wt%Y). The

purpose of the bond coat is to protect the substrate from oxidation and hot corrosion and also

improve the adhesion between the substrate and the top coat. The industry standard material used

as a top coat is zirconia, usually stabilised with yttria. During service a thin thermally grown oxide

(TGO) layer forms as a result of bond coat oxidation.

TBCs are commonly applied by plasma spraying. In plasma spraying, powder particles are injected

into a plasma flame where they are melted and accelerated towards the substrate. The molten

particles impact the substrate, spread out, cool and contract to form lenticular splats which built up

the coating. Advantages of the plasma process are that it is relatively inexpensive, offers high

deposition rates, and is capable of producing durable and reproducible coatings. The application of

thermal barrier coatings for combustor parts, augmentors, vane nozzles, transition pieces and

stationary turbine components today is widely practiced. However, the use of plasma-sprayed

TBCs as a design element in hot rotating parts of the turbine engine is still quite limited. The

application of TBCs as an integral part of such components requires reliable and predictable TBC

performance.

Failure of TBCs usually occurs by buckling and spalling of the coating. Plasma sprayed coatings

are relatively stable against spallation driven by residual or externally applied stresses. The main

reason for this is that the zirconia has a very low stiffness. This prevents large stresses, and hence

Chapter 1: Introduction

2

large driving forces for spallation, from being generated in the top coat. This low stiffness and high

strain tolerance is largely a consequence of the presence of many fine microcracks and pores in the

zirconia layer. In addition to this beneficial effect on the stiffness, these features reduce the thermal

conductivity, which is also desirable. Recent work has shown that another significant change

during prolonged heating is substantial stiffening of the top coat. This arises from sintering

processes, which can raise the stiffness.

The present research aims to investigate the thermophysical properties of plasma sprayed top coats

in TBCs. Together with the widely used zirconia-stabilized with yttria, other novel compositions,

such as dysprosia-stabilized zirconia, yttria-lanthana-stabilized zirconia and ceria-stabilized zirconia

are explored, as well the effect of impurities. The research is focused on the changes that occur in

the phase constitution, microstructure, stiffness and thermal conductivity upon exposure to high

temperatures. The differences in the sintering behaviour and thermal conductivity of differently-

stabilized zirconia top coats are investigated. In-service TBCs operate, in the presence of a

through-thickness thermal gradient, and interest is focused in particular on the behaviour of top

coats under such thermal gradients.

Chapter 2. TBCs for Gas Turbine Applications

3

2 Thermal Barrier Coatings for Gas Turbine

Applications

2.1 The gas turbine engine

A turbine is a rotary engine that converts the energy of a moving stream of fluid into mechanical

energy. The fluid can be water, steam, or gas. The basic element in a turbine is a wheel or rotor

with paddles, propellers, blades, or buckets arranged on its circumference in such a fashion that the

moving fluid exerts a tangential force that turns the wheel and imparts energy to it. This

mechanical energy is then transferred through a drive shaft to operate a machine, compressor,

electric generator, or propeller. Turbines are classified as hydraulic or water turbines, steam

turbines, or gas turbines. Most modern passenger and military aircrafts are powered by gas turbine

engines, which are also called jet engines. Today’s turbine-powered generators produce most of the

world's electrical energy. There are several different types of gas turbine engines, but all have some

parts in common. The main parts of the gas turbine are the fan or inlet, the compressor, the

combustor or burner, the turbine and the exhaust or nozzle (see Figure 2.1).

All gas turbine have an inlet or fan to bring air into the engine. The incoming air enters the

compressor, which is made up from a series of aerofoil blades attached to a shaft. These control the

airflow rate and, as air goes through progressively smaller areas, the air pressure increases. The

pressure of the air can rise up to 40 atmospheres. This increases the potential energy of the air. The

high-pressure air is forced into the combustion chamber. In the combustor, air is mixed with fuel

that is sprayed into the air stream by a ring of fuel injectors. The fuel is generally kerosene,

propane or natural gas. The air/fuel mixture is ignited and burned at temperatures in a range of

1450-2000oC [1, 2 , 3, 4] to generate the maximum possible heat energy. The highest temperature

in a turbine engine occurs in the combustion chamber followed by the inlet temperature at the first

blades of the turbine. The combustion produces a high temperature, high pressure gas stream that

enters and expands through the turbine section. The turbine is an intricate array of alternate

stationary and rotating aerofoil-section blades. As hot combustion gas expands through the turbine,

it spins the rotating blades. The turbine is linked by a shaft to the blades in the compressor and the

inlet fan. Some of the energy from the rotating blades is used to drive compressor and the inlet fan

via the shaft. For power generation the rest of the energy is used to spin a generator to produce

electricity. For jet engines in aircrafts the combustion gases are expelled through the exhaust or

nozzle. The resultant expelled exhaust gas generates the thrust which is the force that moves a jet


4

aircraft through the air. In the case of the turbine used in a power plant, exhaust gases are vented

through an exhaust pipe or are used for a heat exchanger either to extract the heat for some other

purpose or to preheat air before it enters the combustion chamber.

2.1.1 Materials for Gas Turbines

The first gas turbine operated at a temperature which was based on the limitations imposed by the

high-temperature strength of the existing materials. The need for development of new materials

with strength at high temperature arises from the fact that temperature and pressure directly

influence the efficiency of the gas turbine cycle. Both thermal efficiency and power output are

influenced by the maximum cycle temperature and pressure ratio. Broadly speaking, the maximum

temperature has the more important effect on power output whereas the pressure ratio has greater

influence on the thermal efficiency [5].

Higher pressure ratios and higher maximum temperatures are used in aeroengines than in industrial

gas turbine. Turbine gas entry temperatures today are in the range of 1100-1350oC for industrial

gas turbines and around 1500oC for advanced aircraft engines [1, 3, 4, 6-9]. Adjacent to the blades

there is layer of cooler gas that keeps the surface of the blades at lower temperatures. The

operational life requirements are different for aircrafts and industrial gas turbines. Industrial gas

turbines operate for substantially longer cycles and times between overhauls. A typical aero engine

operates for 5000-10,000 cycles between overhauls and cycle times are between 0.5-13 hours.

Industrial gas turbines operate 30,000-50,000 hrs between overhauls and cycle times are 100-

2500 hours [7, 9-11]. In aircraft gas turbine the weight must be kept low and the shorter life allows

higher stresses and temperatures to be used than would be permissible in long-life industrial plants.

From the aspect of material selection, different design parameters apply for different parts of the gas

turbine. Compressor materials must have adequate fatigue strength, in order to be able to withstand

the centrifugal and vibrational stresses. Additionally, they must have satisfactory impact resistance,

since there is some likelihood of the injection of foreign objects. The initial compressor stages

operate at relatively low pressures and temperatures, so that light alloys or composites can be used.

However, for the later stages, where temperature and pressures are higher, creep and oxidation

resistance become important and stronger alloys are necessary.

In the combustor, the fuel is ignited and the air/fuel temperature is in a range of 1450-2000oC [1, 2 ,

3, 4]. The combustor walls need to be continuously cooled in order to maintain the temperature at

safe levels (below1100°C) [10]. Materials in the combustor experience high stress conditions.

However they need to maintain their strength at high temperatures. Furthermore oxidation

http://science.howstuffworks.com/hydropower-plant.htm


5

resistance and thermal shock resistance are important. Failures in the combustor chamber occur

mainly by cracking and buckling at hot spots.

Figure 2.1 Pressure andtemperature of the gas as it passes through the various stages of a jet engine. (after

[1]).

The turbine is the most demanding environment from the material’s performance point of view.

Materials operate in an environment with a combination of very high temperature and high stresses

(Figure 2.1). Turbine blades exist in a more hostile environment than compressor blades. They

must have the maximum possible resistance to creep and thermal fatigue, and high tensile strength,

since they experience high centrifugal stresses. Materials with high strength, ductility, high thermal

conductivity and low thermal expansion provide the best resistance to thermal fatigue. Also,

turbine blades must have good oxidation resistance, since they are in contact with the high

temperature combustion products with high oxygen content. In industrial turbines particularly, hot

corrosion resistance is also an important requirement. Since turbine entry temperatures are so high,

often exceeding the melting point of the metal used, blades must be actively cooled. A cooling

system is used in which cool air is continuously circulated through the blade, to keep it relatively

cool. The active cooling of blades establishes high thermal gradients between the inside and the

surface. This introduces variations in the physical properties and thermal stresses. Other required


6

properties for turbine blades include good impact resistance and microstructural stability over long

periods of time. Complex materials have been developed for use in turbine blades [12, 13]. More

details on the superalloys used in turbine blades is given in § 2.2.1.

Turbine disks are also rotating components in the turbine and operate under high temperatures and

centrifugal stresses. Therefore these materials must have very high tensile strength, good creep

resistance and good thermal fatigue resistance. Stator blades, and nozzle guides vanes are not

exposed to the influence of centrifugal stresses and hence the primary material requirements are

oxidation, hot corrosion and erosion resistance.

2.2 Thermal Barrier Coating Systems

Thermal barrier coatings (TBCs) in engines have been used experimentally since the 1960s and by

the 1980s they were being used in certain commercial gas turbine engines in low risk regions [14,

15]. TBCs offer benefits in terms of the performance and efficiency of gas turbines, allowing lower

metal temperatures, reduced requirements for active cooling, higher turbine entry temperatures and

longer component lifetimes. The application of thermal barrier coatings for combustor parts,

augmentors, vane nozzles, transition pieces and stationary turbine components today is widely

practiced [16]. However, the use of TBCs in hot rotating parts of the aircraft turbine engine usually

is not integrated in the design, i.e. TBCs are used to lower the superalloy temperature and thus

prolong the life-time of the blade, but failure of the TBC is not critical. If TBCs are to fulfill their

full potential, reliability is necessary.

A typical TBC system consists of (i) the top coat (TC), a porous ceramic layer that acts as the

insulator, (ii) the bond coat (BC), an oxidation-resistant metallic layer between the substrate and the

TC and (iii) the superalloy or other material substrate that carries the structural load (see Figure 2.2).

TBCs are usually produced either by electron-beam physical vapor deposition (EB-PVD) or by

plasma spraying (PS).

2.2.1 Nickel Based Superalloys

Superalloys have been described as “alloys developed for elevated temperature service, usually

based on group VIIIA elements, where relatively severe mechanical stressing is encountered and

where high surface stability is frequently required” [17]. Superalloys are used in aircraft, marine,

industrial and vehicular gas turbines and also in space vehicles, rocket engines, nuclear reactor,

submarines, steam power plants and petrochemical equipment. However, their most important use

is in the gas turbine aeroengines.


7

High pressure turbine blades operate under severe mechanical and thermal environments. Turbine

blades are made from nickel based superalloys. Wrought turbine blade compositions are primarily

based on 80% Nickel 20% Chromium alloys. Progressively higher amounts of Aluminium,

Titanium and associated compositional modifications increased the temperature capability of the

alloys [2]. Nickel based superalloys can currently operate up to homologous temperatures of 0.8Tm

(about 1100oC) [8, 18].

Te

mp

era

ture

~1100oC

~600oC

~1300oC

Substrate Thermo-mechanical loading

Bond Coat Oxidation protection/bonding

Top Coat Thermal Insulation

Ni-base superalloys

ZrO2+(6-8)%Y2O3

Hot Gases

MCrAlY or Aluminides

Coolant

Figure 2.2 Schematic of typical structure and qualitative temperature profile through a TBC.

Figure 2.3 Increase in operational temperature of turbine component made possible by alloy development,

manufacturing technology and TBCs [18].


8

Nickel provides a face-centered cubic(FCC) lattice, which has a high modulus, 5 independent slip

systems and low diffusivity for substitutional solutes. The formation of coherent gamma prime

precipitates and carbides lead to strengthening. Advanced turbine blade materials have a relatively

high volume fraction of gamma prime (Ni3Al), around 70% [19]. Aluminium forms a stable

alumina scale during exposure to high temperature and also strengthens the alloy by forming

gamma prime. Other alloying elements are added for different reasons. Chromium may form an

adherent oxide scale which also protects against oxidation. Titanium can replace some of the

Aluminium in gamma prime as well as forming titanium carbides, which also contribute to

precipitate strengthning. Elements such as C, B, Zr and Hf provide grain boundary strengthening in

polycrystalline alloys[2, 12].

Superalloys were conventionally produced by casting methods. However, superalloys produced by

casting methods often did not exhibit consistent creep properties. This problem led to the

development of directional solidification (DS), which produces castings with grains aligned in the

direction of maximum stress and few grain boundaries normal to this direction. Directional

solidification is achieved by ensuring that the heat during solidification of the casting is removed in

a direction parallel to the desired growth direction, while a liquid/solid interface perpendicular to

the solidification direction is maintained. Several techniques have been successfully developed in

which the production cost of DS casting is no greater than that of conventional castings. DS

resulted in a significant increase in the creep strength of these superalloys, relative to

conventionally cast alloys, and led to an increase in the temperature capabilities of superalloys.

A further important development in the processing of superalloys was the production of single

crystal castings. This technology was achieved by extending the DS technology and adding a

simple modification to the casting mould, either a crystal selection device or a “seed” crystal.

Single crystals have improved creep properties relative to DS superalloys. The complete absence of

grain boundaries also slightly increases the solidus temperature of the alloy. This permits higher

temperature solution heat treatment, which is beneficial because it leads to reduced chemical

segregation, more uniform gamma prime distribution and consequently increased temperature

capability. Furthermore, elements such as C, B, Zr, and Hf, which are normally necessary in

polycrystalline castings to produce grain boundary strengthening, can be eliminated in single crystal

superalloys [12, 20].


9

2.2.2 Bond Coat

The bond coat protects the underlying substrate from oxidation and improves adhesion between the

ceramic and the metal. Oxidation occurs due to oxygen reaching the bond coat by diffusion through

the lattice of the top coat and permeation through the pores [21]. The yield and creep characteristics

of the bond coat are thought to be significant for the performance of the TBC system.

Commonly used bond coats can be divided in two categories: MCrAlY (where M= Co or Ni or both)

and Pt-modified aluminides [14]. These coatings were developed for use as protective coatings

against oxidation and hot corrosion [22]. When exposed to an oxidizing environment, they form a

stable dense alumina layer in preference to other oxides. This alumina, often termed the thermally

grown oxide (TGO) prevents further attack of the underlying material, due to its low oxygen

diffusivity and its good adherence. MCrAlY bond coats are usually deposited by low-pressure

plasma spraying and consist of two phases (β-NiAl and either γ-Ni solid solution or γ’Ni3Al) [23].

Small amounts of Y are added in order to improve TGO adherence [24]. Yttrium additions have

been found to inhibit void formation at the TGO/BC interface. In addition, Y-rich oxide protrusions

are formed in the oxide that mechanically peg the oxide to the alloy. Furthermore, yttrium has the

effect of decreasing the grain size of the TGO and thus raising its mechanical strength [25].

Pt-modified aluminides are usually fabricated by electroplating a thin Pt layer on the superalloy and

then aluminizing by chemical vapor deposition or pack cementation. These coatings usually consist

of a single-phase-β with Pt in solid solution. Platinum additions improve the spallation resistance

of conventional aluminide coatings. However, the mechanisms by which this occurs are not fully

understood.

Optimum adhesion between the bond coat and the top coat is attained differently in plasma sprayed

and EB-PVD coatings. In plasma sprayed coatings, it is achieved by mechanical interlocking of the

two interfaces, so the surface roughness of the bond coat is an important parameter [26]. In contrast,

EB-PVD coatings achieve maximum durability when applied to a smooth (preferably polished)

surface, free of absorbed gases or loose oxides. Asperities in the BC/TGO interface are thought to

serve as nucleation sites for cracks that cause coating spallation when they coalesce [23, 27].

MCrAlY bond coats creep at temperatures above 800oC [28-30]. At this temperature, stresses in the

BC are relieved and it is non-load bearing. The creep behaviour of the BC can have a significant

influence on the stress state of the TBC and thus on the failure mechanisms. More detailed

discussion on the influence of creep on the failure mechanism of TBCs is given in section 3.3.


10

2.2.3 Top Coat

The top coat provides thermal insulation for the underlying substrate. The specifications for this

coating require a material that combines low thermal conductivity and a coefficient of thermal

expansion (CTE) that it is as similar as possible to that of the substrate, so that generation of

stresses during thermal cycling can be minimised. The preferred material for this application is

zirconia. Zirconia may exist as three solid phases, that are stable at different temperatures [31]. At

temperatures up to 1200oC, the monoclinic phase (m) is stable. Zirconia transforms from the

monoclinic to the tetragonal phase (t) above 1200oC and above 2370

oC to the cubic phase (c).

Transformation from m to the t phase has an associated volume decrease of 4% [32]. To prevent

catastrophic cracking as a result of the volume changes accompanying the t→m transformation,

which occurs at temperatures within the range of the working environment in gas turbines,

stabilizers are added to the zirconia. These stabilize zirconia into its cubic or tetragonal phases.

Early attempts used MgO to stabilize zirconia in its cubic state, by adding 25 wt% MgO [33].

However, during heat treatment the zirconia reverts to its monoclinic form and the stabilizing oxide

precipitates out from solid solution, affecting the thermal conductivity [34]. Zirconia can be fully

stabilised to its cubic phase by adding 20 wt% yttria. However, such fully stabilised zirconia

coatings perform very poorly in thermal cycling tests [35]. Typically 7-9wt% yttria is used to

partially stabilise zirconia, although other stabilizers have been used as well. Other stabilizers

include CaO, MgO, CeO2 Sc2O3 [36-38]. Rare earth dopants such as Dy2O3 and Yb2O3 have also

generated some interest [39, 40]. The basic criteria for the selection of a suitable stabiliser include a

suitable cation radius, similar to that of zirconium, and a cubic crystal structure. In spite of the

addition of a stabilizer in order to ensure phase stability of the top coat, phase changes in the top

coat might still be induced during service. This is described in more detail in section 3.6.

An important aspect of the performance of top coat material is its sintering behaviour. After

prolonged heating during service, sintering of the top coat can occur. This will result in healing of

the microcracks and pores that will in turn reduce the strain tolerance of the coating and increase the

likelihood for spallation. More detailed discussion of the influence of top coat sintering on the

lifetime of TBCs is given in the section 3.5.


11

2.3 Coating Systems

2.3.1 EB-PVD Coatings

The EB-PVD process takes place in an evacuated chamber. A high energy electron beam is used to

heat and vaporize the coating material. The vapor formed contains the coating atoms and molecules

[7]. To ensure the correct stoichiometry of the ceramic top coat, a controlled amount of oxygen is

bled into the chamber. The preheated substrate is inserted in the vapor cloud and the vapor

condenses onto the surface of the substrate, forming a coating, usually at a deposition rate of

4-8 m min-1

[6]. EB-PVD is used both for the deposition of bond coats and ceramic top coats.

Ceramic coatings generated by EB-PVD have a columnar microstructure, after an initial thin region

of dense ceramic (Figure 2.4). This structure exhibits high lateral strain tolerance. In addition, they

exhibit a good surface finish which is beneficial for aerofoil applications, where a low coefficient of

friction is desirable. Thermal conductivity of partially stabilized zirconia (PSZ) TBCs produced by

EB-PVD is typically in the range of 1.5-1.8 W m-1

K-1

[41], which is higher than values for plasma

sprayed TBCs. This is mainly due to the characteristic columnar structure of EB-PVD coatings, that

does not hinder thermal heat transfer in the through-thickness direction, since voids and interspaces

are mainly aligned parallel to the heat flux direction.

2.3.2 Plasma Sprayed Coatings

A characteristic of all thermal spray processes is a highly concentrated power source, to which the

coating material is fed in the form of powder, wire or rod. The coating material is melted and

accelerated to the substrate, forming the coating. The coating is formed of many overlapping splats,

solidifying one after another and locking one to another. Due to the high kinetic energy of the

droplets, the splats spread over the substrate, forming a pancake.

Plasma spraying is described in detail section 2.4. It is widely used for the production of TBCs.

The use of PS TBCs in the hot rotating section of aeroengines is still very limited and if they are to

fulfill their full potential, their reliability must be improved [42].

PS TBCs have the necessary strain tolerance required for most of the applications in which such

coatings are currently applied. This is largely a consequence of the presence of many fine

microcracks and pores in the microstructure, which results in low stiffness (Figure 2.4). This low

stiffness prevents large stresses from being generated in the top coat. The thermal conductivity of

plasma sprayed coatings range from 0.5-1.4 W m-1

K-1

[43], which is lower than corresponding

values for EB-PVD coatings. The microstructure of PS TBCs exhibits pores and grain boundaries


12

aligned perpendicular to the direction of heat flux (see Figure 2.4). Grain boundaries and pores

hinder heat transfer [44]. The shape and orientation of porosity with respect to the heat flux are

more critical factors than the total amount of porosity for the thermal conductivity of PS coating

[45]. More details on the thermal conductivity of TBCs are given in section 4.3. EB-PVD coatings

offer benefits over PS coatings in terms of the erosion resistance [46-48]. In PS coatings, the

erosion occurs in the form of removal of the mechanically bonded splats by the erosive material.

Since intersplat porosity is already present, the energy required for this process is low.

The low cost associated with the PS process compared to EB-PVD makes PS TBCs the more

attractive options for many gas turbine components such as combustion chambers, nozzle guides

and abradable seals [16]. However, applications that require excellent strain tolerance, good surface

finish and erosion resistance, such as in aerofoils and aero-gas turbines, EB-PVD coatings will be

favoured [49].

Figure 2.4 Schematic illustrations of the pore morphology of (a) a plasma spray (PS) deposited coating and (b)

an electron beam physical vapour deposited (EB-PVD) coating [50].


13

2.4 The Plasma spraying Process

2.4.1 The Plasma Jet

Plasma Spraying, first conducted by Reinecke in 1939, was advanced in the late 50´s by several

other scientists [51]. Since then, it has become increasingly sophisticated and is nowadays widely

used in surface technology.

The plasma spraying gun consists principally of two electrodes.

Figure 2.5 shows a schematic of the plasma spray gun, with the thoriated tungsten cathode inside

the water-cooled copper anode. A gas, commonly a mixture of argon and hydrogen, is injected into

the annular space between the two. To start the process, a DC electric arc is stuck between the two

electrodes. The electric arc produces gas ionisation, i.e. gas atoms lose electrons and become

positive ions. Electrons move with high velocity to the anode, while ions move to the cathode. On

their way, electrons and atoms collide with neutral gas atoms and molecules. Hence, the electric arc

continuously converts the gas into a plasma (a mixture of ions and electron of high energy). The

plasma is on average, electrically neutral and characterized by a very high temperature [52]. The

kinetic energy of the plasma (mostly carried by free electrons) is converted into thermal energy

during collisions between ions, electrons and atoms. In this way, the plasma is capable of producing

temperatures up to approximately 104K [53]. The hot gas exits the nozzle of the gun with high

velocity.

Powder material is fed into the plasma plume. The powder particles are melted and propelled by

the hot gas onto the surface of the substrate [54]. When individual molten particles hit the substrate

surface, they form splats by spreading, cooling and solidifying. These splats then incrementally

build the coating.

Plasma plumes exhibit radial temperature gradients. Whereas particles that pass through the central

core of the plasma tend to be melted, superheated or even vaporised, particles that flow near the

periphery may not melt at all. This will affect the final structure of the coating, which may contain

partially molten or unmelted particles. Voids, oxidised particles and unmelted particles can appear

in the coating (see Figure 2.6). These effects may be desirable, or they may be unwanted, depending

on the requirements of the coating.


14

Water-cooledcopper anode

Tungsteninsert

Water-cooledcopper cathode

Insulating backplate with angled

gas channels

Plasmagases

Arc

Powder

Plasma

Figure 2.5 Schematic of Plasma spray gun [55].

2.4.2 Coatings Microstructure and Properties

Plasma sprayed coatings are built up particle by particle. Molten droplets arrive one at a time and

impact with the underlying material, which consist of previously solidified droplets (splats), and

there they solidify. Hence, the microstructure of the PS coatings is different from that of most other

materials [56]. Plasma sprayed coatings have a non-homogeneous, layered structure, consisting of

splats with a pancake-like shape. Also, partially molten particles, oxides and voids are present in a

typical PS structure (see Figure 2.6).

Porosity levels in plasma sprayed ceramic coatings are generally in the range 3-20% [57]. Porosity

might be in the form of inter splat porosity (fine gaps between one lamella and another) or larger

irregularly shaped void which result from incomplete conformation of splats to the topography of

the impact site. For thermal barrier coatings, high porosity may be desirable, since the thermal

conductivity of the coating is decreased by the presence of porosity [58]. Certain applications

require low porosity and this can be achieved by controlled spraying conditions so as to raise

droplet velocities.

Plasma sprayed coatings also have different mechanical properties from corresponding bulk

material. Strength values of the coatings are usually only a small fraction of the values of

corresponding dense materials. The Young’s modulus of metallic coatings obtained by plasma

spraying has been found to be of the order of 1/3 that of bulk material [59], though ceramic coatings

give lower modulus. For PSZ, typical values of Young’s modulus for as-sprayed coatings are


15

10-40 GPa in comparison to 210 GPa for dense zirconia [60-65]. Elastic anisotropy has also been

observed in plasma sprayed deposits which is attributed to the preferred orientation of planar crack-

like defects for PS zirconia [66].

The thermal conductivity of PS coatings is usually lower than corresponding values for the bulk

material. The coatings exhibit pores and planar defects which inhibit heat transfer [67]. The

coefficient of thermal expansion, however, is not influenced by the coating morphology since pores

and voids do not contribute to expansion. This was confirmed by Kuroda and Clyne [59], who

measured the CTE of plasma-sprayed coatings and found it to be close to that of bulk material.

Schwingel et al [68] also reported an average value of ~10 10-6

K for YSZ at room temperature,

similar to that of sintered YSZ. Ahmaniemi et al [69] observed linear thermal expansion of plasma

sprayed 8wt%-YSZ in the temperature range 50-1000C and reported a CTE of ~ 9.9 10-6

K-1

.

For plasma sprayed ceria stabilized zirconia the same was found to be ~ 10.8 10-6

K-1

. The CTE

of plasma sprayed Lanthanum hexaaluminate was found to be lower (7.7-9.3 10-6

K-1

) than YSZ

coatings (10-11.1 10-6

K-1

) in the temperature range 100-1300C[70].

Figure 2.6. Schematic diagram showing typical microstructural features in plasma sprayed coatings.

Adhesion of plasma sprayed coatings to the underlying substrate is achieved mainly by mechanical

interlocking. Therefore, the surface roughness of the substrate often plays a role in interfacial

adhesion. Solid-state diffusive bonds occurring at a molecular level is also believed to play a minor

role in the adhesion of PS coatings. These non-mechanical bonding mechanisms are not yet fully

understood [54].

During plasma spraying, stresses are generated in the coating as consequence of the rapid

solidification of the molten droplets (quenching stresses) and during subsequent cooling as a result


16

of mismatch of the thermal expansion coefficients of substrate and coating. More detailed analysis

of these stresses is presented in section 3.1.

2.4.3 Parameters Affecting the Process

In plasma spraying, there are many parameters that need to be adjusted and therefore the task of

optimisation and process control is complex. In general, the properties of plasma sprayed coatings

depend on processing variables [71-73]. These include stand-off distance, plasma gas composition,

plasma power and the powder injection rate. However, these parameters are subject to certain

limitations, arising from the requirement that the powder can be melted. The effects of these

variables are now briefly considered.

The power absorbed in the plasma depends on the arc current and the voltage drop across the

electrodes. Also, the way the electric power is applied (i.e. with or without transferred arc, positive

or negatively charged) is an important factor. The power supply affects the temperature and

velocity of the plasma flame and therefore influences the melting and acceleration of the powder

particles. The plasma arc power affects the spraying efficiency. The heat available in the arc, and

the arc temperature for a set of spraying conditions, will be controlled by the arc power, which in

turn depends on the gas mixture. Low power may result in incomplete melting of the injected

powders whereas too high a power can cause vaporization of the particles. The optimum power

settings for spraying will depend on the thermophysical properties of the material sprayed.

The stand-off distance between the nozzle and the substrate influences the deposition of the powder

particles. Depending on the distance, the residence time of the powder particles in the plasma

plume, differs. This results in different velocities and temperatures of the powder particles when

impacting the substrate.

The powder variables that affect the spraying process depend mainly on the physical properties of

the material sprayed and the characteristics of the powder. The physical properties include the

melting and vaporization temperature of the bulk material, its heat transfer properties (specific heat,

thermal conductivity and latent heat of fusion) and its density. Density will affect the particle

trajectory, for a given particle size and plasma plume characteristics. Powder characteristics

include particle size and shape, which are important from the heat-transfer viewpoint. Larger

particles may be incompletely melted, whereas small particles may vaporize. Therefore, for a given

set of plasma conditions, there is an optimum particle size. In general, powders with a narrow range

of particle sizes are preferred.


17

Plasma spraying offers technical advantages over other coating processes [74]. It can deposit metals,

ceramics or combinations of these, since the high plasma temperatures permit the spraying of

materials with high melting points. In addition, high particle velocities result in generating

microstructures with relatively low porosity and high bond strength.

Economically, most important is atmospheric plasma spraying (APS). It is often recommended,

provided oxidation of the coating material is not an issue, e.g. in cases of oxide ceramics or where

oxidation to a certain degree does not harm the performance of the sprayed coating. In cases where

oxidation of the coating has to be suppressed, the production can take place in an inert or protective

environment, provided a spray chamber is available. However this requires greater initial capital

investment.

Chapter 3. Review of Thermomechanical behaviour of TBCs

18

3 Review of Thermomechanical Behaviour of Thermal

Barrier Coatings

The most common failure mode of TBCs is spallation of the top coat (TC), usually by cracking

along the interface between the thermally grown oxide (TGO) and the bond coat and/or cracking in

the interface between TGO and TC. There are different factors that affect the thermo-mechanical

stability of TBCs. Significant factors are the stress state in the TC, the TGO thickness and the

interfacial adhesion between BC/TGO/TC amongst others. A brief description of the properties of

TBCs and the induced changes during service conditions will be presented here.

3.1 Stresses in Plasma Sprayed Coatings

Residual stresses in plasma sprayed coatings have been extensively studied. They can result in the

deformation of coated pieces and contribute to cracking and spallation of the coating. In this

section, an overview of the mechanisms of generation of residual stresses is presented. The

techniques used in order to measure residual stresses are assessed and numerical modelling

approaches for the prediction of residual stresses are described.

3.1.1 Mechanisms of Stress Generation

3.1.1.1 Quenching Stress

During coating formation, stresses are generated, the so called “deposition” or “quenching stresses”.

These stresses are produced when a molten droplet of the coating material impacts the cold

substrate, spreads and solidifies. As the molten splat strikes the substrate, it spreads and rapidly

loses heat through conduction to the underlying material. When it solidifies, its thermal contraction

is restricted by this material and thus a tensile stress is generated. It has been shown that the

magnitude of the quenching stress also depends on substrate temperature [59], although it is

independent of the substrate material [75]. Assuming perfect interfacial bonding and elastic

behaviour of the splat, the maximum value that the quenching stress can reach is given by [59]:


19

( )

1

m sq d d

d

T TE

(3.1)

where αd is the coefficient of thermal expansion of the deposit, Ed is Young’s modulus for the

deposit, d is Poisson’s ratio for the deposit and Tm, Ts are the deposit melting temperature and

substrate temperature respectively. This value is commonly much larger than the yield stress; thus

plastic flow and other stress relaxation mechanisms may operate. These mechanisms are illustrated

in Figure 3.1. Stress relaxation mechanisms are different for metallic and ceramic coatings. For

metallic coatings, the yield stress is often low and the temperature of the substrate material can

often be a considerable fraction of the melting temperature of the coating material, so that creep and

yielding are common stress relaxation mechanisms. On the other hand, ceramic coatings have high

melting points and therefore do not readily relax by creep or yielding. Ceramic coatings, however,

commonly relax stress by microcracking.

Figure 3.1 Quenching stresses and relaxation mechanisms [59]..

Quenching stresses have been measured experimentally by Kuroda and Clyne [59]. The results

confirm that quenching stresses are usually much lower than the maximum theoretical value

(typically ~1 GPa). For metallic materials, the quenching stress can be quite high, especially for

materials with high yield and creep strength. For ceramics and other brittle materials, the

quenching stress is usually relatively small, as a consequence of microcracking (Table 3.1).


20

Property Bond Coat (CoNiCrAlY)

ZrO2-8wt% Y2O3 Substrate

(Nimonic 80A)

CTE (α) 10-6

K-1

13.0 (298 K)

17.1 (773 K)

10.5 (283 K)

11.5 (811 K)

13.0 (298 K)

16.6 (773 K)

Thickness (mm) 0.1 0.3 10

Quenching Stress (MPa)

115 5 -

Table 3.1 Quenching stress in plasma sprayed deposits [76].

3.1.1.2 Differential Thermal Contraction

The coefficient of thermal expansion is in general different for the substrate and the sprayed

material. During plasma spraying, the temperatures of the substrate and deposit are often high.

Therefore, during subsequent cooling of the sample, the mismatch in thermal expansion properties

will generate residual stresses. This mismatch in the coefficients of thermal expansion (CTE) will

also give rise to stress whenever the specimen is heated and cooled.

In addition to the magnitude of the thermal expansion mismatch, the stress state will depend upon

the thermal history, as well as the clamping arrangement during the spraying process (whether the

specimen is rigidly clamped or is allowed to bend). Assuming elastic behaviour and a thick

substrate (compared to the coating), such that the strain is accommodated by the coating alone, the

bi-axial residual stress in the coating (assuming transverse isotropy) can be estimated from the

expansivities of the substrate and the coating. Consider two bonded plates cooled from a stress-free

state through a temperature interval T. A misfit strain =T will be generated and stress

generated in the coating as a result of the constraint of such strain:

)1(

)(

d

dsd

d

TE

(3.2)

where E is Young’s modulus, α is the coefficient of thermal expansion, is Poisson ratio and the

subscripts d and s refer to the deposit and substrate respectively. However, the assumption made

above in not always valid and properties of the substrate material may also be significant in the

generation of residual stresses (particularly for thin substrates).

In addition, large thermal gradients can arise during the spraying process, which may generate

stresses. These also tend to cause curvature during spraying. Thus the sample may not be in a

stress-free state when cooling commences. Evidently, the details of the generation of stress due to

differential thermal expansion can be complex and often numerical modelling is required.


21

Typically, ceramics have low CTEs, compared to most metallic materials, and hence large

differential thermal contraction stresses can be produced. Residual stresses present in ceramic

coatings are therefore governed by stresses generated due to differential thermal contraction, since

quenching stresses in ceramics are in general low.

The generation of residual stresses has been thoroughly investigated for thermal barrier coatings

(TBCs) and in particular for partially stabilized zirconia (PSZ) coatings [77-80]. A gradient in the

stress state throughout the thickness of the coating has commonly been reported, with the surface

stress being more tensile and the stress at the interface more compressive [77]. The residual

stresses in the as-deposited top coat changes from tensile to compressive with increasing substrate

temperature [80, 81].

3.1.1.3 Solid State Transformations

Solid state transformations may occur during spraying and in such cases they may have an effect on

the stress field of the coating. Volumetric changes associated with solid state transformations will

induce misfit strains in the coating. A solid state transformation causing an increase in volume, will

induce compression, whereas a decrease in volume will cause tension. One such process that may

result in the generation of stresses is oxidation. This is particularly important in the case of spraying

metallic materials, such as the bond coat of TBCs, where oxidation is thought to play a significant

role in determining the coating’s lifetime. To minimise oxidation effects during spraying, the

process can be carried out at low pressure (Low Pressure Plasma Spraying, LPPS). The mechanism

of bond coat oxidation and its significance for coating lifetimes is discussed in section 3.3. Solid-

state transformations may occur during annealing of zirconia top coats used in thermal barrier

coatings. This effect is discussed in section 3.6.

3.1.2 Measurement of Residual Stresses

There has been extensive research on experimental methods for measuring residual stresses in

sprayed coatings, but there is no entirely satisfactory method to achieve this. A brief overview is

given below.

3.1.2.1 Curvature Methods

Stresses produced during spraying generate a misfit strain. Applying elastic beam theory, this misfit

strain will cause an unbalanced bending moment in the sample. If the specimen is free to bend, it


22

will acquire a curvature in order to minimize the strain energy. The curvature of the specimen is

hence a source of information on its stress state.

The curvature retained by the sample after the spraying process has been utilized by Hobbs and

Reater [82, 83] in order to deduce information on the stress state. However this method relies on a

single curvature measurement and there are many ambiguities associated with it. Kuroda et al [84,

85] were amongst the first to perform continuous measuring of curvature during the spraying

process and this has proved to be a powerful technique. This was realized by attaching a pair of

light contacting knife edges to a strip-shaped substrate and a contacting displacement meter at the

rear surface of the substrate. This technique was employed to calculate the quenching stress for

different materials and substrate temperatures [86, 87]. Stoney’s equation [88] was used for

estimating the quenching stress:

2

6 1

s sq

s d d

E h

h h

(3.3) (Stoney equation)

where E is the Young’s modulus, h refers to thickness, ν is the Poisson ratio, is the curvature and

the subscripts s and d refer to substrate and deposit respectively.

To ensure that the substrate temperature was kept constant during spraying, a pair of air jets was

provided by a pair of nozzles attached to the plasma torch. For the vacuum plasma spraying (VPS)

process a non-contacting method was employed [59, 87]. The coating material was sprayed onto

substrates of the same material. In this way, no thermal residual stress between the two layers is

generated, since there was no mismatch in the CTE for the two layers. Thus, the post spraying

curvature of the sample can be directly linked to the quenching stresses associated with the process.

The equation used for the calculation of the quenching stress is given below [89]:

1.25

6

ds s s d

s

q

d

EE h h h

E

h

(3.4)

in which the variables are as described for (3.3). Further development in the curvature-monitoring

technique was achieved by Gill and Clyne [90, 91]. A non-contacting method of curvature

measurement was employed with the aid of a video camera. The video camera records deflection of

the free end of the sample during spraying and subsequent cooling. The information recorded can

be transformed into curvature data. This method has the advantage of eliminating any concern about

interference of the previously used instruments with the free bending of the sample and a wide


23

range of temperatures can be examined. The technique has allowed validation of a numerical

process model described in section 3.1.3.

3.1.2.2 Diffraction Methods

Spacing between lattice planes can be determined using diffraction methods. Residual stresses will

induce strains that will alter these spacings [92]. This principle is depicted schematically in Figure

3.2. Changes in lattice parameters, and thus the strain, can be obtained from a single Braggs

measurement [93, 94]. If the elastic constants of the material are known, the corresponding stress

can be calculated. In practice, the sin2 technique is more commonly used. This technique involves

measurement of the shift in a peak as a function of the angle of the specimen (Figure 3.2).

Diffraction methods have the advantage of being non-destructive. On the other hand, they have a

limited penetration depth and are sensitive to sample geometry and surface roughness, which can

present a problem for sprayed coatings, where the surface roughness is often large [92]. Most

diffraction techniques can only give information about the stress state near the surface. To obtain a

through-thickness profile of the residual stress, successive layers need to be removed. This

procedure will itself induce further residual stresses and affect the measurement. Furthermore, the

non-destructive nature of the technique is lost.

Figure 3.2 Schematic diagram of the XRD technique for measurement of residual stresses [95].

The most common diffraction techniques used for the purpose of measuring residual stresses are

X-ray diffraction, synchrotron X-ray diffraction and neutron diffraction [96]. Penetration depths for

X-ray diffraction are normally in the range 10-40 m [97] which is often in the order of the surface

roughness of the coating. This introduces systematic errors in the measurements. Synchrotron

normal to surface

and bisector

180o-2

180o-2

normal to

surface

bisector


24

X-ray diffraction has higher beam intensity and lower divergence, facilitating through-thickness

profiling [97, 98]. However, penetration depths still remain in the order of 60 m. Neutron

diffraction allows much greater penetration depths, and so is more suitable for stress profiling of

thick samples [92, 99]. Scardi et al [77, 78, 100] have investigated all three aforementioned

diffraction techniques for evaluation of the residual stress state in sprayed coatings. Their work

highlights the importance and usefulness of neutron diffraction as a technique for full through-

thickness of residual stress measurement.

3.1.3 Modelling of Residual Stresses

Several researchers have used numerical models to predict stresses caused by the spraying process

[101-103]. Modelling of residual stress generation is a complex task, and many parameters need to

be considered. Understanding of the mechanisms of stress generation is critical. Accurate

knowledge of the thermal and elastic properties of the substrate and the deposit is essential. These

properties are often temperature dependent. Accurate prediction of the parameters which define the

heat and mass flux received by the specimen is also required. Furthermore, the time dependent

nature of the thermal field during spraying has to be considered and included in the model. Finally

the effect of creep should also be incorporated.

Assumptions of temperature invariant elastic properties [89, 104] or use of bulk material properties

[83] will introduce large errors in the modelling of residual stresses. Moreover, ignoring the time-

dependent thermal profiles [105] and ignoring the effects of quenching stress [83, 106] in the

generation of residual stresses can lead to errors.

A model developed by Clyne, Gill and Tsui [103, 107, 108] was used in the present work for the

prediction of residual stresses. The basic features of this model are presented in this section.

The model is based on a finite difference formulation, using a fully implicit solution scheme. The

heat flux from the plasma gun and the mass flux from the droplets determine the temperature rise in

deposit and substrate. Convective heat flux from the plasma gun and droplet mass flux were

apportioned radially symmetric Gaussian distributions, taking into account the movement of the

spraying gun. The gun movement is specified and thus the gun position in relation to the selected

modelling point can be determined at any time. The deposition is modelled by thickening the top-

most element, as splats are added, and generating a new element when the thickness exceeds the

standard element thickness (see Figure 3.3). The interface between the coating and the substrate is

placed at an element boundary.


25

An equal biaxial stress-state with no through-thickness stresses, is assumed. Perfect bonding and

perfect heat conductance at the substrate-deposit interface are also assumed. Temperatures are

mapped into a mesh used for stress calculations. The temperature dependence of CTE, thermal

conductivity and specific heat capacity are taken into account. The mechanical behaviour for

metallic materials was assumed to be elastic-perfectly plastic, with temperature-dependent Young’s

modulus. The effect of creep is also simulated. For ceramic materials, the effect of microcracking

is simulated by introducing a critical stress, beyond which microcracking would inhibit further

increases in stress.

Curvature and in-plane stresses are predicted after each time step. For the calculation of stresses

and curvature, the concept of a relaxed (stress-free) element width is used (see Figure 3.3). The

complete stack of stress elements is allowed to expand or contract laterally in order to set the

summation of lateral forces to zero (force balance). For equilibrium conditions, the requirement of

a zero overall bending moment also needed to be met (moment balance). Therefore the resulting

curvature was calculated. The algorithm for the model is depicted in Figure 3.4.

Calculate gun position

Determine mass transfer and thickness increment

Is a new element required?

t = ti

t = ti +Δt

Mass Input

Figure 3.3 (a) Deposition process as simulated by the model [109], (b) Schematic representation of the calculation

of stress using the relaxed width method. Upon cooling from temperature T1 to T2, (I) substrate and coating contract

by differing amounts, due to the CTE mismatch (II). Substrate and coating elements are then constrained to have

the same width (III) (i.e. application of the force balance with no bending allowed). Finally a moment balance is

applied, such that the curvature with the least associated strain energy is adopted (IV) [76].

The numerical model also has the option of depositing additional coating layers over a coating

system which is already residually stressed, by using the files generated during the previous

deposition processes as starting files. Heat treatment of the sprayed specimen can also be simulated.

The starting point for this calculation is the generated file after the deposition process has been

(I)

(IV) (III)

(II)


26

completed. Heat treatment conditions can be isothermal, i.e. the whole sample being at any

arbitrary temperature, or a thermal gradient can be simulated. A thermal gradient is simulated by

using a stationary heat flux with no mass flux, to the top of the substrate. Cooling at the rear surface

is achieved by decreasing the ambient temperature or raising the rear face heat transfer coefficient.

In this way, a steady-state heat flux is established that sets up the desired thermal gradient across the

coating. This feature of the model can be utilized for the calculation of the thermal conductivity of a

produced coating. A thermal gradient through a coating can be established and monitored

experimentally throughout the thickness of the coating. By knowing the top surface and rear

temperatures, the thermal gradient can be simulated using the model. There are no arbitrary

adjustable parameters, except the quenching stress (which can be measured separately). Sintering

of the top coat under service conditions can raise its stiffness and hence reduce its strain tolerance.

Sintering mechanisms of the top coat are discussed in more detail in section 3.5. The numerical

model has been adapted to allow prediction of residual stress development incorporating the effects

of top coat sintering during heat treatment.


27

Figure 3.4 Flow chart illustrating model algorithm [76].


28

3.2 Mechanical Properties of TBCs

The mechanical behaviour of dense ceramics is well known. Ceramics typically have high tensile

strength and fail in a brittle fashion. Ceramics behave differently in tension and compression and

their compressive strength is usually much higher than their tensile strength. The presence of flaws,

such as pores or inclusions, is known to affect the mechanical behaviour of ceramics. The

mechanical properties of TBCs have been studied extensively by many researchers [60, 63, 110-

112]. In this section a brief overview of the mechanical behaviour of PS TBCs and the attempts to

predict the mechanical response are presented.

3.2.1 Mechanical Behaviour of PS TBCs

Literature values show substantial scatter for the elastic modulus of APS YSZ top coats, varying

from a few GPa to values close to those of dense ceramics. An overview of the different values

reported in the literature, and the method employed for the measurement, is given in Figure 3.5.

This is because the mechanical properties of PS deposits depend very strongly on their

microstructure, especially on the porosity and the interlamellar contacts. These features depend in

turn on the processing parameters and initial powder morphology [68].

The evaluation method employed to measure the elastic properties should always be specified.

Values of the effective Young’s modulus may differ considerably for the same coating using

different evaluation methods. Measurements performed by microindentation or nanoindentation

with sharp pyramid (Berkovich) tips give values approaching that of pore-free zirconia [60]. This is

due to the fact that with such techniques the region undergoing deformation is relatively small and

should be devoid of features such as splat boundaries and microcracks, which are responsible for

the low values obtained by other techniques in which greater regions are examined. Siebert et

al[113] carried out measurements using instrumented depth-sensing micro-indentation with a

Vickers diamond pyramid indenter and reported Young’s Modulus of as-sprayed YSZ in the range

of 94-146 GPa. Due to the relatively small area of the indentations, the influence of large pores and

microcracks on the Young’s modulus is not taken in consideration using this technique.

Nevertheless, increases in the Young’s modulus due to sintering of annealed TBCs were detected.

The changes were accompanied by a changing microstructure. The Knoop indentation method has

also been used to measure the elastic modulus of YSZ plasma sprayed coatings. The indenter in

this method is an elongated diamond pyramid. Leigh et al [114] reported a modulus of 44±6 GPa

for an as-sprayed YSZ coating using Knoop indentation, Zhu et al [112] about 70 GPa and Vasquez

et al [115] reported values in the range of 25-56 GPa. In spherical indentation the radius of the


29

indenter is usually large enough (2-3 mm) so that the indentation covers a suitable range of

deformed material with several representative microstructural features such pores, cracks and splats.

Wallace et al [111] reported values for the Young’s modulus in the range of 22-38 GPa for as-

sprayed YSZ coatings measured with spherical indentation and Eskner et al [63] a value of

38±4 GPa for a YSZ top coat.

0

20

40

60

80

100

120

140

160

180

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Youn

g's

Mo

dulu

s (

GP

a)

ca

ntile

ve

r[6

0]

sp

he

rical in

de

nta

tio

n [1

11

]

sp

he

rical in

de

nta

tio

n [6

3]

kn

oop

in

de

nta

tio

n [1

15

]

kn

oop

in

de

nta

tio

n[1

14

]

ten

sile

[116

]

kn

oop

in

de

nta

tio

n[1

12

]

mic

roin

de

nta

tio

n[1

13

]

na

noin

de

nta

tio

n[6

0]

mic

roin

de

nta

tio

n [1

17

]

mic

roin

de

nta

tio

n[1

17

]

ultra

son

ics[1

15

]

mic

roin

de

nta

tio

n [1

17

]

fou

r p

oin

t b

en

din

g [6

8]

Figure 3.5 Measured Young’s modulus data of as sprayed YSZ coatings from literature.

Methods such as cantilever bending [60] and four point bending [68], have been used to measure

macroscopically the elastic properties of the plasma sprayed coatings. The values for the elastic

modulus measured by these techniques are usually lower than those measured by indentation

techniques and arguably more representative for understanding the behaviour of these coating in

service. Thompson et al [60] measured the Young’s modulus of as-sprayed detached top coats by


30

cantilever bending at 10±2 GPa. Schwingel [68] reported values in the range of 2-20 GPa for as-

sprayed YSZ top coats measured by four point bending.

PS coatings have different Young’s moduli along the directions parallel and perpendicular to the

surface, with higher moduli observed in the direction perpendicular to the surface [114, 117, 118].

Duan et al [117] reported a Young’s modulus of attached YSZ coatings in the direction parallel to

the substrate of 30 GPa measured by microindentation whereas in the cross section the Young’s

modulus was 61 GPa. This is attributed to the relative amount of total surface area of pores and

cracks. The surface area of the pores which are aligned parallel to the substrate is greater

compared to the area of the cracks [119].

Plasma sprayed TBCs behave differently in tension and compression (see Figure 3.6). In tension,

the microcracks will open and some splat shear is likely to occur. Under compression the

microcracks will tend to close, and at sufficiently high compressive strains, they will be fully closed.

To further deform the specimen much higher stresses will be required [62, 120, 121]. For a typical

PS PSZ top coat the microcrack width is in the order of 0.1-0.2 μm, while the lateral separation of

microcracks is about 10 μm. Hence, it might be deduced that strain in the order of 1-2% would be

required in order to close the microcrack [60, 122]. At higher strains, microcrack closure would

occur and an increase in stiffness is expected.

Figure 3.6 Schematic representation of the variation of coating stiffness with imposed strain (after Thompson

[60]). In tension, (a) opening of microcracks results in a low Young’s modulus. (b) at low compressive loads,

the same is true (c) at higher compressive loads an increase in stiffness is expected, due to closing of

microcracks.

Several researchers [109, 120, 121, 123, 124] have observed that PS coatings exhibit non-linear

load-strain curves under either tension or compression, resulting in an appreciable to moderate

hysteresis. The degree of non-linearity appears to be less in tension than in compression, and the


31

hysterisis remained almost unchanged after a few cycles. This irreversible behaviour can be

explained in terms of the microstructure. Microcrack propagation, mechanical interlocking or

loosening and splat sliding may take place. Due to this non-linearity, different approaches have

been taken in order to rationalize the elastic properties. Choi et al [123], introduced an

“instantaneous” elastic modulus, whereas, Harok et al [121] used a method of partial unloading,

which gave values comparable ultrasonic tests. Hysterisis is less marked in samples that have been

heat treated [109, 123].

Heat treatment of PS YSZ top coat can result in sintering, particularly in the form of changes in the

bonding between splats and healing of microcracks. These changes strongly affect the mechanical

properties of PS YSZ TBCs [60, 111]. Sierbert [113] observed a rise in a Young’s modulus after

annealing for only 2 h at 1100oC from 94 GPa to 144 GPa and reaching 177 GPa after 100 h.

Similarly, Eskner et al [63] reported an increase from 38 to 68 GPa after 150 h at a temperatures as

low as 1000oC measured by spherical indentation. Zhu et al [112] investigated the evolution of the

Young’s modulus under high heat flux conditions, where the top coat was under a thermal gradient

with the surface at about 1100oC. The Young’s modulus increased after 120 h from about 70 GPa

to the final value of 125 GPa.

The effect of heat treatment on stiffness has been investigated by Thompson and Clyne [60]. The

rise in stiffness can be divided in two regimes, a rapid initial increase, followed by a more

progressive rise (see Figure 3.7a). Microstructural observations revealed the mechanisms

governing this behaviour. At short heat treatment times, there was evidence of grain growth and

bridging between splat boundaries (stage I stiffening). These changes cause an appreciable change

in stiffness. Aging for longer times, causes healing of the microcracks (stage II stiffening), which is

particularly significant at temperatures above 1100oC. The two stages of the stiffening process are

shown schematically in Figure 3.8. More details on the sintering of YSZ plasma sprayed top coat

and its effect on the microstructure is given is section 3.5.

Sintering and stiffening behaviour of the top coat is different when it remains attached to the

substrate. The higher CTE of the substrate results in the top coat being under tension during high

temperature heat treatment [60]. Stiffness data obtained for heat treated top coats while attached to

the substrate [60] showed that the presence of the substrate reduced top-coat stiffening (see Figure

3.7b). In fact, at lower temperatures and for short times at higher temperatures, the Young’s

modulus apparently decreased. The tensile stress state in the top coat during heat treatment opens

up microcracks and inhibits the sintering process. At higher temperatures and for longer times,

owing to diffusional processes, sintering causes the stiffness to rise.


32

(a)

0

10

20

30

40

50

60

70

80

0 20 40 60 80 100 120 140Time (hours)

1400oC

1300oC

1200oC1100oC

1000oC

E (

GPa)

(b)

0

5

10

15

20

0 20 40 60 80 100

E (

GP

a)

Time (hours)

1100oC

1200oC

1300oC

1000oC

Figure 3.7 Young’s modulus data obtained by applying the cantilever beam bending test to detached YSZ top coats subjected

to various prior heat treatments (a)while detached from the substrate (b) while still attached to the substrate [60].

Figure 3.8 Schematic diagram showing the microstructural defects and healing processes which result in the two

stage stiffening observed experimentally (after Thompson and Clyne [60]).

3.2.2 Prediction of Mechanical Properties of PS Materials

The mechanical properties of porous materials are different from those of dense materials and many

attempts have been made to model their mechanical response. Gibson and Ashby [125] modelled

the bulk elastic properties of a porous material, considering it to be composed of an idealised open

foam cell. Mackenzie [126] derived an expression for the Young’s modulus of a porous material,


33

relating it to the Young’s modulus of the dense material and the porosity fraction, for a material

with homogenous and isotropic elastic properties and isolated, randomly distributed spherical pores.

The Eshelby model [127] addressed the problem for ellipsoidal inclusions. Based on the Eshelby

method, many researchers have modified the approach and assumed zero stiffness for the inclusions

in order to understand the mechanical response of thermal barrier coatings [128 Wang, 2003

#10277].

Leigh and co-workers [128] used a set of explicit constitutive equations for the effective elastic

constants of transversely isotropic porous materials and developed them by modelling the pores as

unidirectional aligned and two-dimensional randomly oriented ellipsoids. They obtained

quantitative microstructural data from stereological analysis, which was used to calculate the elastic

constants based on the model. Their results highlighted the strong dependence of elastic constants

on microstructural features such as void concentration, void aspect ratio and crack density.

Experimental data lay below the model predictions.

Sevostianov and Kachanov [129, 130] followed a similar approach and investigated the effect of the

crack-like shapes of the observed pores in plasma sprayed deposits. Their work indicated that crack

parameters such as crack densities and orientation play a more important role in determining the

effective elastic properties than does the overall porosity. Their work also directly related the

scatter in crack orientation to the anisotropic properties exhibit by PS coatings. However, in

practice it can be difficult to obtain reliable values for the crack density and its orientation in plasma

sprayed deposits. Kulkarni and co-workers [119, 131] used small angle neutron scattering (SANS)

in order to acquire such quantitative microstructural information of the microstructure of plasma

sprayed deposits produced from different feedstock materials. The microstructural data extracted

from SANS was incorporated in a finite element analysis and used to predict the elastic properties

of the different coatings. Their results were compared with experimental data obtained by

microindentation [119, 131]. Experimental results were lower than the predicted values and this

was attributed to the presence of splat boundaries in plasma sprayed deposits, which was not

included in the model and have an important influence on the overall modulus of the coating [132].

Generally, there is a great interest in quantitatively characterizing the microstructural features of PS

coatings and being able to relate them to their effective properties, and different techniques and

models have been investigated [133-137]. There is particular interest in the evolution of the

microstructure during sintering [138, 139]. More detail on the sintering of YSZ plasma sprayed top

coats and its effect on the microstructure is given is section 3.5.


34

3.3 The Thermally Grown Oxide

Upon exposure to high temperature, an oxide layer develops at the interface between the bond coat

and the top coat, due to bond coat oxidation. The oxide layer is usually referred to as a “thermally

grown oxide” (TGO) and consists mainly of -alumina. Alumina-forming alloys are preferred due

to the superior adherence (stability) of alumina and its low oxygen diffusivity. The development of

the TGO is thought to play a critical role in the durability of TBCs.

The growth rate of the TGO is essentially parabolic and depends on the oxygen activity, as well as

on time and temperature [140]. The growth is controlled mainly by the inward ingression of

oxygen anions, rather than the outward diffusion of cations. Ingression of oxygen towards the top

coat/bond coat interface can occur by two mechanisms, namely oxygen gaseous permeation through

the network or interconnected cracks and voids of the TC or oxygen diffusion thorough the lattice.

Fox [21] has shown that, over the range of operating temperatures of TBCs, oxygen transport

through the top coat by permeation is expected to dominate. However, the rate of TGO thickening is

largely controlled by the oxygen diffusion through the TGO itself.

A typical microstructure of a TGO layer formed in a MCrAlY-ZrO2 system is depicted

schematically in Figure 3.9. The main oxide formed is α-alumina. There are two zones, a columnar

zone and an equiaxed zone. In the columnar zone close to the substrate intergranular pores,

dislocation pile ups and low angle grain boundaries are present. In addition, very fine spinels and

chromium oxides are detected in this region. The thin chromia layer and spinels near the BC/TGO

interface probably formed as a result of aluminium depletion in the BC. Oxides other than alumina

and spinels are also present in the equiaxed zone near the TC. These oxides probably are formed

during the early stages of oxidation and are pushed back as the growth of alumina progresses.

Yttrium segregation to the grain boundaries is also observed. Microstructural features such as

spinels and transition oxides, segregation of species to the grain boundaries, can embrittle the

interface region. Therefore, the energy required to cause spallation of the coating decreases.

The TGO layer is typically only a few microns in thickness. However, there are thought to be large

residual stresses associated with the TGO layer. Firstly, differential thermal contraction between the

TGO and the underlying substrate can give rise to large compressive stresses in the TGO. These

stresses have been estimated to be of the order of 3-6 GPa [141]. Also, stresses arise due to the

volume changes associated with phase transformations occurring during the creation of the TGO

(~1 GPa).


35

Figure 3.9 Schematic of thermally grown oxide structure [140].

According to Evans et al [23], the sequence of events that causes the failure of TBCs related to

stresses in the TGO are: (i) stresses develop in the TGO upon cooling due to differential thermal

contraction. These stresses are relieved either by buckling of the TGO or by visco-elastic

deformation of the bond coat. The relaxation processes introduce imperfections. (ii) Stresses normal

to the interface develop around imperfections and cracks are nucleated. Coalescence of these cracks

eventually leads to failure by delamination. On the other hand Tolpygo et al [142, 143], have found

that average and residual stresses in the TGO do not change significantly during thermal cycling

prior to failure. Similar results have been reported by Clarke et al [144].

Typical thicknesses of TGO are 3-10 m and failure commonly occurs when thickness of the TGO

is approximately 6-8m. Several approaches to determine the lifetime of TBCs use the time it takes

for the TGO to reach a critical thickness value as the upper limit for its durability.

3.4 Creep of Bond Coat and Thermally Grown Oxide

Creep is defined as the time-dependent deformation of a material under constant load. Creep in

polycrystalline materials occurs as a result of the motion of dislocations within grains, grain

boundary sliding and diffusion processes. Generally, creep becomes significant at temperatures

above about 0.6Tm and at these temperatures materials will creep at stresses much lower than their

yield stress. Thus, at high temperatures creep often governs the stability of materials. Creep of the


36

bond coat and the TGO is thought to play a role in the failure of TBCs. In this section the effect of

bond coat and TGO creep on the stresses in the TBC system will be discussed.

3.4.1 Creep in the Bond coat

Temperatures during typical service conditions of TBCs are above 0.6Tm of the bond coat materials

used commonly. Hence, stresses in the bond coat will relax due to creep with time at high

temperature. It has been suggested that at temperatures above 800oC the bond coat becomes

essentially non-load-bearing after short times [34]. On cooling down from elevated temperatures,

stresses regenerate due to differential thermal contraction and stress levels in the bond coat might be

higher than originally, especially at high cooling rates [145]. This is because at high cooling rates,

typical of those experienced under service conditions, a limited degree of creep relaxation can take

place.

The creep strain rates of two bond coat materials, namely CoNiCrAlY and NiCrAlY have been

measured experimentally by Thompson et al [28]. The measured steady state creep rates conform

to power law expressions, with respective stress exponents of about 2.9 and 4.5. The effect of creep

of the aforementioned bond coats on their residual stress state after heat treatment of a TBC system

was predicted using a numerical model. At temperatures above 700oC stresses in the bond coat

relax with time due to creep. On returning to room temperature, at high cooling rates, enhanced

stress levels existing in the bond coat. The level of the stresses depends on the relative values of

expansivity, creep parameters and modulus of the bond coats. This larger stress is likely to enhance

the initiation of microcracks and debonding of the TBC leading to spallation [145].

3.4.2 Creep of the TGO

The high melting point of alumina, the main constituent of the TGO, makes creep of TGO less

significant than bond coat creep. However, creep relaxation may take place in the oxide during

prolonged heat treatment or thermal cycling at high temperatures and several researchers have

looked into TGO creep and its effect on TBC failure.

A conclusive argument on the effects of TGO creep has not been established. Rösler et al [30]

maintain that creep deformation of the TGO scale can significantly reduce growth stresses and

therefore, relaxation by creep will improve TBC life. However, it is possible that creep at high

temperatures could lead to higher stresses on subsequent cooling. On the other hand, Evans et al

[29], maintain that creep of the alumina oxide layer has no significant effect on interfacial crack

propagation rates, but that creep in the underlying NiCrAlY coating, on contrary has a significant


37

effect in retarding wedge crack along the interface with the oxide. Tolpygo and Clarke [143, 146]

relate the importance of creep relaxation in the oxide to substrate thickness and maintain that TGO

creep is major relaxation process on thick substrates, whereas in thin substrates metal deformation

predominates. In all cases the effect of cooling rate on TGO creep is evidently significant, with

higher cooling rates preventing creep relaxation. Residual stresses in the oxide become smaller

with decreasing rate of cooling after oxidation, due to relaxation by creep during cooling.

3.5 Sintering of the Top Coat

The microstructure of a typical PS TBC incorporates many through-thickness microcracks,

interlamellar pores, globular voids and other defects. As a results of these, top coats have low

stiffness and hence high strain tolerance. This high strain tolerance enables the top coat to

withstand the thermal strains imposed during thermal cycling of the TBCs. However, at the

temperatures experienced under-service conditions, defects and microcracks might heal as a result

of sintering, reducing the strain tolerance and increasing the driving force for top coat spallation.

Additionally, the thermal conductivity of the top coat will tend to increase. In this section, a brief

overview of sintering effects in the top coat is presented.

3.5.1 Sintering mechanisms in ceramics

3.5.1.1 Solid State Sintering

The definition of sintering follows: sintering is a thermal treatment for bonding particles into a

coherent, predominantly solid structure via mass transport events that often occur at the atomic

scale. The bonding leads to improved strength and lower system energy [147]. The driving force

for sintering is the reduction of free surface energy and interfacial energy acting over curved

surfaces. For two adjacent particles the free surface is associated with high energy and particles

sinter by atomic level events to eliminate free surface energy. Surfaces with high curvatures have

high surface area per unit volume and mass flows towards regions of lower curvature; hence

sintering tends to eliminate surface curvature. Reduction of free surface energy occurs also by

removal or coalescence of pores, consequently densification and coarsening often occur during

sintering [148]. Similarly, grain boundaries have an energy associated with them and grain growth

during heat treatment reduces the grain boundary energy and area [149].

In response to the driving forces for sintering there is mass transport. There are several possible

mechanisms of mass transport. Possible mass transport mechanisms of sintering are summarized in

Table 3.2 and illustrated in Figure 3.10 [150]. The arrows indicate the path along which the material


38

flows for each mechanism. These mechanisms are separated mainly in two classes, surface

transport and bulk transport.

Figure 3.10. Schematic representation of the possible mechanisms of sintering (after Ashby [150]).

Possible sintering mechanisms

Mechanism Number

Transport path Source of matter Sink of Matter

1 Surface diffusion Surface Neck

2 Volume/lattice diffusion Surface Neck

3 Vapour diffusion Surface Neck

4 Grain boundary diffusion Grain boundary Neck

5 Volume/lattice diffusion Grain boundary Neck

6 Plastic flow/lattice diffusion/ Dislocation Neck

Table 3.2 Possible diffusion mechanisms during sintering [147, 150].

Surface transport processes produce neck growth without densification or shrinkage, the source and

sink of mass in these processes being the particle surface. These mechanisms are surface diffusion

(1), volume/lattice diffusion from the particle surface (2), and vapor diffusion (3) [147, 150].

Surface diffusion involves the motion of atoms from and to surface defect sites. Surface diffusion

is thermally activated and usually the activation energy is less than other mass transport

mechanisms. Therefore, surface diffusion initiates at lower temperatures in comparison with other

mass transport mechanisms and it is an initial contributor to the sintering process. Volume

diffusion from the particle surface, through the particle interior and subsequent emergence at the

particle surface is effectively a transport path form a surface source to a surface site. However this

diffusion path does not occur at very significant levels [147]. During vapor transport atoms at the

surface evaporate and condensate on a nearby surface. This process usually has a very small

contribution, however it can be significant for materials with high vapor pressures or those that

form volatile species.


39

Bulk transport processes include grain boundary diffusion (4), volume/diffusion (5) and plastic flow

or lattice diffusion through dislocations (6) [147, 150]. These mass transport processes result in

densification and shrinkage and in general are active at higher temperatures. Grain boundary

diffusion is usually an important mechanism in the densification of most material systems. Grain

boundaries, due to their defective character provide a low energy path for diffusion. Mass is

removed from the grain boundary and deposited at sinter neck. The activation energy for grain

boundary diffusion is usually intermediate between that for surface and volume diffusion. When the

grain boundary has high energy, sintering may be inhibited [151, 152]. Grain boundary energy

might be lowered by impurities or additives that segregate at grain boundaries [153, 154]. If

sintering is grain boundary diffusion-controlled, most typically active species will segregate at grain

boundaries [155]. Volume diffusion involves the motion of atoms from the interparticle grain

boundary through the crystalline structure, to the neck surface, or a reverse vacancy flow, with

subsequent shrinkage and densification. The volume diffusion rate depends on temperature,

composition and curvature. Essentially, the vacancy population of the polycrystalline material,

which depends on temperature, determines the volume diffusion contribution to sintering. For ionic

materials, stoichiometry is also a controlling parameter [156]. The vacancy concentration in ionic

materials will be affected by their stoichiometry. The total contribution of volume diffusion is then

determined by the combined action of the thermally induced vacancies and vacancies induced as a

result of changes in stoichiometry. Finally, plastic flow or lattice diffusion by dislocation climb can

also contribute to sintering and densification. Dislocations essentially interact with vacancies and

improve mass transport. This transport process is most likely important during heating and their

role decreases as dislocations are annealed out.

It is worth noting that actual sintering involves a mixture and cooperative occurrence of the above

processes described. During the different stages of sintering different mechanisms shift in

dominance.

Solid state sintering can be divided into three stages. The first stage involves particles approaching

one another and the formation of a sinter bond or neck at the point of contact. In this stage, the

number and size of necks between individual particles is sufficiently small to consider that there is

no interaction between neighboring necks [157, 158]. The second or intermediate stage of sintering

is characterized by great shrinkage and densification, pore rounding and grain growth [159]. As

sintering progresses grain growth becomes increasingly active. By the end of the intermediate stage,

no open porosity remains in the structure, all remaining pores are closed. The final stage of

sintering is a slow process [159]. It involves the collapse of the pore structure, pore spheroidization,


40

simultaneous coarsening and densification. Pores become isolated and move away from grain

boundaries. Once pores become spherical and isolated, further densification is difficult. Various

additives such as second phase particles and segregants can slow grain growth and assist

densification during the final stage sintering.

Numerous models have been developed in order to describe the sintering process. Most of them

describe the sintering of spherical particles of the same size. Ashby [150] developed sintering

diagrams which identify, at a given temperature, particle size and neck size, the dominant

mechanism and show the rate of sintering produced by all the mechanisms acting together. The

initial stage of sintering has been the focus of many of the sintering models [147, 150, 151, 160-

163]. For spherical particles, equal in size, the initial stages of sintering can be described in a

simplified form, by relating the neck size to particle ratio to the sintering time by the following

equation:

1

n

m

B tr

R R

(3.5)

where r is the neck radius, R is the particle radius, t is the sintering time, B1 is a constant, and n and

m are constants that depend on the mass transport mechanism. The constant B1 is a constant that

collects material and geometric constants and depends on the sintering mechanism. Depending on

the assumption used, different values are assigned to the exponents [164]. As mentioned earlier the

initial stage of sintering occurs mainly through surface diffusion and the diffusion coefficient

follows an Arrhenius temperature dependence embedded in the parameter B1. The above equation,

even though not very exact, is a good approximation for neck size ratios up to 0.3. In addition it

illustrates the relative importance of the parameters affecting the initial stage of sintering, such as

the high sensitivity to the inverse of particle size and exponential temperature dependence in

comparison with the relatively small effect of time. Kingery and Berg [160] considered the

approach of the centers of spherical particles by bulk transport mechanism and derived an equation

that related the linear shrinkage to the sintering time t, in a relationship of the following type:

2

n

o

LB t

L

(3.6)

where ΔL is the length change of a material with initial length Lo, and the parameter B2 includes

physical constants and material parameters such as surface energy, volume diffusion coefficient,

atomic size, etc. The parameter B2 is also exponentially dependant on temperature. The exponent n

takes a value of 2/5 for spherical particles. This method is an attempt to separate the mechanisms of


41

surface mass transport to bulk mass transport in the initial stages of sintering and is valid only for

small shrinkages (3%). Due to the many simplifying assumptions this equation is not always

applicable to the sintering of real powders. Other efforts have also been made in literature for

quantitative descriptions of shrinkage occurring during the sintering process. A critical review of

these models is given by Exner [163, 165].

The intermediate process comprises of densification, grain and pore growth. Due to complexity of

these phenomena, there are very few attempts to model the process. Coble [159] first proposed a

microstructure model based on a highly-interconnected pore structure network. The microstructure

comprises of grains in the shape of tetrakaidecahedron with cylindrical pores occupying the grain

edges. The pore network is fully interconnected. With this model a relation between the remaining

porosity in the material and the sintering time was derived:

3 lno

o

tp p B

t (3.7)

where po and to are the porosity and time at the onset of the second stage, t is the isothermal

sintering time (greater than to) and p is the porosity at sintering time t. The parameter B3 includes

physical constants of diffusional sintering and grain growth such as surface energy, diffusivity,

atomic volume, curvature and Boltzmann’s constant and is also exponentially dependant on

temperature The above equation is often termed the “shrinkage law”. Coble further developed his

model for intermediate stage sintering and modified it to apply it to final stage sintering as well. He

assumed spherical pores located at grain corners and related the porosity to the pore diameter and

the tetrakaidecahedron grain edge length. Shi [148] proposed a microstructure pore model for

intermediate and final stage sintering based on spherical pores. He derived densification equations

relating density and densification rate to pore size and grain size ratio. The derived equations can

also be used to explain the effect of pore size distribution and the agglomeration properties. Harmer

et al [166, 167] developed sintering diagrams which highlight the conditions where pores remain

attached to grain boundaries during final-stage sintering.

Particle size has a significant effect on the sintering. During the initial stage of sintering, the

significance of particle size becomes apparent from the neck growth equations for each transport

mechanism, which are proportional to a power of the inverse particle size. Some mechanisms of

mass transport are more sensitive to particle size than others. Surface and grain boundary diffusion

are very sensitive and are favored by smaller particles sizes because of the higher surface area

content per unit volume. Volume and vapor diffusion are less sensitive to particle size changes.


42

During intermediate and final stage sintering, there is a similar dependence on particle and pore size.

Small particles and small pores have a higher energy per unit volume, more available surface area,

and higher curvature, all of which contribute to higher sintering rates. Herring [168] was among the

first to report this effect and he established a scaling law, where he correlated the degree of sintering

of different sized powders to the sintering time. For smaller sized powders, less sintering time is

needed to reach a certain amount of sintering. However, is worth noting that despite enhanced

sintering of powders with smaller particle sizes, nanoscale powders tend to form agglomerates

which prove difficult to sinter [169, 170]. In general, large pores with lower curvature and higher

dihedral angle are difficult to sinter [149]. Particle size distribution also plays a role in sintering. A

wide particle size distribution results in inhomogeneities during packing, which can result in a

coarser, more porous microstructure at long sintering times [171]. Another factor that affects

sintering of compacts is the particle packing, or green compact density. Small packing densities

and packing inhomogeneities result in sintered dense regions surrounded by enlarged pores. Higher

packing densities result in more initial points of contact between particles, smaller initial pores, and

higher final densities. In general, there are many factors that affect the solid state sintering and

careful consideration of these is necessary in order to be able to control the process.

3.5.1.2 Liquid Phase Sintering

Sintering in the presence of a liquid leads to densification by quite a different process. Liquid phase

sintering occurs when a liquid is present at the sintering temperature. There are three different

stages in the densification process after the liquid is formed, illustrated schematically in Figure 3.11.

Stage 0 represents the preliminary stage, where the reactive liquid is formed. In many systems prior

to the formation of the liquid considerable densification occurs by solid state sintering. On

formation of the liquid phase, rearrangement of the liquid and solid occurs due to capillary forces

(stage I). The rearrangement of particles proceeds in the direction of reducing porosity. If the

volume of the liquid phase is sufficient, complete densification can take place in this stage. On the

other hand, at low liquid contents the solid skeleton inhibits densification, requiring the

participation of solution-reprecipitation events for further densification. At the end of the

rearrangement stage, further densification occurs by a solution-reprecipitation process (stage II),

where mass transport through the liquid controls densification. Material dissolves at the contact

points between particles and diffuses out through the liquid film between the particles. This process

causes particle centres to move closer together. The final stage of liquid-phase sintering is

controlled by grain growth and densification of the solid skeleton [172]. Other processes such as


43

coalescence of grains and pores, diffusion of the liquid components into the solid and phase

transformations may also place.

The temperature of liquid-phase formation and the liquid composition can be determined from the

phase diagram for a specific system of materials. A decrease in the liquidus and solidus

temperatures with alloying indicates a tendency for solute segregation. Solute segregation is greater

for greater separation of the solidus and liquidus curves. Impurities often show up preferentially at

grain boundaries and cause the formation of an inadvertent liquid film [173].

One key factor in liquid-phase sintering is the solubility of the solid in the liquid. For high

densification, the solid solubility in the liquid is high, but the solubility of the liquid in the solid is

low to avoid formation of a transient liquid. In addition, good wetting between the solid particles

and the liquid is required [147, 174, 175].

0

I

II

III

Figure 3.11 Schematic diagram of the stages of liquid phase sintering, 0: melting; I: rearrangement; II: solution-

reprecipitation; III: pore closure (after [147]).

3.5.2 Sintering of Zirconia based top coats

The sintering behaviour of commercial yttria-stabilized zirconia powders has been investigated by

various researchers and a wide range of sintering additives that enhance sintering have been

explored. It is clear that such additives should be avoided for TBCs made of zirconia-based

materials.

The mechanisms controlling sintering kinetics in stabilized zirconia are disputed in literature. Some

authors [169, 176] reported initial stage sintering kinetics to be dominated by grain boundary and

surface diffusion whereas other authors support that volume diffusion is the dominant mechanism

[177-179]. The rate of volume diffusion is determined by the species diffusing most slowly through


44

the lattice, which in zirconia is the cation [180-182]. The mechanism of cation diffusion in

stabilized zirconia is also disputed in literature [180, 183, 184].

The addition of up to 6 mol% Y2O3 and up to 12 mol% CaO stabilizers in zirconia powder

compacts has been reported to enhance densification [185]. Segregation of Y+3

, La+3

, Ca+2

and Ce+4

solute dopants at the grain boundary in zirconia has been observed by various researchers [186-190].

Segregation of solid solution additives at the grain boundaries lowers grain boundary energy and

aids shrinkage of pores by grain boundary diffusion [149, 155, 191]. Larger, low valence cations

appear to be more effective in reducing the grain growth, in agreement with the solute drag

models[153, 154].

The sintering rate of stabilized zirconia is strongly dependant upon impurity content. MgO [185],

Bi2O3 [192], Fe2O3 [192], SiO2 [193] CuO [194] and Al2O3 [195, 196] have been shown to promote

sintering in stabilised zirconia powder compacts when added in small amounts . The most common

mechanisms involved are segregation of the impurity at the grain boundary, formation of a second

phase at the grain boundaries, or formation of a liquid phase, which migrates to the grain boundaries.

De Souza et al [173], observed the development of a silicate glassy phase during sintering of

isostatically pressed powder discs with 0.5 wt % Silica. The presence of this phase is thought to

promote liquid phase sintering. Similar results have been reported for the influence of impurities,

especially SiO2, in PS coatings. Eaton and Novak [197] reported a fivefold increase in shrinkage on

going from 0.2 wt % to 4 wt % silica. This was attributed to the presence of a silicate glassy phase

that promoted liquid phase sintering. An increase in shrinkage with increasing silica content (in the

range of 0.02-0.12 wt %) was also reported by Vassen et al [198]. It is important to note that ranges

of impurity content in the studies mentioned above are significantly different. Work conducted by

Klocker et al [199] indicates that, for the impurity levels encountered in baseline YPSZ powders

(0.3-0.4 wt%), there is no apparent effect on the sintering behaviour. In addition to impurity levels,

non-stoichiometry also affects sintering. Higher sintering rates have been observed by Zhu et al in

Ar + 5%H2 atmosphere compared to air, due to enhanced cation interstitial diffusion under reducing

conditions [182].

The effect of particle size and temperature on the densification rate was studied in detail by Rhodes

[169]. He confirmed that using fine powders, free of agglomerates enhanced sintering kinetics.

Typically, PSZ compacted powders exhibit relatively rapid sintering at temperatures between

1100-1300oC [200], depending on the powder composition and morphology. Sintering effects in PS

TCs appear to be significant at temperatures above about 1000oC [197]. As a result of sintering,

there can be significant increase in the stiffness and thermal conductivity of the top coats [60, 112].


45

Ilavsky et al [201] reported a significant decrease in the surface area voids (up to 33%) in PS YSZ

top coats as a result of sintering, at temperatures as low as 1000oC. Cracks appear to sinter first, at

lower temperatures, followed by sintering of intersplat pores at higher temperatures.

3.6 Phase changes in the Top Coat

As mentioned is section 2.2.3, zirconia exists as three solid phases that are stable at different

temperatures [31]. At temperatures up to 1200oC, the monoclinic phase (m) is stable. Zirconia

transforms from the monoclinic to the tetragonal phase (t) above 1200oC and above 2370

oC to the

cubic phase (c). Transformation from m to the t phase has an associated volume decrease of 4%

[32]. A phase diagram for the zirconia-yttria system is presented in Figure 3.12a.

Zirconia-based TBCs are doped with trivalent or tetravalent cations, in order to stabilize the

tetragonal zirconia polymorph and prevent catastrophic cracking as a result of the volume changes

accompanying the t→m transformation, which occurs at temperatures within the range of the

working environment in gas turbines. Stabilization of the tetragonal or cubic zirconia phase by the

addition of trivalent dopants is believed to be associated with the generation of oxygen vacancies

[202]. The trivalent dopant cations substitute the Zr ion in the cation network and oxygen vacancies

are created for charge compensation. The tetragonal and cubic polymorphs are stabilized when the

doping level is sufficient to create enough oxygen vacancies to reduce the co-ordination number of

the Zr ion from 8 to around 7.5 [203, 204]. The Zr ion in fully stabilized and high temperature

cubic phase has a coordination number of 7, due to the strong covalent nature of the ZrO bond and

the small size of the Zr ion.

The tetragonality c/a is an important factor in the stability of the tetragonal phase and it has been

shown to be independent of the species of the dopant, but dependant on the content of the dopant

for oversized trivalent rare earths [205, 206]. Tetragonality decreases with increasing dopant

content for oversized trivalent cations and vanishes at 11 mol% M2O3 regardless of the ionic sizes,

since tetragonality is aided by the creation of oxygen vacancies only [207]. The decrease in

tetragonality is associated with increasing stability of the tetragonal phase. The lattice parameters

on the other hand of the tetragonal phases increase systematically with the ionic radii of the dopant

atoms [205, 206].

Tetravalent dopants such as ceria still stabilize the tetragonal and cubic phases of zirconia.

However, the stabilization is not the result of the generation of oxygen vacancies, as is the case with


46

trivalent dopants. The Ce4+

ion substitutes the Zr4+

ion in the cation network, without creating any

oxygen vacancies, since both ions are tetravalent [208, 209]. In this case, the stabilization effect is

thought to be caused by the slight dilation of the cation network by the oversized Ce4+

ion which

decreases the tetragonality and stabilizes the tetragonal phase to room temperature [208]. The

tetragonal lattice in zirconia has an inherent lattice strain caused by the smaller ionic size of the

zirconia atom. This strain can be relieved either by the presence of oxygen vacancies associated

with the Zr ions, as in the case of trivalent dopants, or in the case of tetravalent dopants by slightly

dilating the lattice by introducing a oversize ion. The tetragonality c/a of zirconia doped with

tetravalent cation decreases with increase in dopant content, as is the case with trivalent dopants.

However this effect is not so pronounced [208].

The conventional material used for the top coat in TBCs is ZrO2 stabilized with 7-9 wt% Y2O3.

Partially stabilised zirconia with 8 wt% yttria consists of non-transformable metastable t’ tetragonal

phase. A metastable tetragonal phase (t’) results from a diffusionless shear transformation directly

from the cubic state. During plasma spraying, the rapid cooling of molten splats to the substrate

temperature often results in the formation of t’ phase. This phase is thought to be stable up to

1400oC i.e. it does not transform directly to monoclinic on cooling to room temperature. The

resistance of the non-transformable tetragonal to transformation into the monoclinic phase is

ascribed to its smaller tetragonality. The “non-transformable” t’ tetragonal phase in PSZ with 6-

12 wt% yttria will decompose during annealing at elevated temperatures into the equilibrium high

temperature phases, cubic (c) and tetragonal (t) [38, 210, 211]. These phases will be retained at

room temperature, either as high-yttria and low-yttria t phases or as cubic and low-yttria t phase.

The rate of transformation is dependent on composition and temperature, and is slow at

temperatures below 1400oC. When annealing at temperatures above 1400

oC, the tetragonal phase

can transform to monoclinic upon cooling [212, 213]. The cooling rate has a significant effect on

the t→m transformation [214]. Slower cooling and heating rates will allow the t→m transformation

to take place (see Figure 3.12b). Furthermore, the size of the tetragonal precipitates also affects the

stability of the t phase. Larger precipitates transform spontaneously to the m phase, whereas below

a critical grain size the metastable phase remains stable [214]. Finally, residual stresses also have

an effect on the t→m transformation. It is thought that residual stresses generated during air

quenching can partially suppress the t→m transformation. On the other hand, stress fields

associated with defects can promote a stress-induced t→m transformation [215, 216].

The phase stability of PS top coats after prolonged heat treatment is an important issue. Different

stabilizers have been used to ensure that the tetragonal phase is retained. The basic criteria for the


47

selection of a suitable stabiliser include a suitable cation radius and a cubic crystal structure. Early

attempts used MgO to stabilize zirconia in its cubic state, by adding 25 wt% MgO [33]. However,

during heat treatment the zirconia reverts to its monoclinic form and the stabilizing oxide

precipitates out from solid solution, affecting the thermal conductivity [34]. Zirconia can be fully

stabilised to its cubic phase by adding 20 wt% yttria. However, such fully stabilised zirconia

coatings perform very poorly in thermal cycling tests [35].

CeO2-stabilized TBCs have been investigated by various researchers [36-38, 217, 218] and the

existence of the non-transformable t’ tetragonal phase similar to that in the YSZ has been agreed

upon. At high CeO2 weight percentages a second metastable cubic phase c’ also exists. In reducing

atmospheres the Ce4+

ion reduces to Ce3+

, which tends to stabilize the cubic phase [219].

Gadolinia is in general a less effective stabilizer than Yttria, although partial substitution of Gd for

Y does enhance phase stability, if Y remains the dominant species [220, 221]. ZrO2-Er2O3, ZrO2-

Sm2O3 and ZrO2-Nd2O3 PS TBCs have been reported to consist of non-transformable t’ tetragonal

phase up to compositions of 6 mol% of stabilizer, and above that, coatings are entirely cubic [39,

222, 223]. Erbia stabilized coatings showed better thermal phase stability than samaria-stabilized

and neodymia-stabilized coatings, which decomposed after heat treatment to the monoclinic and

tetragonal phases. Scandia-yttria stabilized TBCs have been reported to show better phase stability

than conventional YSZ TBCs up to temperatures of 1400oC [224, 225]. Dy2O3 and Yb2O3 rare-

earth dopants have also been reported to stabilize the tetragonal phase [40, 206, 226]. Zirconia

doped with lanthana has been reported to form the cubic pyrochlore-structured phase, especially at

high mol % of stabilizer [227-229]. The cubic pyrochlore structure is stable up to its melting point

and hence it is believed that these materials might have potential as TBCs [230]. However coatings

of this material yield shorter thermal fatigue cyclic life than YSZ due to their relatively low CTE

[231].


48

Figure 3.12 (a) Phase diagram for the zirconia-yttria system [232], (b) phase flow diagram showing behaviour of plasma-

sprayed 8 wt% Y2O3 PSZ on cooling after aging [214].

Chapter 4. Review of Thermal Conduction in TBCs

49

4 Review of Thermal Conduction in TBCs

Trends towards ever increasing turbine inlet temperatures and the need for coatings to protect hot

sections of the turbine have increased the demand for TBC materials with low thermal conductivity.

This review is concerned with briefly describing the thermal conduction mechanisms in ceramic

oxides and reviewing the methods for determining the thermal conductivity. Also, a review of the

thermal conductivity of TBC materials and various materials with a potential use as TBC is

presented as well as an existing microstructural model for the prediction of the thermal conductivity

of TBCs.

4.1 Thermal Conduction in Ceramics

Thermal conductivity is a measure of the rate of flow of thermal energy through a material in the

presence of a temperature gradient. In oxide ceramics thermal energy transport is due to lattice

vibrations (phonons) or electromagnetic waves (photons).

4.1.1 Phonon Conduction in Ceramics

The contribution to the thermal conductivity from phonon conduction, as derived by Debye, can be

expressed by the general form [174, 233] :

1

3ph v pk C l (4.1)

where Cv is the specific heat at constant volume, ρ the density, ν the mean phonon velocity

(approximately equal to the velocity of sound in the crystal) and lp the mean free path for scattering

of phonons. The mean free path of phonons is the distance they travel before being scattered.

Scattering of phonons in solids can occur by interaction with each other as well as by interaction

with defects, such as impurities, vacancies, grain boundaries and interfaces. Scattering of phonons

by direct interaction amongst themselves occurs by Umklapp-processes (sometimes called U-

processes), which is the name given to these interactions by Peierls [233].

At very low temperatures, phonon-phonon interactions are limited, and the mean free path is

determined by scattering from defects and grain boundaries. At very high temperatures U-processes

are very frequent and they are effectively the main source of thermal resistance. At intermediate

temperatures both U-processes and defect scattering are important.


50

The temperature dependence of phonon thermal conductivity in a non-metallic solid can be

predicted from equation 4.1. At very low temperatures, the mean free path is essentially constant

and the heat capacity Cv varies as T3, so the temperature dependence of the thermal conductivity

reflects the T3 behavior of the specific heat. At high temperatures, the specific heat is nearly

constant, and the mean free path is inversely proportional to the temperature, and hence the thermal

conductivity is proportional to 1/T. In the intermediate range of temperatures (since the incidence

of phonons rises with temperature), the thermal conductivity is determined by the intersection of the

behavior in the other temperature ranges, and is essentially represented by an exponential of 1/T

[234]. A typical conductivity curve for a non-metallic crystal is shown in Figure 4.1.

Figure 4.1 Temperature variation of phonon and photon conductivity for ceramic crystal [234].

In ceramic polycrystalline materials, the mean free path of the phonons, and hence the thermal

conductivity, may be influenced by the dimensions of the grains, the crystal structure, porosity or

impurities present. If the mean free path is of the same order of magnitude as the grain size, then

grain boundaries become active scattering centers and may reduce the mean free path of the energy

carriers and hence the thermal conductivity. For many ceramics, this is the case at low temperatures.

If the mean free path is generally smaller than the grain size, then the crystal structure is the major

influence, rather than the grain size. In this case, increased structural order results in an increase of

the mean free path and hence the thermal conductivity. The mean free path when there are a

number of independent scattering processes is given as [235]:

1 1

ip il l (4.2)

kph - phonon conductivity kr - photon conductivity kt - total conductivity

kr~T3lr

kt = kph + kr

kph~T3

kph~e-1/T

kph~T-1

The

rma

l C

on

du

ctivity

Temperature


51

where the subscript i denotes the various scattering processes operating. The above equation

suggests that the mean free path of a phonon is determined by the smallest of the mean free paths

due to the various independent scattering processes. In the case of substitutional atoms of mass

m+Δm in place of a normal atom of mass m, Klemens [235, 236] has shown that their contribution

to the mean free path is:

2 3 4

3

1

4pd

m aC

l m

(4.3)

where lpd is the mean free path due to scattering of phonons by point defects, C is the concentration

of point defects per atom, a3

is the atomic volume, ν is the mean phonon velocity and ω is the

frequency of the phonon lattice waves. Effective values of Δm/m can be obtained for other defects,

including vacancies. Porosity, grain boundaries and impurities also reduce the mean free path of

phonons and hence the thermal conductivity.

4.1.2 Radiative Heat Transfer in Ceramics

In addition to phonon conduction, heat may be transferred through solids by radiative heat transfer.

Radiative heat transfer is energy transport through transmission, or absorption and reradiation of

electromagnetic waves (photons). Radiative heat transfer can occur between two surfaces and it

does not require a heat transfer medium, it can occur in vacuum. In solid materials photons can be

reflected, absorbed and emitted by the material itself. The contribution to the thermal conductivity

by photon conduction can be expressed as [234]:

rr lTnk 32

3

16 (4.4)

where σ is the Stephen-Boltzmann’s constant, n is the refractive index, T is the absolute temperature

and lr the mean free path of photons or penetration depth. The refractive index for most oxides

usually varies very little with wavelength or temperature within the region of interest. The mean

free path of a photon is the average distance a photon travels in the material between scattering

events or before being absorbed by the medium. The total mean free path is equal to the reciprocal

of the extinction coefficient. The mean free path lr (or penetration depth) in equation (4.4) is given

by,

λlr

1 (4.5)


52

where λ = γ + κ is the extinction coefficient and is equal to the scattering (γ) plus the absorption (κ)

coefficients [237]. Generally, the extinction coefficient depends on the absorption and scattering

characteristics of the material, the wavelength distribution of the incident radiation and the

temperature of the material.

The scattering coefficient will depend on the microstructural features for polycrystalline materials.

If the radiation penetration depth is much greater than the grain size, then grain boundaries become

active scattering centers, as do pores [234, 238]. The radiation heat flux in this case will be uniform

through the material and the pores, but pores will change the scattering characteristics by decreasing

the penetration depth, relative to that of fully dense material. The dependence of the scattering

coefficient on porosity is thus expected to be quite strong, and porosity at a level of few % may

result in changes of an order of magnitude in the scattering coefficient [234, 239].

The total conductivity of oxide ceramic is the summation of phonon and photon conduction:

t ph rk k k (4.6)

In a perfectly transparent material or medium, the photon mean free path is infinite, and radiation

energy transfer occurs through the medium, without any interaction with the medium. In an opaque

material, the photon mean free path is extremely short, and all the radiation is absorbed and emitted

over extremely short distances. Therefore, the photon transport contribution to total energy transport

is small compared to the contribution from phonon conduction. If the photon mean free path in a

solid material is large compared to the phonon mean free path, as can be the case for

semitransparent materials, radiation conduction becomes significant. However, as can be seen from

eq.(4.4), the effect of temperature is more significant than the effect of photon mean free path on the

radiative heat transfer.

4.2 Methods for Measuring the Thermal Conductivity of Solids

4.2.1 Steady State Methods

Steady state methods for measuring the thermal conductivity generally consist of measuring a

temperature gradient and a heat flux across a sample and obtaining the thermal conductivity from:

Tq kA

x

(4.7)


53

where is q the is the input power, A is the cross-sectional area, ΔT is the temperature drop, Δx is the

distance over which the temperature drop occurs, and k (W m-1

K-1

) is the thermal conductivity.

The simplest steady state arrangement comprises a sample of uniform cross-sectional area A, to

which heat is supplied at one end, and removed at the other end, establishing a thermal gradient

across the length of the sample (linear flow method). An alternative is a sample which surrounds

the heat source either cylindrically or spherically, so the heat flow is radial. Further terms are then

needed in equation (4.7).

In the linear flow method, the sample is placed between a heat source and a heat sink, and so the

heat flow is linear and parallel to the sample axis and the isotherms are planar and parallel to the

cross-sectional area. The temperature difference ΔT=T2-T1 along the length Δx=x2-x1 is measured

using thermometers or thermocouples. Since k usually varies with temperature, the derived value of

k will be that corresponding to an average temperature between T2 and T1.

The steady-state linear flow method can only be applied if all the heat supplied by the heat source

travels though the sample to the colder end, with negligible heat losses from the periphery of the

sample. Heat losses are difficult to prevent, especially at higher temperatures. However, they can be

usually minimized, so that they become insignificant compared to the heat conducted through the

sample. Heat losses can be minimized by performing the measurement in vacuum and surrounding

the sample with insulation. In addition, the thermocouples should, if possible, fit tightly in the hole

and be small in cross-section, in order to minimize disruption of the heat flow. The choice of

specimen geometry, controlled mainly by the expected thermal conductivity value to be measured,

is also important. In order to ensure that the temperature difference ΔT is sufficiently large, the

ratio of length to area must high. However, relative lateral heat loses are proportional to the surface

area, so to minimize them, a short sample with a large cross section is required. In practice, long

bars are used for good conductors and thin discs for poor conductors.

At higher temperatures, heat losses due to radiation can become very significant and it may be

necessary to place a radiation shield around the sample. Ideally the shield should be heated

independently of the sample and have a temperature distribution similar to that along the sample.

However, this is difficult and in general, the use of this method is limited to temperatures where

radiative heat loss is not significant.

A standard method for the measurement of the thermal conductivity using steady state linear flow is

the guarded hot plate method. The equipment consists of a central transverse heater and two

similar samples either side of it. The samples and heater are sandwiched in between two heat sinks


54

and the whole assembly is surrounded by a material with low conductivity. The heater is

surrounded by an annular guard, which is independently heated to the same temperature as the

heater, in order to eliminate radial heat losses. Between the heater and the guard there is a small gap

(Figure 4.2). The thermal conductivity of the sample can be calculated using equation (4.7). This

method is particularly applied when measuring material with low thermal conductivity, and

available as thin samples with low length-to-diameter ratio.

Figure 4.2 Schematic diagram of guarded hot

plate apparatus for measurement of the

thermal conductivity [240].

4.2.2 Dynamic Methods

In “dynamic” or non-steady state methods, the temperature distribution varies with time.

The heat balance equation for transient conditions may be written as:

dt

dTρCTk p (4.8)

where Cp is the specific heat at constant pressure and ρ is the density [233, 240]. For homogenous

materials, the thermal conductivity k can be treated as a constant and equation (4.8) can be written

as

dt

dTρCTk p2

or

dt

dTTα 2 (4.9)

where α is the thermal diffusivity and is equal to:

p

k

C

(4.10)

The thermal diffusivity is a measure of the rate at which a material will approach thermal

equilibrium with its environment.

water-cooled heat sink

water-cooled heat sink

main heater guard guard

sample

sample

insulation


55

In dynamic methods, usually both the thermal diffusivity and the specific heat need to be

determined experimentally in order to derive the thermal conductivity. Accurate determination of

the specific heat becomes essential for many materials where the specific heat may vary with

temperature. On the other hand, measurement times in dynamic methods are generally short, so the

influence of heat losses on the measurement becomes less significant. In addition, heat losses can

be taken into account and be included as boundary conditions by incorporating a heat loss

coefficient in the heat balance equation. This heat loss coefficient can be eliminated by making

additional measurements under different experimental conditions. Hence, dynamic techniques are

particularly useful for measurements at higher temperatures. There are two main categories of

dynamic methods for measurement of the thermal conductivity: periodic or transient. In periodic

methods, a periodic thermal variation is imposed to the sample and certain propagation

characteristics of the resulting temperature wave are measured, whereas in transient methods a

single application of heat is supplied and the time for subsequent temperature changes in the sample

is measured.

Transient methods employ transient, non-periodic, heating to the sample, which is initially in

equilibrium with its surrounding environment. The change in thermal flux causes a temperature

change and the thermal diffusivity is evaluated by measuring the changes in temperature of the

sample with time. There are various methods widely employed for measurement of the thermal

conductivity. In general, these methods consist of generating a thermal flux either within the

sample, i.e with a suitable heater sandwiched between two identical samples, or at the surface of the

sample. The temperature distribution throughout the sample as a function of time depends on the

thermal diffusivity of the sample. In order to obtain the theoretical relationship between the sample

temperature distribution with time and the thermal diffusivity of the sample, the diffusion equation

must be solved, subject to the specific boundary conditions imposed by the experimental set up.

The most common transient methods used for thermal conductivity measurement are discussed

briefly below.

4.2.2.1 Transient Hot Strip

The transient hot-strip (THS) [241], transient hot-wire (THW)[242], transient plane-source (TPS)

[243] (or hot-disk) and dynamic plane source (DPS) [244] methods all apply the same principle for

determining the thermal properties of a liquid or solid material, in the sense that a single element is

used simultaneously as both a heat source and a temperature sensor. The experiment generally


56

involves passing an electrical current through the source/sensor, which results in a slight

temperature increase, and recording the voltage or resistance change over the source/sensor.

The THS method employs a thin metal strip as a planar heat source, placed between two identical

samples. A constant current is passed through the strip over a short period of time and the voltage

drop across the sensor over that time is recorded. By supplying a constant current, the output power

remains constant. The voltage change is due to the temperature increase which causes an increase in

the electrical resistance of the metal strip. Assuming an infinitely long planar heat source and no

influence to the temperature increase from the outer boundaries of the sample, the temperature

distribution with time in and around the metal strip can be expressed in terms of the thermal

diffusivity of the sample, the breadth of the heat plane source and the output power per unit area

[241]. The THS method has been applied to anisotropic solids [245], to insulating solids and

liquids [246] and to electrically conducting materials [247] .

4.2.2.2 Transient Plane Strip (Hot Disk)

In the TPS technique, the resistive element consists of a thin metal pattern shaped as a disk (Hot

Disk) or as a square (Hot Square) in-between two thin layers of the disk-shaped or square-shaped

insulator respectively (see Figure 4.3). As in the THS and THW techniques, the resistive element

acts as both the heat source and the temperature sensor. The sensor is sandwiched between two

identical samples. A current pulse is passed through the sensor, sufficient to cause a slight

temperature increase in the sensor and in the sample surrounding it. The length of the pulse is

normally chosen to be short enough so that the outer boundaries of the sample do not influence the

temperature increase of the element by any measurable extent. Hence, the TPS sensor can be

considered in contact with an infinite or semi-infinite solid throughout the transient recording. In

order to ensure that this assumption holds, the distance from the resistive element to the nearest

sample boundary must be equal or larger than the probing depth which is defined as [247]:

21

max )(αtc (4.11)

where c is a constant of the order of unity, α is the thermal diffusivity of the sample and tmax is the

total time of the transient experiment. Assuming that the resistivity of the metal element in the

sensor is temperature dependant, the temperature increase in the sensor will cause a change in its

resistance and a corresponding voltage variation over the sensor. Thus, the temperature increase of

the element can be deduced from recording its resistance or voltage variation over the period of the

transient event. The resistance of the sensor can be expressed by [243]:


57

( ) {1 [ ( )]}o i samR t R β T T τ (4.12)

where Ro is the resistance of the TPS element before the transient recording has been initiated and β

is the temperature coefficient of the electrical resistance of the sensor. ΔTi is the constant

temperature difference that develops almost momentarily between the sensor and the surrounding

sample due to the insulating layer around the sensor. This temperature difference will stay constant

after a short initialization time, Δti, due to the constant power supply (see Figure 4.4). ΔTsam(τ) is

the temperature increase of the sample, which for a disk-shaped (Hot Disk) sensor is given by [248]:

32

( ) ( )sam

qT D

rk

(4.13)

where q is the total power output from the sensor, r is the radius of the sensor and k is the thermal

conductivity of the sample. The dimensionless time function D(τ) [248] varies as a function of τ,

where τ is defined by:

r

αt

θtτ (4.14)

where α

rθ2

is defined as the characteristic time of the measurement, r is the radius of the sensor

and α is the thermal diffusivity of the sample.

For times greater than Δti, the temperature increase ΔT (t) of the sensor is:

1 ( )( ) ( ) 1i sam

o

R tT t T T

R

(4.15)

which can estimated by the recorded change in resistance R (t) of the sensor . By substituting ΔT(τ)

from equation 4.13 into equation (4.15) and defining:

32

qC

rk (4.16)

equation (4.15) can be rewritten as:

( ) ( )iT t T CD τ (4.17)

Hence, plotting the temperature increase of the sensor ΔT (t) versus D (τ), should give a straight line,

for times greater than Δti, and if the characteristic time θ has its correct value. From this particular

value of the characteristic time, θ, the thermal diffusivity, α, can be obtained (α = r2/ θ). The slope

(C) of the final plot is proportional to k-1

. In this way it is possible to determine the thermal

conductivity and the thermal diffusivity from one single transient event. However, in order to be


58

able to determine both the thermal conductivity and the thermal diffusivity from one single transient

event, it is necessary to perform the experiment so that the time period of the transient event is close

to the characteristic time, θ.

For thin samples, the probing depth as defined by equation (4.11) may be greater than the distance

from the sensor to the nearest sample boundary, and hence the assumption that the heat source is

placed in an infinite medium is not valid. Hence, for such cases, it is necessary to take the boundary

condition at one of the outer surfaces into account and consider two-dimensional radial heat flow in

an infinite plate. Such a situation has been described by Gustafsson et al [249].

Figure 4.3 (a) Schematic diagram of experimental set up with the Hot Disk sensor and (b) photo of a

Nickel/Kapton Hot Disk sensor.

Figure 4.4 (a) Schematic diagram of experiment Hot Disk set up showing the heat affected zone and (b) typical

temperature vs. time distribution in sample and sensor.

4.2.2.3 Laser Flash Method

The flash method utilizes an energy pulse (from a laser, an optical flash tube or an electron beam) to

heat the surface of a sample for a short, finite time. The energy pulse is absorbed at the surface of

the sample and it results in a temperature increase in the sample. The time-dependent temperature

on the rear of the sample is measured (see Figure 4.5). This transient response of the sample can be

Heat affected zone

Sensor

Sample

30mm

(a) (b)

(b) (a)


59

related to its thermal diffusivity. Heat losses are minimized by making the measurements in a time

short enough so that very little cooling can take place. As shown by Parker et al [250], the

temperature history T(L,t) of the rear surface of a sample of length L can be expressed in a

dimensionless form as a fraction of the maximum temperature Tmax of the rear surface by:

12

22

max

exp)1(21),(

n

n

L

αtπn

T

tLT (4.18)

where n is an integer and α is the thermal diffusivity of the sample. The rear face temperature

history following a flash heat pulse to the front sample face is shown schematically in Figure 4.5 as

described by equation (4.18). If the heat losses are negligible, the temperature at the rear face of the

sample will reach a maximum value and remain constant. The thermal diffusivity can be deduced

from equation (4.18) in two ways. When T/Tmax = 0.5, then π2 αt/L

2=1.38 and hence:

5.0

2

238.1

tπ

Lα (4.19)

where t0.5 is the time required for the rear surface of the sample to reach half of the maximum

temperature rise. Alternatively, the time axis intercept of the tangent to the time-temperature curve

at the point of maximum gradient occurs at 0.48, which yields the relationship:

xtπ

Lα

2

248.0 (4.20)

where tx is the time axis intercept of the time-temperature curve. Using the measured thermal

diffusivity and knowing the specific heat, Cp, and the density of the sample, ρ, the thermal

conductivity, k, can be calculated as the product of these three parameters (α = k/Cp ρ). Direct

determination of the thermal conductivity is possible, if the amount of energy absorbed, q, at the

front surface of the sample is known. Then the product of the density and the heat capacity can be

deduced using the equation:

max

p

qC

LT (4.21)

The above derivation assumes that heat losses from the sample are negligible. If heat losses are not

negligible, the maximum temperature reached by the sample will be reduced, it will take a longer

time to reach the maximum temperature and it will not remain constant but will fall off with time.

According to Cape and Lehman [251], the numerical factor in equation (4.19) decreases as the heat

losses from the sample due to radiation increase. The heat losses can be reduced by reducing the


60

sample thickness. However, for the derivation of equation (4.19), the energy pulse from the flash

lamp is assumed to be of negligible duration in comparison with the time it takes for the heat pulse

to propagate through the sample length. By decreasing the sample thickness, the duration of the

energy pulse will be comparable to the time it takes for the heat pulse to propagate through the

sample length, and, in such a case, the numerical factor in equation (4.19) increases with decreasing

sample thickness. For equation (4.19) to apply, the time it takes for the heat pulse to propagate

through the sample length must be at least ten times greater than the duration of the energy pulse. If

heat losses remain important after the maximum possible reduction in thickness, corrections to

equation (4.19)must be introduced. This is particularly important for poor thermal conductors or

higher temperatures, since the parameter determining the radiation heat losses is dependent on the

temperature, the emissivity and the thermal conductivity of the sample, and it is greater for poorer

thermal conductors and higher temperatures [252].

Another consideration regarding the flash method is the uniformity of the energy density across the

face of the sample. In the case that the energy density is not uniform, large errors in the

temperature measurement of the rear surface can arise, particularly if the temperature is measured

over a small area.

Moreover, the derived solution relies on the approximation that the energy pulse is absorbed

instantaneously in a small depth and that the temperature rise in the sample is small. For poorer

thermal conductors, in order to produce a measurable temperature increase on the rear face, a larger

energy pulse in the front face will be necessary. The depth at which the energy is absorbed will also

depend upon the pulse duration and thermal diffusivity of the sample, and it will be smaller for

poorer thermal conductors. Therefore, it is possible that the temperature rise in the front face of a

poor conductor would be so large that the aforementioned assumptions will not be valid. Moreover,

for a very large temperature rise at the front face it is possible to induce vaporization of the sample

in the front face.

The flash method produces an effective value for the thermal diffusivity of the sample, however, an

effective value of corresponding temperature is not determined. Usually a time averaged mean

temperature between the front and back surface temperatures up to the time that the rear surface

temperature has reached half its maximum value is taken as the effective temperature. However,

this approximation neglects the variation of thermal diffusivity with temperature. If the temperature

rise in a single measurement is kept small, then this problem is minimized. However, in a poor

thermal conductor, where large thermal gradients can be maintained during the measurement,

assigning an effective temperature to a particular measurement may prove difficult [253].


61

Figure 4.5 (a) Rear face temperature history following flash irradiation of the front face and (b) schematic

diagram of laser flash apparatus for thermal diffusivity measurement.

4.3 Thermal Conductivity of TBCs

The thermal conductivity and diffusivity of polycrystalline or single crystal stabilized zirconia, as

well as APS or EB-PVD zirconia stabilized coatings, has been the subject of numerous

experimental investigations because of the use of these materials as TBCs. Some of the

experimental measurements available in literature for YSZ are presented in Figure 4.6. The black

markers indicate the thermal conductivity of polycrystalline tetragonal 8 wt% YSZ hot pressed

powders, measured by steady state (filled symbols) and transient techniques (open symbols). The

thermal conductivity is in the range of 2.4-2.7 W m-1

K-1

at room temperature and decreases slightly

with temperature, which indicates that phonon conduction predominates [254, 255]. However, in

general, the temperature sensitivity of the thermal conductivity of dense hot pressed YSZ is not

great. The thermal conductivity of cubic zirconia is 2.0-2.4 W/m K and that of monoclinic zirconia

is about 5.4 W m-1

K-1

at room temperature [255].

In Figure 4.6, blue symbols represent conductivity measurements in EB-PVD 8 wt% YSZ TBCs

[256-258] and red symbols measurements on PS 8 wt% YSZ TBCs, both with about 10% porosity

[67, 112, 256, 259]. Filled symbols are for measurements conducted by steady state methods and

open symbols for measurements conducted by laser flash in air or another gas at atmospheric

pressure. The values reported cover a wide range, but in the majority of the measurements, the

variation of the thermal conductivity of YSZ with temperature indicates a predominantly phonon-

1.38 0.48

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8

T' =

T/ T

max

2t/L

2

Temp. recorder

amplifier recorder

trigger

laser tube

sample furnace

(a) (b)


62

based transmission mechanism up to about 1000 K (the conductivity is inversely proportional to the

temperature). At higher temperatures the thermal conductivity increases slightly with temperature,

probably due to photon conduction (radiation).

From the data available, it is obvious that the thermal conductivity is strongly dependent on the

amount and the nature of porosity. Both EB-PVD and PS coatings have lower thermal conductivity

values than polycrystalline YSZ hot pressed powder, mainly due to the presence of about 10-15%

porosity in these coatings. EB-PVD coatings tend to have higher thermal conductivity values, since

their microstructure is comprised of columns aligned parallel to the heat flux direction. Typical

thermal conductivity values for EB-PVD coating are around 1.4-1.7 W m-1

K-1

[256-258]. On the

other hand, in PS TBC splat boundaries and intersplat pores lie perpendicular to the heat flux

direction and offer significant resistance to the heat flow. The thermal conductivity of PS TBCs has

usually been reported to be in the range of 0.7-1.3 W m-1

K-1

[67, 112, 256, 259].

The wide range in the thermal conductivity values quoted in literature, even within coatings

produced with the same production method, is indicative of the sensitivity of the thermal

conductivity to the nature of porosity and the exact microstructural features in each coating,

although it must be recognized that experimental measurement errors are sometimes quite high. For

PS coatings, different spraying conditions may result in different microstructures and, even for the

same total porosity, may have quite different thermal conductivities. Therefore, it is very important

to assess or quantify these features in order to be able to make useful comparisons between different

coatings or materials.

For EB-PVD coatings, the thermal conductivity varies through the coating thickness, since the

microstructure changes progressively in this direction. For the first few tens of micrometers of the

coating, in the nucleation zone close to the bond coat interface, the grain size is small and there are

interfaces which hinder heat transfer. Further from the interface, the grains are columnar and

parallel to the heat flux [41]. Hence quoted thermal conductivity values should include coating

thickness for PVD coatings.

Furthermore, the technique used to evaluate the thermal conductivity must be considered when

assessing the reliability of the values obtained. Features of the techniques are discussed in section

4.2. Often the thermal conductivity is indirectly measured via the thermal diffusivity, using the

(volume) specific heat. While the specific heat is not usually very sensitive to microstructure, its

value may be difficult to obtain with high accuracy. Also, measurements are often carried out in

vacuum or under reduced pressure, which could lead to an underestimate-see section 8.4.


63

0.5

1

1.5

2

2.5

3

3.5

0 500 1000 1500 2000

Dense YSZ (steady state measurements)

Dense YSZ (laser flash measurements)

EB-PVD YSZ top coat (steady state measurements)

EB-PVD YSZ top coat (laser flash measurements)

PS YSZ top coat (steady state measurements)

PS YSZ top coat (laser flash measurements)

The

rma

l C

on

ductivity (

W m

-1 K

-1)

Temperature (K)

Figure 4.6. Thermal conductivity measurements for tetragonal dense 8 wt% YSZ , 8 wt% YSZ PS coatings

and 8 wt% YSZ EB-PVD coatings measured by steady state and laser flash experiments in air. Data are

from references [112, 254, 258-263].

Heat treatment can lead to an increase in the thermal conductivity of TBCs, both for PS [112, 259,

264, 265] and EBPVD coatings [254, 257, 262]. This is due to sintering of cracks, pores and

interfaces during heat treatment. In PS coatings, poor contact between lamellae in close physical

proximity can be improved by diffusional processes at high temperatures. These effects are clearly

of potential practical significance. If representative values of the thermal conductivity need to be

used for TBC design, the thermal conductivity should be determined after a heat treatment that will

result in a structure that will remain relatively stable in service.

Microstructural defects such as pores or interfaces constitute obstacles to through-thickness heat

transfer and can be deliberately introduced in order to lower the thermal conductivity. For PS

coatings, such features are inherent and can be optimized with tailored spraying parameters. For

EB-PVD coatings, morphologies designed to reduce in the thermal conductivity can by achieved by

generating inclined columns [50, 266, 267]. Such “zig-zag” microstructures have increased


64

porosity between the boundaries and results in a reduction of 20-40% in the thermal conductivity.

Another effective way of reducing the thermal conductivity of EB-PVD coatings is by producing

fine micro-crystallite structures, such as in the nucleation region of EB-PVD coatings, close to the

bond coat interface. This can be achieved by periodic re-nucleation of the EB-PVD coating growth

[268, 269]. Another approach involves layering the EB-PVD coatings by introducing interfaces or

density changes parallel to the ceramic bond coat interface. This is achieved by periodically

switching between high and low direct current bias during deposition, thus altering the degree of ion

bombardment [41, 268, 270]. Thermal conductivity reductions up to 40% have been achieved.

Changing the chemical composition and structure of the TBC can lead to a reduction of the thermal

conductivity if the changes result in increased phonon scattering or photon scattering/absorption-see

eqs.(4.1) and (4.4). The mean free path due to scattering of phonons by defects is given by equation

(4.3). It can be deduced that thereduction in the thermal conductivity due to phonon scattering

increases with defect concentration. Scattering by solute or impurities increases with the difference

in mass between host ions and foreign ions. The difference in ionic radius between the foreign ion

and the host ion, and also changes in atomic bonding, can affect the strength of the reduction in

thermal conductivity [271]. Clarke et al [271], argued that low thermal conductivity in TBCs is

favoured by: (a) high molecular weight, (b) a complex crystal structure, (c) non-directional bonding

and (d) a large number of different atoms per molecule. Consequently, the thermal conductivity of

TBC top coats can be reduced by introducing atomic level defects, such as vacancies, or

substituting Zr atoms with atoms of different mass or radius. Also a reduction in the thermal

conductivity can be achieved by the addition of dopant materials that increase the opacity of TBCs,

particularly in the infrared region, and hence reduce the radiative transport (colouring) [268].

The thermal conductivity of YSZ decreases with increasing yttria percentage. The introduction of

the stabilizer with a lower valence is accompanied by the incorporation of extra vacancies for

charge compensation, which provide an effective source of scattering of phonons [270]. The

thermal conductivity of stabilized zirconia is believed to be dominated by oxygen vacancy

scattering [262, 272]. Dopant ions, regardless of their valence, also create lattice strains due to the

difference in size between host and dopant atom, which provide an additional source of phonon

scattering.

For zirconia-based TBCs, many dopants (mainly rare earths) have been investigated. Possible rare

earth dopants for zirconia-based TBCs that have been considered include Ytterbia (Yb2O3) [18],

Erbia (Er2O3) and Gadolinia (Gd2O3) [220]. Replacing Yttria with 12 mol% Dysprosia, for an EB-

PVD coating resulted in a stable cubic phase and a 40% reduction in thermal conductivity [273].


65

Replacing Yttria with 8 mol% Erbia, Gadolinia, Ytterbia and Neodymia in EB-PVD coatings all

resulted in lower thermal conductivity, with reductions up to 45%, compared to 8 mol% Yttria, with

GaSZ having the lowest conductivity followed by NdSZ and YbSZ [268]. The effect was explained

as being due to the presence of ions of differing ionic radius and mass inducing phonon scaterring

and additionally by increased radiation absorption (colouring). Lanthana additions of 1-5 mol% to

YSZ top coats have also been reported to decrease the thermal conductivity by up to 60% at room

temperature [274].

The rare earth dopants mentioned above have lower valence than the zirconium host ion, so that

oxygen vacancies are introduced, contributing to lowering the thermal conductivity by promoting

phonon scattering. The difference in mass and atomic weight between host and dopant ion is also

believed to reduce the thermal conductivity, also by promoting phonon scattering. However,

studies by Raghavan et al [272, 275], which involved introducing heavier (Ta+5

) and lighter (Nb+5

)

pentavalent ions of similar sizes in zirconia, suggest that phonon scattering due to mass difference

alone has little effect on thermal conductivity. The introduction of pentavalent dopants in zirconia

does not introduce any oxygen vacancies, and no reduction in the thermal conductivity was

observed by doping with Ta+5

or Nb+5

. This suggests that such vacancies have strong phonon

scattering effects.

Multi-component zirconia-based TBCs of Y2O3-ZrO2, co-doped with additional paired rare earth

oxides (Nd2O3-Yb2O3 or Gd2O3-Yb2O3 or Sm2O3-Yb2O3), with the paired oxides added in a ratio of

1:1 and with a total dopant concentration of about 10 mol%, showed decreases in thermal

conductivity of up to 50%, compared to YSZ with similar dopant concentration. The added paired

oxides form defective oxide clusters, which are thermodynamically stable and reduce the thermal

conductivity, as well as being sinter-resistant [276, 277].

Alternative TBC materials with lower thermal conductivities have also been investigated. Hafnia-

based TBCs with 27wt% Yttria has been reported to have lower thermal conductivity and be more

sinter-resistant [278]. Pyrochlore structure rare-earth zirconates are reported to have lower intrinsic

thermal conductivity and higher thermal expansion coefficient than zirconia [230, 279]. Coatings

of Pyrochlore/YSZ PS double layers reportedly show improved temperature capaibilities and cyclic

lifetime, compared to conventional PS YSZ coatings [280].

4.4 Modelling of Heat Flow in Plasma-Sprayed TBCs

The problem of predicting the effective thermal conductivity of materials with inhomogeneities or

inclusions is of great interest for various areas of materials science and it has been addressed by a


66

large number of authors. A few models will be mentioned briefly, focusing on models for materials

with gas-filled inhomogeneities. Materials with randomly distributed pores have been modelled by

several authors [281-283]. Shafiro [283] and Sevostianov [282] derived the effective thermal

conductivity of materials with diverse ellipsoidal inclusions, non-randomly oriented, assuming non-

interacting inclusions. This approximation is valid for materials with dilute concentration of

inclusions. For higher porosity levels, self-consistent [284, 285] and effective medium [286]

models have been developed. Bristow [287] derived the effective thermal conductivity (along with

the effective elastic moduli) in the case of crack-like pores, and in the limit of small crack density.

Other approaches include modelling incorporating a contact resistance [288-290] or modelling

periodic structures [291].

McPherson [288] developed a model specifically for plasma sprayed coatings, and other layered

systems composed of solid lamella with small contact areas between them. The model assumes two

independent heat fluxes, one through the solid contacts areas and another through the pores. A

thermal resistance is assigned to the contact areas. This model was further developed to incorporate

thermal resistance of the lamella [289] and oxidation of the contact areas [290]. Lu et al [292]

extended the original shear lag model [291] developed for composite periodic structures for thin

cracks in a uniform matrix.

An analytical and numerical model has been developed by Golosnoy et al [293] for simulating the

heat flow in plasma sprayed thermal barrier coatings. This model will be described in more detail

in the next section.

4.4.1 Morphological representation of PS TBCs

For modelling the effective thermal conductivity, the complexity of the porous space has to be

reduced to several dominant elements. The plasma sprayed TBC structure consists of pancake-

shape splats of high aspect ratio, with poor contacts between them. This structure is represented by

a periodic array of infinite lamellae, separated by pores with bridges between them. Pores in

successive layers were assumed to have aligned arrangement, as depicted in Figure 4.7. The

structure can be described by 6 characteristic parameters (4 geometrical and 2 physical). The

geometrical parameters are the height Lv and thickness Lh of the unit cell, the thickness of the pore

dv, and the inter-bridge or interpore distance d. The two physical parameters are the thermal

conductivity of the solid lamella material k0, and the thermal conductivity of the pore kp. For the

above simplified structure geometry, the normalized thermal conductivity of the material as a whole

keff/k0, is a function of three geometrical parameters (the normalized bridge area (1-d/Lh)2, the


67

normalized inter-bridge distance Lh/Lv, and the normalized bridge height dv/Lv) and one physical

parameter (the normalized thermal conductivity of the pore kp/k0).

lamellae

interlamellar void

vertical crack contact bridge

Figure 4.7 Modelled periodic lamellar structures, showing the pore configurations.

L v

L h

( L h - d ) d/2

d v k p

k 0

Figure 4.8 Modelled unit cell for open porosity (isolated bridges) showing (a) perspective view and (b) vertical

section.

4.4.2 Physical Assumptions

Heat transfer in the polycrystalline solid occurs through phonon conduction and in the gas in pores

by molecular collision conduction. In addition, radiation heat transfer occurs through the

semitransparent material and the gas in the pores. The total heat flux through the coating can hence

be split in two parts: phonon conduction through the polycrystalline solid (or collision conductivity

in the gas) and radiative heat transfer. There are no free electrons in the constituents to consider.

Typically, TBCs consist of tetragonal Zirconia partially stabilized by 8wt% yttria. For this material,

the thermal conductivity is only weakly dependent on temperature over the range of interest, and for

dense polycrystalline solids has a value of about 2.25 W m-1

K-1

[255, 294].

The gas in the pores conducts heat through molecular collisions. Since pore thickness, dv, is

comparable to the mean free path of molecules in a typical gas at atmospheric pressure ( λ~100nm),

convection in pores can be neglected. The thermal conductivity of the gas in the pores, kg, can be

estimated by a simple analytical expression given by Lu et al [295]:

(a) (b)


68

PdBT

kk

v

g

g

1

0

(4.22)

where kg0 is the normal conductivity of the gas at the temperature concerned, P is the pressure, T

is the temperature (in K) and B is a constant, with a value of 2.5 10-5

Pa m K-1

for air. Figure 4.9

shows measured values for kg0 for air [296], together with calculated values for kg, for different pore

thicknesses and pressures. It can be seen that both pore thickness and gas pressure can have

significant effects on the pore conductivity, as well as temperature.

0

0.05

0.1

0.15

400 600 800 1000 1200 1400 1600 1800 2000

kg (d

v=0.1, P=1)

kg (d

v=1, P=1)

kg (d

v=0.01, P=1)

kg (d

v=0.1, P=0.01)

kg (d

v=0.1, P=40)

kg

0, Incropera & DeWitt, 1996

Gas

ther

mal

conducti

vit

y,

kg (

W m

-1 K

-1)

Temperature, T (K)

B = 2.5 10-5

Pa m K-1

Figure 4.9 Dependence on temperature of the thermal conductivity of the pores showing the effect of pore

thickness and gas pressure on the conductivity of air[296], according to equation(4.22).

The usual approach for incorporating the contribution of heat transfer by radiation in such materials

is to introduce a radiative conductivity which is additive to that of conduction [234, 237]. The

radiative conductivity is given by equation (4.4) and it is assumed that the total thermal conductivity

of the assemblage is the sum of the effective conductivity of the assemblage due to phonon

conduction and the radiative conductivity (given by equation 4.4). The mean free path lr (or

penetration depth) in equation (4.4) is given by,

λlr

1 (4.23)


69

where λ = γ + κ is the extinction coefficient and is equal to the scattering (γ) plus the absorption (κ)

coefficients [237]. Using this formulation, uniform radiation scattering is assumed and any

wavelength dependence or boundary effects are neglected. In typical plasma sprayed TBC material,

scattering is very strong and it tends to dominate, so that λ ~ γ.

For sintered zirconia, the penetration depth has been reported [297] to be about 50 μm. For plasma

sprayed TBCs it is expected that the penetration depth will be lower than that for sintered zirconia.

There is some data available for the refractive indices, n, and the penetration depths for zirconia and

PS YSZ [4, 234, 298, 299]. From this data it can be inferred than appropriate values for the

penetration depth lr, and the refractive index n for plasma sprayed zirconia are of the order of 10 μm

and 2.2 respectively. Using these figures, the radiative conductivity and the conductivity of the gas

in the pores as a function of temperature are shown in Figure 4.10, for a pore thickness of 0.1 μm

and a gas pressure of one atmosphere.

0

0.05

0.1

0.15

0.2

400 600 800 1000 1200 1400 1600 1800 2000

Normal gas conductivity, kg

o

Conductivity of gas within pores, kg

Radiative conductivity, krad

Ther

mal

conducti

vit

y,

k (

W m

-1 K

-1)

Temperature, T (K)

Air in pores

P = 1 atm

dv = 0.1 µm

penetration depth, = 10 µm

refractive index, n = 2.2

Figure 4.10 Dependence on temperature of the contributions to the thermal conductivity showing the

radiative conductivity and conductive of the gas within the pores, for air at atmospheric pressure with a pore

thickness of 0.1μm, according to equation 4.22.

4.4.3 Numerical Model

The unit cell is a parallelepiped. The modelled volume incorporates the main microstructural

features. A 3-D Cartesian mesh with uniform spacing was superimposed on the cell, with

approximately 100 volume elements in each direction, giving a total of about 106. The steady state


70

temperature distribution was obtained by relaxation of the transient heat transfer equation [300, 301]

(see equation (4.8)). The temperature dependence of the thermal conductivity k has been neglected

in most of the current work, except that values have been employed corresponding to low (~300 K)

and high (~1500 K) temperature regimes. The focus here is mainly on the influence of geometric

and microstructural factors. The evolving temperature distribution was obtained using an implicit

method based on the ADI technique [300, 301], with a 1% convergence criterion.

The main input parameters concerning geometry are the splat thickness (=Lv-dv), inter-splat bridge

area (Sbr/Stot), inter-splat pore thickness, (dv) and the inter-splat pore volume fraction, fp, (which

determines the bridge spacing, Lh). These values can be inferred from various types of

experimental measurements, including inspection of micrographs, densitometry and small angle

neutron scattering (SANS). The SANS technique [119, 136, 137, 302] gives information about the

spacings between planes which act as strong scattering centres, such as splat surfaces.

In general, the splat thickness depends only weakly on spraying conditions and the sprayed material

[303-305]: a typical value for sprayed zirconia would be about 2 µm and this figure was used in the

present work. A reference value used for the inter-splat pore thickness was 0.2 µm, which is

consistent with microscopy [134] and SANS [119, 136, 137, 302] evidence for sprayed zirconia.

From densitometry and SANS data, the inter-splat pore volume fraction can be estimated to be

about 3%. The relationship between the bridge area, pore volume fraction, splat thickness and

inter-splat pore thickness can be expressed

v pbr

tot v

1L fS

S d (4.24)

In practice, the bridge area and the inter-bridge spacing might vary over quite a range, particularly

as a consequence of post-deposition sintering effects, so the effects of altering these parameters was

investigated. Typical initial values [134] might be 10-30% and 5-10 µm respectively.

4.4.4 Two Flux Region Analytical Model

An analytical two flux region (TFR) model was developed [293] in order to describe the heat

transfer in the porous periodic structure. The unit cell is divided into two regions. The first zone

includes the bridge, and in this region it is assumed that heat flux takes place only through the solid

material, with no conduction through the pore. The effective conductivity of this region depends on

the area S, which is the area of this “contact” region. The second region contains two layers of solid

material and the pore between them (see Figure 4.11). The sizes of these regions are not known a


71

priori. They are established by maximizing the conductance of the assembly. The fluxes in each

region are assumed to be independent, so that:

21 )()( kS

SSSk

S

SSk

tot

tot

tot

eff

(4.25)

where keff(S) is the effective conductivity of the material, and k1(S) and k2 are the thermal

conductivities of the corresponding regions and Stot is the total cross sectional area of the unit cell.

k1(S) is a function of the area S, whereas k2 is independent and can be given by a standard series

addition expression:

1

0

2

111

pv

v

v

v

kL

d

kL

dk (4.26)

where k0 is the total conductivity of the lamellae, kp is the total conductivity of the pore, dv is the

pore thickness and Lv the height of the unit cell. The required condition of maximum conductance

of the assembly can be expressed as:

)(max],[

Skk effSSS

efftotbri

(4.27)

In order to find keff, k1 must be modelled as a function of the size of the region (area S), since k2 is

independent of S. The total conductivity of contact region k1 contains two terms: the conductivity

of the bridge itself, and the conductivity of the solid region above and below the bridge. The

conductivity of the bridge itself, kbr, can be given by:

0kS

Sk br

br (4.28)

where Sbr is the cross section area of the bridge. The heat flow in the region above and below the

bridge is being funneled towards (or away from) the bridge. The effective conductivity in this case

is often referred to as a contact resistance, emphasizing that it is the value for the resistivity of an

infinitesimally thin bridge. Contact resistances of this type have been investigated by several

authors [288, 290] [289, 295]. According to McPherson [288] and Li [289], for the limiting case of

small contact areas (d~Lh), the effective thermal conductivity, referred to as contact resistance kres,

is given by:

12

0)(2

1

vh

h

resLdL

Lkk (4.29)


72

where Lh is the width of the unit cell and d is the pore length between bridges. For the limiting case

of large contacts (d<<Lh), the modified shear-lag model [295] [291]gives:

1

23

23

23

0

)(2

)2(1

vhh

hres

LLdL

dLdkk (4.30)

A simple interpolation process between equations (4.29) and (4.30) gives the contact resistance as a

function of the region surface and bridge (contact) areas:

1

21

23

0

212

1)(

brv

br

br

res

SL

S

S

SS

SS

kSk (4.31)

The conductivity of the first (contact) region can be derived by a simple series addition expression:

1

0

1)(

11)(

kS

S

L

d

SkL

dSk

brv

v

resv

v (4.32)

Consequently, equation 4.32 can be solved, since all functions are known. The appropriate value

for the area S of the first (contact) region can be determined so that the conductance of the unit cell

is maximized. The maximum will occur at some intermediate value of the fractional bridge area or

at the total bridging limit (Sbr = Stot). This is illustrated in Figure 4.11.

Figure 4.11. Schematic depiction of the Two Flux Regions (TFR) Model, showing (a) heat flow paths and (b)

plot of how the effective conductivity varies with the area S through which the heat flux is funneled in the

vicinity of a bridge.

(a) (b)

region

1

region

2

max


73

4.4.5 Predicted Conductivity

A comparison is made in between predictions from the numerical model, the TFR model and two

other analytical models [288] [292], using typical input data for a TBC in the absence of radiation

(radiative contributions have not been incorporated into the two models from previous work). It can

be seen that the TFR model is in good agreement with the numerical modelling predictions, while

the other two can lead to some substantial errors, depending on the regimes of bridge area and inter-

bridge distance concerned. This arises from certain simplifications and approximations

incorporated in them.

0

0.5

1

1.5

2

0 0.2 0.4 0.6 0.8 1

Current work, 3D numerical

Lu et al, modified shear lag

Current work, TFR model

McPherson, contact resistance

Eff

ecti

ve

therm

al c

on

du

ctiv

ity

, k ef

f (W

m-1

K-1

)

Normalised bridge area, Sbr

/ Stot

Lh/L

v = 1

dv/L

v = 0.1

kp = 0.00008 W m

-1 K

-1

k0 = 2.25 W m

-1 K

-1

(a)

0

0.5

1

1.5

2

2.5

0 0.5 1 1.5 2 2.5 3

Current work, 3D numerical

Current work, TFR model

Lu et al, modified shear lag

McPherson, contact resistance

Eff

ecti

ve

ther

mal

co

nd

uct

ivit

y,

k eff,

(W m

-1 K

-1)

Normalised inter-bridge distance, Lh / L

v

Sbr

/Stot

= 20.25%

dv/L

v = 0.1

kp = 0.00008 W m

-1 K

-1

k0 = 2.25 W m

-1 K

-1

(b)

Figure 4.12. Effective conductivity for different models [288] [292], with no radiative contribution, as a function

of (a) bridge area and (b) inter-bridge distance.

Chapter 5. Experimental Procedures

74

5 Experimental Procedures

5.1 Characterization of Starting Materials

Powder size and morphology, as well as composition, can have a significant effect on the behaviour

during spraying and thus influence the properties of the coatings. Therefore, characterization of the

starting materials is important in order to understand coating properties.

5.1.1 Chemical Composition

The powders utilised were supplied by SULTZER METCO Inc. The bond coat material was

CoNiCrAlY (Amdry 995C). A summary of compositions and acronyms of the top coat powders is

presented in Table 5.1. Chemical compositions of powders, as given by the supplier, are

summarised in Table 5.2. The powders were manufactured by the homogeneous spherical-powders

method (HOSP).

Product Composition Acronym

204NS-1 yttria-stabilised zirconia YSZ

204NS yttria-stabilised zirconia YSZ

205NS ceria-stabilised zirconia CSZ

SPM6-2444 dysprosia-stabilised zirconia DSZ

AE8321 yttria-lanthana stabilised zirconia YLSZ

AE8170 Al2O3-doped Yttria-stabilised zirconia Al2O3-doped YSZ

AE9018 Al2O3-doped Yttria-stabilised zirconia Al2O3-doped YSZ

Table 5.1. Acronyms and compositions for the different top coat powders.

5.1.2 Particle Size Distribution

A Malvern P580 Mastersizer E machine was used for particle size analysis. Particles were

suspended in an aqueous solution and kept in suspension by a means of a stirrer. The particle size

distribution is obtained by shining a laser beam through the suspension. The suspended particles

scatter the laser beam at different angles, depending on the size: small particles scatter at large

angles, whereas large particle scatter at lower angle. A detector gathers the scattered light and the


75

angular distribution is analysed to give the particle size distribution. The distributions are presented

in Figure 5.1. The mean particles sizes are between 50-70 μm for all powders.

The microparticle population in the powders is shown in Figure 5.2. Microparticles do not

contribute much to the total mass of the powder, and do not significantly affect much the mean,

median and mode. However, they contribute to nozzle build-up during spraying, by clogging up the

nozzle and solidifying at the nozzle exit. To avoid nozzle build up, an additional carrier gas flow

rate was introduced into the nozzle, which forced particles through the nozzle and prevented them

from depositing at the nozzle exit (see Figure 5.5). All powders, expect 204NS-1, needed

additional carrier gas injection at long spraying times to avoid nozzle build-up. Powders with large

particle size distribution where more prone to nozzle built up. These two attributes are thought to

be responsible for the nozzle built up. It is concluded that in order to avoid nozzle build up, it is

important to have a narrow particle size distribution and a low population of microparticles. The

average particle size does not appear to influence build up, however, for successful melting of the

ceramic powder a relatively small average particle size is desired.

Figure 5.1 Particle size distributions for zirconia-based top

coat powders.

Figure 5.2 Distribution densities of microparticles in

zirconia-based top coats.


76

Powder Product

Chemical

Composition

(wt %)

204NS-1 204NS 205NS SPM6-2444 AE8312 AE8170 AE9018

Al2O3 0.01 <0.01 <0.01 0.02 0.88 0.8

CaO 0.04 <0.01 0.02 0.01 <0.01 <0.02

CeO2 25.15 <0.02 <0.02

Cr2O3 <0.01 <0.02

Fe2O3 0.01 <0.01 <0.01 0.03 0.01 0.01 0.01

HfO2 1.7 1.51 1.2 1.5 1.54 1.59 1.59

Dy2O3 10.1

La2O3 6.58

MgO 0.02 <0.01 0.01 <0.01 <0.01

Na2O 0.01 <0.01 <0.01

P2O5 0.01

SiO2 0.3 0.07 0.0 0.10 0.04 0.05 0.05

TiO2 0.04 0.09 0.01 0.10 0.10 0.10 0.10

U + Th <0.01 <0.01 0.01 <0.01 <0.01 <0.01

Y2O3 7.8 7.6 2.4 6.50 7.51 6.5

ZrO2 balance balance balance balance balance balance balance

Table 5.2. Chemical composition of ceramic powders supplied by SULZER METCO Inc.

5.1.3 Powder Morphology

The morphology of the top coat powders used in this study was examined using a JEOL 5800LV

scanning electron microscope. The acquisition of the images was done using a Noran Voyager

Analysis Software. To prevent charging, coatings were sputtered with gold prior to analyses in the

SEM. Figure 5.3 shows SEM micrographs of the top coat powders. All powders consist of

spherical particles with a range of different morphologies: hollow spheres, spheres with a dense

outer shell and a porous core, agglomerate spheres with smaller particles in the core and spheres


77

with a shell-structure. Representative micrographs of individual particles showing the different

particle morphologies present in all powders are shown in more detail in Figure 5.4. AE8170 and

AE8321 powders consist mainly of spheres with a dense shell and a porous core, while the YSZ

(204NS, 204NS-1) powders, mainly contain agglomerate spheres with smaller particles in the core

(Figure 5.4). However, quantitative evaluation of the morphology of the spherical particles could

not be achieved by SEM and was beyond the scope of the present study. A spherical geometry is

beneficial for good flow in the powder feed line, which in turn can lead to high feed and deposition

rates.

(a) (b) (c)

(d) (e) (f)

Figure 5.3 SEM micrographs showing the morphology of the zirconia-based top coat powders: (a) YSZ

(204NS-1), (b) YSZ (204NS), (c) CSZ (205NS), (d) DSZ (SPM6-2444), (e) YLSZ (AE8321) and (f) Al2O3-

doped YSZ (AE8170).

(a)

(b)

(c)

(d)

Figure 5.4 SEM migrographs of zirconia-based top coat powders showing the range of particle morphologies: (a)

agglomerate sphere (204NS), (b) hollow sphere (SPM6-2444), (c) shell-structured sphere (SPM6-2444) and (d) sphere

with a porous core.


78

Figure 5.5. Photograph showing nozzle build-up after spraying.

5.1.4 Phase Constitution of Starting Powders

X–ray diffraction (XRD) measurements were carried out using a computer-controlled Phillips

PW1710 diffractometer, with CuK radiation (=0.154 nm), 40 kV accelerating current and a

40 mA filament current. The incident beam optical conditions were set up with an anti-scatter slit

of 1o and a 15 mm horizontal mask. For the diffracted beam, a 0.2 mm receiving slit and a 1

o

divergence slit were used. The apparatus was regularly calibrated using a silicon control sample.

Quantitative phase analysis was performed using Rietveld analysis.

Phase (%) /

Powders

Monoclinic

zirconia

Cubic

zirconia

Tetragonal

zirconia

Ceria

204NS-1 (YSZ) 16 84 -

204NS (YSZ) 32 68 -

205NS (CSZ) 12 28 53 7

SPM6-2444(DSZ) 31 69

AE8321(YLaSZ) 8 28 64

AE8170 (Al2O3-dopedYSZ) 12 23 65

Table 5.3 Phase constitution of as-received top coat ceramic powders.

The X-ray diffraction spectra of the as-received zirconia-based top coat powders are shown in

Figure 5.6. The phase constitution of the as-received powders is, in general, in good agreement

with the expected compositions according to the available phase diagrams [231, 306-308]. Mostly,

the powders consist of a mixture of monoclinic and tetragonal zirconia. The x-ray data confirm that

the additives result in partial stabilization of the tetragonal phase of the zirconia polymorph. For the


79

Al2O3-doped YSZ powder (AE8170) no traces of Alumina were detected. This is because of the

very low Alumina content (0.88 wt%), which is not enough to show in x-ray scan. In the CSZ

(205NS) powder, some of the ceria is not into solution and appears as a separate peak in the scan.

0

2000

4000

6000

8000

10000

12000

14000

25 35 45 55 65 75

YSZ (204NS-1)

YSZ (204NS)

CSZ (205NS)

DSZ (SPM6-2444)

YLSZ (AE8321)

Al2O

3-doped YSZ (AE8170)

Cou

nts

2 (°)

Figure 5.6 X-ray spectra of as-received top coat ceramic powders.

5.2 Coating Production

5.2.1 Substrate Preparation

The substrate materials used in the present study were mild steel and a Nickel base superalloy.

Substrates of mild steel had thicknesses of 1-1.5 mm. The Nickel base superalloy was Nimonic80A

(Special Metals Limited, Hereford, UK) and it was used in rectangular blocks of 50 mm in

thickness. A chemical analysis of the substrate for Nimonic80A, as given by the supplier, is

presented in Table 5.4.

cubic Zr2O

tetragonal Zr2O

monoclinic Zr2O

cubic CeO


80

Plasma spraying requires careful surface preparation. The preliminary treatment consisted of grit

blasting and cleansing. The main purpose of grit blasting is to obtain enough surface roughness to

ensure proper mechanical anchoring of the coating. Also, inorganic contaminants and oxides, that

always exist on metal substrates, can be removed by grit blasting. The blasting equipment utilised

was a Guyson Blast cleaning cabinet, with Al2O3 500 m grit and a blasting pressure of 100 psi.

When grit blasting was performed on substrates in sheet form, the process introduced curving of the

substrate away from the blasted surface, as a result of residual compressive stresses. To remove this

deformation, the back surface was grit blasted as well. The cleansing process removed residual grit,

using compressed air and involved thorough cleaning with a degreasing agent. This process was

done immediately before spraying was carried out, to avoid recontamination. Over-sprayed coatings

were sprayed directly onto the sprayed coating, without any prior surface preparation.

For certain experiments, in order to facilitate top coat debonding after spraying, an aluminium

plasma sprayed coating was deposited onto the substrate, prior to the deposition top coat. The

aluminium powder used for this purpose was 1002 powder supplied by Newmet Koch. The

presence of the 100 m thick Aluminium coating facilitated detachment of the plasma sprayed top

coat, as explained in section 5.3.1.

Chemical composition of Nimonic Alloy 80A

weight % (except where stated ppm)

C Si Mn P S Al Ag (ppm) B Bi (ppm)

0.081 0.32 0.07 <0.003 <0.001 1.4 0.1 0.003 <0.1

Co Cr Cu Fe Ni Ti Pb (ppm) Ni+ Co

0.08 19.42 0.03 0.78 Bal 2.52 1.7 75.28 Table 5.4. Chemical composition of Nimonic 80A alloy, as provided by the supplier.

5.2.2 Plasma Spraying

Spraying was performed using a Plasma Technik AG VPS unit. The spraying unit is contained in a

double skinned, water cooled, cylindrical stainless steel chamber, with diameter of approximately

1.5 m and length of around 2 m. The plasma spraying was performed either at low pressure (after

evacuating the chamber-LPPS) or in air (APS).

The gun pattern and speed are programmed from a control console. The spray unit is controlled by a

dedicated computer system. A Windows based control software allows adjustment of the different

parameters of the spraying process. Table 5.5 gives a summary of the spray parameters for

deposition of coatings used in this study. For some of the ceramic powders, nozzle build up was

observed during spraying. In order to overcome this, additional gas (Ar) was injected at the nozzle


81

exit and some of the spraying parameters were modified, as specified in Table 5.5. The spraying

parameters used for depositing aluminium pre-coats were identical to those used for the

CoNiCrAlY bond coat- see Table 5.5.

To promote homogeneous coverage, gun motion in the x-y plane was programmed using a raster

pattern (see Figure 5.8). The raster pattern was repeated for a number of cycles. The number of gun

cycles, and the waiting time in-between each cycle was varied according to the requirements of the

coating being deposited. During APS, jets of compressed air were in some cases applied to the

backsurface of the substrate, for cooling purposes.

Figure 5.7 The plasma spray rig [309].

Figure 5.8 Raster pattern in the x-y plane for one cycle of gun movement.

substrate

gun

movement

one cycle

end point start point

one pass

gun


82

Table 5.5 Parameters employed in the plasma spraying process. Coatings AE8321 and AE9018 were

sprayed by Sulzer Inc.

5.3 Sample Preparation Procedure

5.3.1 Detaching the Top Coat

To carry out characterisation of the top coats, it was necessary to detach them from the substrate. In

order to achieve this, an aluminium coating of about 100 m thickness was sprayed on to substrate

prior to the deposition of the top coat. The attached coatings were placed in a hydrochloric acid (32 )

solution at 50 oC, which dissolved the aluminium layer, allowing the top coat to detach (after

15-20 min). In the absence of the aluminium coating, the hydrochloric acid dissolved the substrate,

Material

Parameters

CoNiCrAlY YSZ

(204NS-1)

DSZ (SPM6-2444) /

CSZ (205NS) /

YSZ (204NS)/

Al2O3-doped YSZ

(AE8170)

YLaSZ (AE8321) /

Al2O3-doped YSZ

(AE9018)

Chamber pressure (mbars) 200 (Ar) Atmospheric

(air)

Atmospheric (air) Atmospheric (air)

Stand- off distance (mm) 270 105 105 120

Plasma

Arc Current (A) 500 750 750 580

Volatge (V) 50 50 50 64

Gun speed

(mm s-1

)

100 55 55

Nozzle diameter (mm) 8 8 8 6

Plasma gas mixture

Argon flow Rate (NLPM) 50 50 50 74

Hydrogen flow rate (NLPM) 10 8 8 23

Powder Feed

Carrier gas flow rate (Ar)

(NLPM)

2 5 5 12.8

Additional gas flow rate (Ar)

(NLPM)

2

Stirrer speed (gr min -1

) 66 80 80 80

Disc speed (%RPM) 17 20 20


83

but, the procedure was more time consuming (45-60 min) and often led to contamination of the top

coat with iron chloride. To remove any residues of iron chloride, the top coat was placed in a

hydrochloric acid solution (1 %) at 50oC, which was stirred continuously using a magnetic stirrer

for about 1 h, or until all residue was removed. Subsequently the top coat was thoroughly cleaned

with a degreasing agent in an ultrasonic bath and then dried.

5.3.2 Metallographic preparation

Cutting of the top coat was performed using a diamond cutting wheel in a slow speed CAPSO Q35

precision cutting machine, which allowed parallel cuts to be made, ensuring no brittle fracture of

the top coat while cutting. Fracture surfaces for examination by SEM were produced by brittle

fracture of detached top coats using three-point-bending. Mounting of TBCs for microstructural

evaluation was performed using Struers Epovac vacuum impregnation equipment. The epoxy used

was Struers Caldofix. In order to ensure low viscosity of the resin for better impregnation, the resin

was preheated to 100o C. Vacuum impregnation was necessary in order to ensure minimum

smearing of small pores and material removal during subsequent metallographic preparation.

Metallographic preparation of the surface and/or cross section of detached top coats was required

for some of the experiments. The procedure involved grinding with a coarse SiC grinding paper

(120 grit), until a homogenous surface was achieved, followed by finer grinding paper (800 grit and

1200 grit). Diamond polishing to 1 μm finish was performed with polycrystalline diamond paste,

using polishing cloths for hard ceramics. Care was taken to minimise the polishing and grinding

times, in order to avoid unnecessary material removal and smearing of small pores during the

procedure.

For measurement of the thermal conductivity by the steady state method, samples of detached top

coats with lateral cross section of 35 30 mm were prepared. The detached TBCs were slightly

curved due to residual stresses. In preparing samples for the bi-substrate technique, special care

was taken to ensure that sample curvature was minimized during the procedure. In addition, an

attempt was made to ensure that the thickness of the sample remained uniform during preparation.

The thickness of the detached coatings after polishing was measured at 16 different points. In

general, point to point variations were below 5%. Following this procedure, it was possible to

achieve a good surface finish on detached coatings, with a roughness Ra of less than ~1 m. The

debonded coatings had thicknesses ranging from 0.5 to 2.0 mm.


84

5.4 Coating Characterisation

5.4.1 Surface roughness

Surface roughnesses were determined using a WYKO RS-2 optical interferometric profilometer.

This apparatus characterises the surface roughness and produces a topographic map of the surface.

An incident monochromatic optical beam is scanned across the sample surface and a set of fringes

are produced inteferometrically. The fringes result from constructive or destructive interference

depending on the optical path difference between a reference beam and that reflected from the

sample. The fringe pattern is analysed and a three-dimensional topographic image of the surface is

generated. The lateral resolution is about 1 μm and the depth resolution about 5 nm.

Roughness can be characterized by the values of Ra and Rz. The average surface roughness, Ra, is

the average distance of departure of the profile from the surface mean, and is the most commonly

used parameter for describing surface roughness. The average maximum roughness, Rz, is the

average height difference between the ten greatest peak-to-valley separations in the evaluation area,

and thus gives a measure of the extremes of the surface roughness and the surface texture. The

average surface roughness, Ra, was used in the present work to characterize coatings, mainly in

order to evaluate surface roughness induced by the metallographic sample preparation.

5.4.2 Scanning Electron Microscopy

Microstructural studies were performed on fracture surfaces and polished cross sections using a

JEOL 5800LV scanning electron microscope. The acquisition of the images was done using Noran

Voyager Analysis Software. To prevent charging, coatings were sputtered with gold. For higher

magnification images carbon-sputtered samples were studied using a JEOL JSM-6340F Field

Emission Scanning Electron Microscope.

5.4.3 Dilatometry

Dilatometry experiments were performed on as-sprayed free-standing top coats in a NETSCH DIL

402C dilatometer. A schematic diagram of such a dilatometer can be seen in Figure 5.9. The

specimen under investigation is placed in the head assembly and an alumina push road applies a

constant force of 30 cN. The sample is heated in a controlled manner and changes in its dimensions

are transmitted via the alumina thrust rod and detected by an electrical transducer. The resolution is

approximately 0.1 μm. The push rod and the sample holder will also undergo dimensional changes

during measurement so a calibration run is performed using a block of alumina similar in size to the


85

specimen, in order to quantify this and factor it out of the result. All tests were carried out in air.

Specimens were placed in the head assembly and heated at a heating rate of 30-50 oC min

-1. The

dwell times at these temperatures varied from 24 to 150 h. Cooling rates were about 30 oC min

-1.

Figure 5.9 Schematic diagram of the push-rod dilatometer.

Dilatometry tests were performed on free-standing top coats. As shown schematically in Figure

5.10, specimens were tested in both in-plane and through-thickness directions. Specimen

dimensions differed according to the direction being examined. Typically, to study the in-plane

characteristics, specimens approximately 10 mm in length were used. In order to examine the

sintering behaviour through the thickness of the coating, the specimen length, (coating thickness)

was about 2 mm.

Figure 5.10 Specimen configuration for dilatometry experiments.

5.4.4 X-Ray Diffraction

X–ray diffraction (XRD) measurements were carried out using a computer-controlled Phillips

PW1710 diffractometer, with CuK radiation (=0.154 nm), 40 kV accelerating current and a

40 mA filament current. The incident beam optical conditions were set up with an anti-scatter slit

in-plane

through-thickness

10 mm

2 mm


86

of 1o and a 15 mm horizontal mask. For the diffracted beam, a 0.2 mm receiving slit and a 1

o

divergence slit were used. The apparatus was regularly calibrated using a silicon control sample.

Scans over the range 2θ = 25-80o were carried out with a step size of 0.02

o and 4 s counting time at

each step. Two regions were of particular interest for the identification of the phases present. The

first of these was the region 2θ = 27-32o, where ( 1 11)m and (111)m peaks, and the (111)cub/tet peak

are found. Slow scans were performed over this range, with a step size of 0.02o and a dwell time of

10 s. The second region examined by a slow scan was 2θ = 72-75o, in order to examine the (400)tet,

(004)tet and the (400)cub peaks. Slow scans were performed over this range with a step size of 0.01o

and a dwell time of 35 s.

The mole fractions of tetragonal, monoclinic and cubic phases were calculated from the peak

intensities of the slow scans, using the following equation, from Miller [232] and Toraya [310, 311]:

/

111(111)

(111)

0.82m m

c t

m

c t

I IM

M M I

(5.1)

400

004 400

0.88 c

t t

c

t

IM

M I I

(5.2)

1m c tM M M (5.3)

where Mm, Mc and Mt are the mole fractions of the monoclinc, cubic and tetragonal phases

respectively and I is the integrated intensity corresponding to the peak concerned. The integrated

intensity was calculated after peak deconvolution and profile fitting, performed using Phillips

PROFIT software. The software can fit Lorenzian, Gaussian, pseudo-Voigt and Pearson-Seven

profiles, all of which can be symmetrical or asymmetric, and it will assign the K2s. This

procedure allows accurate determination of peak positions, peak areas and widths.

Alternative, quantitative phase analysis was in some cases performed with the full x-ray diffraction

pattern, using the Rietveld method based on the Phillips X’pert Plus software. The Rietveld code in

the software is based on [312]. In order to perform the analysis, all phases must be identified and the

quality data for the structure of the phases present must be available. The Reitveld method permits

simultaneous refinement in each of the present phases of both structural (lattice parameters) and

microstructural (phase percentage) parameters.

The (400)tet, (004)tet and (400)cub peak positions were used for calculation of lattice parameters, a

for the cubic phase and a and c for the tetragonal phase. The Y2O3 content of the tetragonal and


87

cubic phases were calculated from these measured lattice parameters, respectively using the

following equations [232, 306], where a and c are expressed in nanometres

1.5

1.0223

mole % YO in tetragonal phase0.001309

c

a

(5.4)

1.5

- 0.5104mole % YO in cubic phase =

0.0204

a (5.5)

5.4.5 Porosity Measurements

Porosity measurements were performed on as-sprayed and heat-treated free standing top coats,

using a liquid immersion technique. Mass measurements were performed using a Sartorius

RC120P microbalance, with a sensitivity of ± 10 g. Firstly, the mass of the sample was measured

and recorded (mair). Then, the specimen was coated with a viscous lacquer of known density ρlaq

and its new mass measured (mc). The lacquer was used in order to prevent the filling of any surface-

connected cracks or pores by capillary action of the liquid. Finally, the coated specimen was

immersed in liquid and its mass was recorded (mliq). For the purpose of recording the specimen

mass while immersed in liquid, an appropriate set up was utilised, which is shown schematically in

Figure 5.11. The liquid used was Flutec PP9 (per fluoro-1-methyl decalin), a dense, low surface

tension liquid with known density ρliq. The following equation was used to determine the density of

the sample:

laq

airc

liq

lidc

aird

mmmm

m

(5.6)

The porosity fraction of the specimen was calculated by comparing the measured density with the

theoretical density for the bulk material ρth, following the equation below:

1 df

th

p

(5.7)

The theoretical density of each of the coating materials was calculated in the following way. From

x-ray data, the lattice parameters of the crystal structure were deduced and the volume of the unit

cell was calculated. From the weight percentage of the dopant, and the atomic weight of the

elements present, the atomic mass of the unit cell was calculated. It was assumed that dopant

cations substitute for the zirconia ions in the structure. The presence of impurities was not taken


88

into consideration. The theoretical density was calculated from the atomic mass and the unit cell

volume. The estimated error for this technique is approximately 3-4 %. The main sources of

experimental error arise from the weight measurement when the sample is immersed in the liquid

and from uncertainties in the density of the liquid and the lacquer.

Figure 5.11 Schematic diagram of set-up for density and porosity measurements (after Murphy [313]).

5.4.6 Stiffness Measurement

5.4.6.1 Cantilever Bending

Young’s moduli of the detached top coats was determined using the Cantilever Bend Test. This is a

modified version of the cantilever beam test used by Tsui [76] and has been found suitable for

determining the Young’s modulus of materials with relatively low stiffness, such as plasma sprayed

top coats [309]. A schematic diagram of the loading configuration used is shown in Figure 5.12. A

fixed incremental load P was applied at a distance L from the clamped end. The displacement in the

y direction, at distance x from the fixed end, was measured using a non-contact laser scanning

extensometer (Lasermike). The stiffness was determined using the equation below:

62

32 xLx

yI

PE (5.8)

where I is the second moment of area of the sample and the other variables are as described in

Figure 5.12.


89

Figure 5.12 Loading configuration for the cantilever bend test.

5.4.6.2 Nanoindentation

Nanoindentation was carried out using a MicroMaterials NanoTest 600 indenter. Nanoindentation

testing consists of loading and unloading a diamond indenter into the sample under investigation

and monitoring how the indentation depth changes. All indentations were done with a Berkovich

diamond indenter, which is a three-sided pyramid. The indenter was pressed into the surface with a

specified loading rate. A dwell time at maximum load was included, to prevent gross errors due to

creep [314]. In this case, the maximum load was 100 mN, the loading rate was 5.1 mN s-1

and the

dwell time at maximum load 10 s. Displacement was measured via a parallel plate capacitor. The

location of the indents and the indentation stage is accurately controlled by an actuator.

The deformation on loading/unloading of the indenter is a combination of both elastic and plastic

strain [315]. Both the hardness and modulus of the material can be determined by this method.

During, loading there is a combination of both and elastic and plastic deformation, whereas, during

unloading the material recovery is principally elastic. Young’s modulus was determined from the

unloading curve, using the Oliver and Pharr [315] technique. The reduced modulus of the material,

Er, is determined from the initial gradient of the unloading curve, S, using the following equation:

4r

dP AS E

dh (5.9)

where A is the indent contact area, which is determined from the displacement, h, according to a

pre-determined diamond area function. The Young’s modulus of the sample, Es, can then be

estimated using the Poisson’s ratio of the indenter, νi, and the sample material, νs, and the Young’s

modulus of diamond, Ei, following the equation:

L x P

y


90

2 21 11 s i

r s iE E E

(5.10)

The diamond indenter is commonly taken to have a Young’s modulus of 1147 GPa and Poisson

ratio of 0.07. Nanoindentation tests were performed on polished cross-sections of free-standing top

coats. The data were analysed using a program supplied by NanoTest, which uses calibration

indentations performed regularly on fused silica.

5.5 Heat Treatment of Thermal Barrier Coatings

5.5.1 Isothermal Heat Treatment

Isothermal heat treatments were performed on free-standing APS top coats with typical thickness of

0.5-2 mm, at temperatures in the range 1200-1400˚C. The heat treatments were performed in air in

a Lenton furnace, with a heating rate of about 30oC/min. Samples were removed from the furnace

once the heat treatment was completed and were air cooled at a rate of about 30Co/min.

5.5.2 Heat Treatment with a High Thermal Gradient

Heat treatments of TBCs under high thermal gradient were performed in a specially designed rig, in

which a high thermal gradient through the top coat could be established. The heat treatments were

carried out on Nimonic80A superalloy substrate block, about 50 mm thick, with a CoNiCrAlY bond

coat about 40 m thick and top coats 1-2.5 mm thick. A schematic of the rig is shown Figure 5.13.

The rig was constructed within a controlled atmosphere chamber. The heating source was provided

by a graphite susceptor heated by an induction coil. Typical power input into the induction coil was

3-5 kW. The base of the substrate was brazed to a water-cooled copper block, with a silver-based

braze Easyflow No.2. The flow rate of the cooling water was 4 l/min. Insulation was placed

around the specimen to minimise lateral heat losses. The graphite susceptor could be lifted away

from the induction coil by a hydraulic ram, allowing thermal cycling of the specimen. The thermal

gradient was monitored and controlled by four thermocouples embedded in the substrate. The

temperature in the top coat was monitored using a thermocouple, which was inserted into a laser

drilled hole, parallel to the bond coat/top coat. The laser drilled hole was 2 mm deep and was

drilled using a pulsed Nd-YAG laser with wavelength =1.06 m, 0.5 ms pulse duration and energy

2.5 J per pulse.


91

The holding time at temperature, for each thermal cycle, was 1 hour and samples were allowed to

cool to room temperature before being re-heated. Heating rates were ~20 K min-1

and cooling rates

~30 K min-1

. Typical top coat temperatures during the gradient heat treatments applied in the

present work ranged from about 1500˚C at the free surface to 900˚C at the interface. A vacuum of

2 10-4

mbar was maintained throughout. The chamber incorporated a viewing window, which

allowed automated periodic image capture with the aid of a digital video camera.

(a) (b)

Figure 5.13. The high thermal gradient rig: (a) schematic diagram and (b) photo taken during operation.

5.6 Measurement of Thermal conductivity

5.6.1 Steady State Rig

5.6.1.1 Experimental Set Up

Figure 5.14 shows a schematic of the setup. A sample, with lateral cross section of 35 30 mm,

was sandwiched between two Nimonic 80A alloy substrates (Special Metals Limited, Hereford,

UK). These substrates act as flux meters, each being instrumented with four K-type thermocouples

(T1 to T8), inserted into holes at known distances, drilled to the centerline of each block. In

principal, only two temperature readings are required from each substrate to determine the heat flux,

but by taking additional readings, an improved average value for the flux can be obtained. Since

the (temperature-dependent) thermal conductivity of Nimonic 80A is well known over a wide


92

temperature range [316], the heat flux across each substrate can be accurately determined. Of

course, the substrate can also be made of other materials and with different cross-sections.

Temperature gradients are generated by heating the lower substrate with an electric resistance

heater, while heat is continuously removed from the upper substrate via a water-cooled copper heat

sink. To maintain consistent heat supply and removal, the heater power and cooling water flow rate

are fixed throughout the experiment. The setup is insulated with a low conductivity glass wool, to

minimise lateral heat losses and promote one-dimensional heat flow. A check can be made about

this, since significant losses would result in the heat fluxes through the two substrates being

different.

To eliminate air gaps and raise the interfacial thermal conductance (h), two types of substrate-to-

sample interfacial materials were tested:- (i) a silicone-based thermal compound (HTSP

Electrolube), with conductivity of 3.0 W m-1

K-1

and (ii) Sil-Pad®2000 (Bergquist), a high

performance conductive pad, with thickness of 0.38 mm and conductivity of 3.5 W m-1

K-1

. The

suitability of these two interfacial materials is discussed in more detail in section 8.1.1. To ensure

that the interface can be reproduced for different runs, a fixed torque of 3 N m was applied onto the

test column via the bolt. This torque generated an axial force of 750 N, corresponding to a nominal

pressure of about 0.72 MPa for the current sample cross-section. During the experiment, the

temperature readings (T1 to T8) were continuously logged. A steady state was considered to have

been established when the temperature fluctuations were within ±0.5ºC, for more than 30 minutes.

These readings were then employed in calculating the thermal conductivity of the sample using

equation (4.4) for steady state heat flux.


93

Figure 5.14. Schematic of the steady-state bi-substrate thermal conductivity setup. T1 to T8 designate the

eight thermocouples used for monitoring the temperature within the two substrates (flux meters).

Figure 5.15. Photo of the bi-substrate steady state thermal conductivity set up and data logging

equipment.


94

5.6.1.2 Data Analysis

The data analysis performed in order to calculate the thermal conductivity from the measured

temperature profile and sample thickness is presented here. By assuming one-dimensional heat

flow across the setup, and by taking into account changes in thermal conductivity of the blocks with

temperature, the mean heat flux, Q, can be found using eqns.(5.11) to (5.13):

Qupper

1

6

Ti T

j

xi x

j

ksub

Tave

ij j1

3

i2i j

4

, T

ave

ij T

i T

j

2 (5.11)

Qlower

1

6

Ti T

j

xi x

j

ksub

Tave

ij j5

7

i6i j

8

, T

ave

ij T

i T

j

2 (5.12)

where i and j designate locations of thermocouples (see Figure 5.14), while k

subT

ave

ij is the

temperature-dependent thermal conductivity of the Nimonic substrate at the average temperature.

The mean heat flux can be written as:

Q

1

2Q

upperQ

lower (5.13)

The measurement is presumed to satisfy one-dimensional heat flow when:

Qupper

Qlower

Q10% (5.14)

For a typical experiment, the difference was found to be significantly less than 10%.

Figure 5.16 shows a schematic of the temperature profile established across the substrates and

sample (coating) under steady state conditions. Assuming no lateral heat losses, the same flux

flows through coating and interfaces, so that:

Q k

eff

T

x (5.15)

Q hT

i (5.16)

Q k

true

Tc

x (5.17)

where keff is the effective thermal conductivity of the coating, T is the total temperature drop, x is

the coating thickness, h is the interfacial thermal conductance (assumed the same for both interfaces)

and ktrue is the actual coating thermal conductivity, while the subscripts i and c designate interface

and coating, respectively.


95

The total temperature drop, T, consists of temperature drops across the coating and interface, that

is

T T

c 2T

i (5.18)

By substituting eqns (5.16). and (5.17) into eqn.(5.18), and rearranging, we obtain the following

linear equation:

T

Qx

ktrue

2

h (5.19)

Now, by plotting T/Q versus x, the slope and intercept are given by 1/ktrue and 2/h, respectively.

Both ktrue and h can thus be determined.

Figure 5.16. Schematic depicting the different

temperature drops across the substrates and

sample (coating).

5.6.1.3 Validation of Technique

Before the technique can be used with confidence, it must be verified using a standard material of

known thermal conductivity. Fused silica (quartz glass) samples supplied by Heraeus Quarzglas

(Germany), were chosen for this purpose. In the manufacturer’s material datasheet, its thermal

conductivity is reported as 1.46 W m-1

K-1

at 100ºC [317]. Samples with three different thicknesses,

i.e. 0.567, 0.989 and 2.897 mm, were used. Figure 5.17 shows the temperature versus distance plot

for a silica sample with a thickness of 2.897 mm. The mean temperature of the sample was about

100ºC and the total temperature drop (T) across the sample was about 48ºC. Similar profiles were

observed for the thinner samples, but with smaller drops. Using eqns.(5.11) to (5.13), the mean heat

flux was estimated as 26.36 kW m-2

. Subsequently, from eqn.(5.15), the effective thermal

conductivity (keff) was found to be 1.25 W m-1 K-1 .


96

Figure 5.17. Temperature versus distance plot showing the temperature drop across a fused silica sample.

The points (T1 to T8) are thermocouple measurements corresponding to locations in the Nimocic substrate.

The line is calculated using the average heat flux, Q, and temperature-dependent thermal conductivities of

Nimonic substrates.

Figure 5.18 shows experimental data for all the fused silica samples, along with the best linear fit,

as given by eqn.(5.19). It can be seen that the data show relatively little scatter. The values of

actual thermal conductivity (ktrue) and interfacial thermal conductance (h) were found to be

1.54 W m-1

K-1

and 12.88 kW m-2

K-1

, respectively. Of course, ktrue is higher than keff, since

temperature drops across the two interfaces (2Ti) have now been taken into account. The actual

conductivity agrees quite well with the value (1.46 W m-1

K-1

) quoted by the manufacturer [317]. It

may also be noted that the interfacial contact conductance measured here is of a similar order of

magnitude to values reported elsewhere (for a similar surface finish and contact pressure) [318].

Samples with lateral cross section of 35 30 mm of as-sprayed YSZ (204NS) were prepared by

debonding the top coats from the substrates, and subsequently polishing the surface to ~1 m

surface finish. The debonded coatings had thicknesses ranging from 0.5 to 2.0 mm. Care was taken

to ensure a homogenous thickness.


97

Figure 5.18. A plot of T/Q versus x for fused silica samples of three different thicknesses. The points are

experimental data while the line is their best linear fit using eqn.(5.19). The values of ktrue and h can be

obtained from the slope and intercept, respectively. Conductive compound was applied at the interface, and

the mean temperature of the samples was about 100ºC.

5.6.2 Hot Disk

A Hot Disk Thermal Constants Analayser Version 5.0 [249] was used for thermal conductivity

measurements of free-standing top coats. In this technique a disk-shaped sensor, which acts both

the heat source and the temperature sensor, is sandwiched between two identical samples. A

current pulse is passed through the sensor, sufficient to cause a slight temperature increase in the

sensor and in the sample surrounding it. Assuming that the resistivity of the metal element in the

sensor is temperature dependant, the temperature increase in the sensor and the sample surrounding

will cause a change in its resistance and a corresponding voltage variation over the sensor. From

the change in resistivity with time, and assuming constant current and no influence to the

temperature increase from the outer boundaries of the sample, the temperature distribution with

time in and around the metal strip can be expressed in terms of the thermal diffusivity and the

thermal conductivity of the sample. Details on the underlying theory are given in section 4.2.2.2.

The Hot Disk sensor consisted of an electrically conducting double-spiral pattern of Nickel

supported in both sides by an insulating layer of Kapton. The radius of the sensor was 2.001 mm.

The output powder used for the experiments was 0.2 W and the measurement time 2.5 sec. The


98

samples used were approximately 35 30 mm and 2 mm thick. Analysis of the data was performed

using the Hot Disk Thermal Constants Analyser software.

The temperature increase in the sensor for YSZ as-sprayed top coat is presented in Figure 5.19. The

“time window” for analysis was chosen taking into account the sample dimensions and an initial

stabilization time. For the first few data points the temperature difference between the sample and

the sensor due to the interface resistance is not constant, and towards the end the temperature might

be influenced by the size of the sample- see Figure 5.19. The probing depth corresponding to the

time window chosen is 1.98 mm, which is just under the thickness of the samples used.

0

0.5

1

1.5

2

2.5

3

3.5

0 0.5 1 1.5 2 2.5 3

Tem

per

ature

Incr

ease

T

(oC

)

Time (sec)

Analysis

Figure 5.19. Temperature increase of the Hot Disk sensor as a function of time showing time-window used

for data analysis.

For the correct value of the characteristic time of the measurement α

rθ2

, where r is the radius

of the sensor and α is the thermal diffusivity of the sample, the temperature increase recorded by the

sensor can be expressed as a linear function of the dimensionless function D(τ) (see eqns 4.13 and

4.14). Figure 5.20a shows a plot of the temperature increase as a function of D(τ), which is a

straight line. The slope of the line in Figure 5.20a is inversely proportional to the thermal

conductivity of the tested sample (see equations 4.16 and 4.17). The selected data points for

analysis where adjusted to a straight line by varying the characteristic time θ. The random scatter of

the data points around the fitted straight line is displayed in Figure 5.20b, which is a plot of the


99

difference between the measured and fitted temperature versus Sqrt(time). Since the scatter is

random, the correct value for the characteristic time has been used.

(a)

2.6

2.8

3

3.2

3.4

0.2 0.22 0.24 0.26 0.28 0.3 0.32

Tem

per

ature

incr

ease

(oC

)

D()

(b)

-0.003

-0.002

-0.001

0

0.001

0.002

0.003

0.8 0.9 1 1.1 1.2 1.3 1.4 1.5

Tem

per

ature

Dif

fere

nce

T

fitt

ing -

T

exp (o

C)

Sqrt(time) Figure 5.20. Hot Disk measurement: (a) temperature increase as a function of the dimensionless time function D(τ) and (b)

random scattering of experimental data around theoretical fitting as a function Sqrt(time).

5.6.3 Laser Flash

The laser flash method was used in order to calculate the thermal conductivity of free-standing top

coats. Measurements were carried out in the Thermophysical Properties Section of the NETZSCH

Application Laboratory. The thermal conductivity, k, was calculated using the following equation

(see also equation 4.10):

( ) ( ) ( ) pk T T C T (5.20)

where α is the thermal diffusivity, Cp is the specific heat, ρ is the density and T is the temperature.

The thermal diffusivity was measured using a Netzsch model 427 laser flash diffusivity apparatus.

The unit used in this work was equipped with a high-temperature, water-cooled furnace capable of

operation from 25 to 2000°C. The sample chamber is isolated from the graphite heating element by

a protective tube allowing samples to be tested under vacuum or in an oxidizing, reducing or inert

atmosphere. The temperature rise on the back face of the sample is measured using an In-Sb

detector. Data acquisition and evaluation are accomplished using a comprehensive PC software

package. The thermal diffusivity measurements were conducted in a dynamic argon atmosphere at

a flow rate of 100 ml/min between room temperature and 1400°C. The presented thermal


100

diffusivity results are the average values of five individual tests. The samples were free-standing

square disk top coats with side lengths of approx. 10 mm and thicknesses of approx. 2 mm. In order

to increase both the absorption of laser energy and emission of infrared light, the samples were

coated with graphite.

The specific heat measurements were conducted using a Netzsch model DSC 404 C Pegasus

differential scanning calorimeter capable of operation from 25 to 1500°C. The system is vacuum-

tight, and therefore samples can be tested under pure inert, reducing or oxidizing atmospheres, as

well as under vacuum. The specific heat is determined by running a baseline and standard over the

temperature range of interest. The sample is then run and the specific heat is calculated by the

standard ratio method. Instrument control and data acquisition are accomplished via a new 32-Bit

MS®-Windows

TM Thermal Analysis software package. Data evaluation is carried out by a

comprehensive PC software package. The samples tested with the DSC had masses of approx. 200

mg. The measurements were carried out in a dynamic argon atmosphere (gas flow rate: 50 ml/min).

The system was equipped with a temperature-calibrated DSC-cp type S sensor. Platinum crucibles

with lids were employed for the test. The samples were heated between room temperature and

1400°C at a heating rate of 20 K/min.

Chapter 6. Thermal Stability of PS top coats

101

Equation Section 6

6 Thermal Stability of Plasma-Sprayed Top Coats

6.1 Changes in Phase Constitution of Plasma-Sprayed Top Coats due

to Heat Treatment

Phase characterisation of TBCs is of great importance for both studying and designing materials for

specific applications. TBCs are often made of polymorphs whose presence is controlled by many

factors such as stabilizer content, processing parameters and heat treatment. The polymorphism of

zirconia, particularly the volume expansion associated with the tetragonal-to-monoclinic

transformation, has led to the investigation of various stabilized zirconia alloys. The phase stability

of the top coat constitutes an important factor for determining the thermomechanical properties of

TBCs.

6.1.1 As-sprayed

6.1.1.1 Yttria stabilized zirconia top coats

The x-ray diffraction patterns of the as-sprayed top coats are presented in Figure 6.1 to Figure 6.4.

The YSZ as-sprayed top coats (204NS & 204NS-1) are composed of non-transformable T’

tetragonal phase. The K1 and K2 components of the T' peaks can be distinguished. This phase is

formed by a diffusionless transformation of the high-temperature cubic F phase, due to rapid

solidification and cooling during spraying. The T’ non-transformable tetragonal is not an

equilibrium phase and it differs from the regular tetragonal phase, notably for its smaller

tetragonality, higher yttria content and stability against further transformation into the tetragonal

phase [306, 319-322]. However, for crystallographic purposes, the T’ non-transformable tetragonal

phase and T tetragonal phase are fundamentally the same tetragonal polymorph in the zirconia solid

solution [207]. Stabilization of the tetragonal or cubic zirconia phase by the addition of trivalent

dopants is believed to be associated with the generation of oxygen vacancies [202]. The trivalent

dopant cations substitute the Zr ion in the cation network and oxygen vacancies are created for

charge compensation. The tetragonal and cubic polymorphs are stabilized when the doping level is

sufficient to create enough oxygen vacancies to reduce the co-ordination number of the Zr ion from

8 to around 7.5 [203, 204]. The Zr ion in fully stabilized and high temperature cubic phase has a

coordination number of 7, due to the strong covalent nature of the ZrO bond and the small size of

the Zr ion.


102

The position of the tetragonal peaks is known to move as a function of the amount of yttria that the

tetragonal phase contains, due to the changing lattice parameter. Due to the slightly different yttia

content, the tetragonal peaks are not in exactly the same position in the two coatings. Trace

amounts of the monoclinic phase exists in the 204NS-1 YSZ powder (1.5 %), which may be due to

the presence of unmelted particles thatcould contain some monoclinic phase. This is in agreement

with previous work in this area [76, 109]. The different amount of impurities in the starting

powders 204NS and 204NS-1 of these two coatings does not have a noticeable effect of their phase

constitution.

The two YSZ top coats doped with a small amount of Al2O3 (AE8170 & AE9018) also contain the

non-transformable tetragonal phase. The shoulder observed on both the tetragonal peaks are due to

the K1 and K2 components of the T' peaks. The presence of a small amount of Al2O3 does not

appear to alter the phase constitution of the top coat. The solubility of Al2O3 in Zr2O is very low

[323, 324] (<0.5 mol%), so it is possible that the Al2O3 is not in solid solution. However the

amount present is too small to form a distinct peak. It can be concluded that impurities such as SiO2

and Al2O3 do not have an effect on the phase constitution of as-sprayed YSZ coatings.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

M (111)M (111)M (111)

72 73 74 75

Inte

nsi

ty

2 (o)

T'(004)

(b)T'(400)

F(400)

Figure 6.1 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ for as-sprayed YSZ (204NS-1) top coats.


103

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

T'(004)

(b)T'(400)

F(400)

Figure 6.2. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚for as-sprayed YSZ (204NS) top coats.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

T'(004)

(b)T'(400)

F(400)

Figure 6.3. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚for as-sprayed Al2O3-doped YSZ (AE8170)

top coats.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

T'(004)

(b) T'(400)

F(400)

Figure 6.4. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚for as-sprayed Al2O3-doped YSZ (AE9018)

top coats.


104

6.1.1.2 Dysprosia stabilized zirconia top coats

Dysprosia is also believed to be an effective stabilizer in zirconia, up to concentrations of about

15 mol % ( 35 wt%) [308]. When the concentration of Dy2O3 is above 15mol%, the pyrochlore-

structure Dy2Zr2O7 phase and zirconia solid solution are supposed to co-exist. The DSZ top coat

(SPM6-2444) in the present study, which has about 3.5 mol% Dy2O3, comprises the non-

transformable tetragonal T’ with a trace of monoclinic phase (1.5 %). No pyrochlore phase is

observed. The K1 and K2 components of the T' peaks can be distinguished (Figure 6.5).

The tetragonality c/a is an important factor in the stability of the tetragonal phase and it has been

shown to be independent of the species of the dopant, but dependant on the content of the dopant

for oversized trivalent rare earths[205, 206]. Tetragonality decreases with increasing dopant

content for oversized trivalent cations and vanishes at 11 mol% M2O3 regardless of the ionic sizes,

since tetragonality is aided by the creation of oxygen vacancies only [207]. The mol% dysprosia in

the DSZ (3.58 mol%) in the present study is slightly lower than that of yttria in the YSZ top coats

(about 4.3 mol%). Therefore, we expect the DSZ top coat will have a higher tetragonality than that

of YSZ coatings (c/aDSZ = 1.012, c/aYSZ = 1.010). The decrease in tetragonality is associated with

increasing stability of the tetragonal phase and hence the resistance of the non-transformable

tetragonal to transformation into the monoclinic phase is ascribed to its smaller tetragonality.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

T'(004)

(b) T'(400)

F(400)

Figure 6.5. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚for as-sprayed DSZ (SPM6-2444) top coat.


105

6.1.1.3 Yttria-lanthana stabilized zirconia top coats

For the YLaSZ, the deconvoluted spectrum is presented in Figure 6.6b, which shows that the as-

sprayed coating probably consists of two tetragonal phases with slightly different lattice parameters.

Possibly the presence of the two tetragonal phases is a result of two different dopants with different

ionic radii. Whereas tetragonality is independent of ionic radius of the dopant and depends only on

dopant content, on the other hand lattice parameters of the tetragonal phase increase systematically

with the ionic radii of the dopant atoms [205, 206]. If the two tetragonal phase that appear to be

present are a result of the simultaneous presence of yttria and lanthana, then the tetragonal phase

with the greater lattice parameters is a result of the lanthana dopant, whereas the tetragonal phase

with slightly smaller lattice parameters is a result of the yttria dopant, since the lanthanum cation

has a bigger ionic radius (see Table 6.1).

The stabilization effect of lanthana in zirconia is disputed in the literature. As the ionic size

mismatch between the dopant and the zirconium atom becomes greater, the solubility of the

tetragonal and cubic phase decreases and the pyrochlore phase may be formed instead of the

tetragonal or cubic phase. According to the binary zirconia lanthana phase diagram [231] at above

about 2 mol% La the equilibrium phases are the monoclicnic and the pyrochlore phase. No

pyrochlore phase was detected in the present study in the as-sprayed coatings. Din et al [308]

reported that the addition of lanthana to zirconia did not stabilize the T phase. However, a

pyrochlore-type cubic phase La2Zr2O7 is formed. Bastide et. al [227] also found that the solid

solubility of La in monoclinic Zr is 1.5 mol % and above that concentration pyrochlore-structured

La2Zr2O7 phase is formed. However, Li et al [325] reported that 8mol% La2O3 stabilized the

tetragonal phase of La2O3–ZrO2 powder compacts. In the present study the simultaneous presence

of yttria and lanthana stabilizers in zirconia appears to result in two metastable tetragonal phases in

the as-sprayed coatings.

Ion Zr4+

Y3+

Dy3+

La3+

Ce4+

Ce3+

Ionic radius (nm) 0.084 0.102 0.103 0.116 0.097 0.114

Table 6.1. Ionic radii for cations in PSZ top coats [326].


106

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.6. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ for as-sprayed YLSZ (AE8321) top coat.

6.1.1.4 Ceria stabilized zirconia top coats

The CSZ powder also contains the non-transformable tetragonal phase (see Figure 6.7). Tetravalent

dopants such as ceria stabilize the tetragonal and cubic phases of zirconia [38, 190]. However, the

stabilization is not the result of the generation of oxygen vacancies, as is the case with trivalent

dopants. The Ce4+

ion substitutes the Zr4+

ion in the cation network, without creating any oxygen

vacancies, since both ions are tetravalent [208, 209]. The stabilization effect is thought to be caused

by the slight dilation of the cation network by the oversized Ce4+

ion which decreases the

tetragonality and stabilizes the tetragonal phase to room temperature [208]. In Figure 6.7 broad

XRD peaks are observed for the CSZ top coat. This has commonly been observed previously with

CSZ top coats. It is thought to relate to compositional variations, owing to the larger size of the Ce

ion within the zirconia lattice [37]. Some research [219, 327] suggests that the tetravalent Ce ion is

reduced to is trivalent state because of the reducing atmosphere in the plasma plume and the rapid

quenching thereafter of the molten droplets. In such situation, oxygen vacancies would be

introduced to balance the lower valance state of the cerium ion. This would result in stabilization of

the cubic phase and possibly the presence of both non-tranformable tetragonal T’ and cubic F

phases. The presence of some trivalent cerium could account for the large variation in the unit cell

responsible and thus the broad XRD peaks observed. The presence of some cubic phase is also a

possibility. However, the presence of a cubic peak cannot be confirmed in the data presented here

(see Figure 6.7), since the height of any cubic peak is expected to be of the same order as the noise,

even though the signal to noise ratio is satisfactory. Annealing at relatively moderate temperature

(700oC) should result in re-oxidation of the cerium into its tetravalent state and stabilization of the


107

tetragonal phase, with no cubic phase present. The colour of CSZ plasma sprayed coatings in which

the cerium ion has been reduced to its trivalent state is brown-green [38, 218], whereas for CSZ

coatings with the majority of the substitutional Ce ions in their tetravalent state the color is pale

yellow [38]. The samples used in the present work were pale yellow in the as-sprayed state, further

suggesting that they consist mainly of tetravalent cerium and non-transformable tetragonal phase

T’(see Figure 6.9). It is possible that trivalent cerium formed during spraying was re-oxidized

during cool-down, since the plasma sprayed coatings were air cooled to below 120oC inside the

spraying chamber. Previous work [209] suggests that the ceria ions in CSZ plasma sprayed

coatings are preferentially in their tetravalent state.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

T'(004)

(b)T'(400)

F(400)

Figure 6.7. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ for as-sprayed CSZ (205NS) top coat.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

(b)F(400)

Figure 6.8. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ for CSZ (205NS) top coat that has been

annealed in vacuum at 500oC.

The reduction of cerium atoms to its trivalent state is accompanied by a volume change. The Ce3+

ion is larger than the Ce4+

ion, resulting in a decrease in the solubility of CeO2 in the tetragonal

solid solution and causing a decomposition in the T-phase. Figure 6.8 shows X-rays spectra for


108

CSZ after annealing at 500oC in vacuum. The tetragonal phase has decomposed into cubic. In

addition, the colour of the as-sprayed coating changed to dark-brown/green (see Figure 6.9), which

suggests that the cerium cation has been reduced to its trivalent state. Dal Maschio et al [219]

noticed a volume diminution at 340oC during cool-down, believed to be due to the phase

transformation from cubic to tetragonal due to re-oxidation. The valance state of cerium in CSZ top

coats is believed to have a large influence on the stress state in the ceramic coating [219]. The

simultaneous presence of ceria and yttria makes it difficult to estimate the stabilizer content from

the peak position, using known relation between lattice parameters and stabilizer percentage.

Figure 6.9 CSZ (205NS) top coat (a) as-

sprayed and (b) after annealing at 500oC in

vacuum. The change in colour is due to

reduction of Ce4+

to Ce3+

.


6.1.2.1 Yttria stabilized zirconia top coats

After holding the YSZ 204NS-1 top coat at 1300˚C for 100 h, the tetragonal T' phase decomposed

to a mixture of low-yttria tetragonal T'l, high-yttria cubic F and T’2 with an intermediate yttria

composition (Figure 6.10). According to the phase diagram (see Figure 3.12a) YSZ top coats

composed of non-transformable T´ should decompose at high temperature, to a mixture of cubic

and tetragonal phases. On cooling to room temperature, the high yttria cubic phase may be retained,

or it may transform to a high yttria tetragonal phase and the low yttria equilibrium tetragonal phase

may transform to monoclinic. The estimated percentages of the phases present in the current

samples (and the yttria contents in each phase) are shown in Table 6.2. The percentage of yttria in

the tetragonal and cubic phases was deduced from the lattice parameters of each phase, according to

empirical equations (see section 5.3.4). The T’1 low-yttria tetragonal phase has a composition of


109

1.7wt% Y2O3, the high-yttria cubic (F) has 13wt% Y2O3 and the T'2 tetragonal phase has a

composition of 6.3wt% Y2O3. The T´ gradually decomposes at high temperature into the

equilibrium high-yttria cubic and low yttria-tetragonal phases equilibrium, but total decomposition

is not achieved after 100 h and some intermediate T´2 is still present. Decomposition occurs by

gradual segregation of the yttria into lower and higher yttria regions, with corresponding changes in

the c/a ratio. Of course, it must be recognized that some changes in phase constitution, and possibly

in the composition of the phases, might have occurred during cooling to room temperature. The

cubic phase formed at high temperature is retained at room temperature and contains about 13 wt%

yttria, which agrees more or less with the phase diagram, which suggests that at 1300oC the

equilibrium concentration of the cubic phase should be about 12 wt%. This concentration exceeds

the compositional limit of the T´ phase and hence the cubic phase F is retained to room temperature

[217, 328].

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.10 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated YSZ

(204NS-1) for 100 h at 1300oC.

The 204NS YSZ top coat was heat treated for 50 h at 1300o C. The evolution of the phase content

can be seen in Figure 6.11. After 1 h at 1300o C, the coating is still predominantly T´, but the peaks

appear to broaden. This is probably caused by compositional differences in the tetragonal phase due

to gradual diffusion of yttria, and corresponding changes in the lattice parameters. After 10 h of

heat treatment, further broadening of the tetragonal peaks is apparent and a small additional peak

appears which can be identified as the cubic peak. The trend continues for 20 h of heat treatment

and after 50 h it is possible to recognize two distinct tetragonal phases and the cubic phase. Figure

6.12 shows the deconvoluted profile the of 204NS YSZ top coat after heat treatment for 50 h at


110

1300o C. The T´ phase has also decomposed into a mixture of low-yttria tetragonal T'l high-yttira

cubic F and T’2 with an intermediate yttria composition. The percentages of each phase present are

summarized in Table 6.2. As heat treatment progresses, the amount of intermediate yttria T´2

tetragonal phase appears to gradually decrease, while the low yttria T´1 and high yttria cubic phases

F appear to increase. Comparing the percentages of phases present in this top coat to those present

in the YSZ 204NS-1 top coat after heat treatment for 100 h at 1300oC, the amount of cubic and low

yttria phases present is less in this top coat than in the YSZ 204NS-1, since the longer heat

treatment has allowed more decomposition of the high yttria T´2 tetragonal phase. A trace of

monoclinic phase is also evident, which has probably resulted from the transformation of some of

the low yttria T´1 tetragonal phase during cooling to room temperature.

It is not clear why there is some monoclinic phase present in the YSZ 204NS top coat, while none is

present in the YSZ 204NS-1 top coat, which has the higher impurity content. The tetragonal-to-

monoclinic transformation is associated with a volume increase which can be detrimental for the

failure of TBCs, although the amount of monoclinic phase (2.9%) present in the YSZ 204NS top

coat in the current study is probably not significant enough to cause failure. It is worth noting,

however, that the presence of SiO2 impurities probably aids the phase stability in the YSZ 204NS-1

powder.

28 28.5 29 29.5 30 30.5 31 31.5 32

as-sprayed1 hour2 hours5 hours10 hours20 hours50 hours

Inte

nsi

ty

2

72 72.5 73 73.5 74 74.5 75

as-sprayed1 hour2 hours5 hours10 hours20 hours50 hours

Inte

nsi

ty

2

Figure 6.11 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ showing phase evolution for heat treated

YSZ (204NS) up to 50 h at 1300oC.


111

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.12.XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated YSZ

(204NS) for 50 h at 1300oC.

The Al2O3-doped-YSZ top coat (AE9018) was heat treated for 50h at 1300oC. This top coat has a

similar composition and impurity level to the YSZ 204NS top coat, with Al2O3 being the only

significant difference. The T´ tetragonal phase decomposed to a mixture of low-yttria tetragonal T'l

high-yttira cubic F and T’2 with an intermediate yttria composition (Figure 6.14). Even though the

amount of cubic phase present was slightly less than the corresponding level in the YSZ 204NS top

coat given the same heat treatment, it can not be said conclusively that the tetragonal phase appears

to be more stable. In both top coats, the percentage of intermediate yttria T´2 tetragonal phase is

comparable, indicating that approximately the same amount of decomposition into high and low

yttria containing regions took place. The gradual decomposition of the T´2 into higher and lower

yttria regions often results in compositional variations, which make the observed peaks broader, but

not necessarily identifiable as distinct phases of low and high yttria content. The addition of

alumina does not appear to make a significant difference to the stabilization of the tetragonal phases,

although it appears that yttria diffusion appears to be slightly slower in Al2O3-doped YSZ than in

YSZ top coats with similar amounts of other impurities.

The Al2O3-doped-YSZ top coat (AE8170), which has a very similar composition to the Al2O3-

doped-YSZ AE9018 top coat, was heat treated for 50 h at 1350oC (Figure 6.13). The percentage of

cubic phase present in this case is slightly higher than that observed in the AE9018 top coat that was

heat treated at a slightly lower temperature (Table 6.2). This highlights the influence of the heat

treatment temperature on the decomposition rate of the T´ phase into its equilibrium phase. The

temperature difference of 50oC significantly affects the percentage of cubic phase formed. As

mentioned previously, compositional variations resulting from the progressive diffusion of yttria


112

into high- and low-yttria regions results in broadening of the observed peaks and sometimes it is

difficult to differentiate distinct phases and distinguish this from a single tetragonal phase exhibiting

a broad peak. In the current case, is possible to deconvolute the observed pattern into either a low-

yttria T´1, high yttria cubic F and intermediate T´2 yttria phases or simply into a tetragonal and a

cubic phase. Another possibility is the existence of a mixture of phases with small compositional

variations. Thus x-ray analysis cannot in this case be reliably used to establish the phase

constitution. Other techniques, such as TEM might be useful to provide further information.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.13 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated Al2O3-doped YSZ

(AE8170) for 50 h at 1350oC.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.14 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated Al2O3-doped YSZ

(AE9018) for 50 h at 1300oC.


113

6.1.2.2 Dysprosia stabilized zirconia top coats

The DSZ top coat (SPM6-2444) decomposes slowly during heat treatment and gradually a cubic

phase appears. The x-ray spectrum after 50 h at 1300oC is difficult to deconvolute and hence it can

not be said with certainty whether there are one or more tetragonal phases present (Figure 6.15). It

is almost certain that small compositional variations in the cubic and tetragonal intermediate phases

result in the observed broadening of the base of the two tetragonal peaks. According to the phase

diagram of this system [329], both the pyrochlore (P) phase Dy2Zr2O7 and the zirconia solid

solution phases co-exist when the concentration of Dy2O3 is above 15 mol%. No pyrochlore phase

was formed in the present work, due to the relatively low mol% of Dy2O3 present. The x-ray

spectrum of DSZ after heat treatment at 1350oC for 100 h shows a considerably higher amount of

the cubic phase F (Figure 6.16). Since the temperature and heat treatment time were higher, it

comes as no surprise that there is more cubic phase present. A trace of monoclinic is also present.

There are no well-established equations relating the peak position to the phase composition for

dysprosia (as there are for yttria) and hence is not possible to calculate the dysprosia content in

these phases. From the present results it can be concluded that for the current amount of stabilizer

(3.58 mol%), dysprosia successfully stabilized the tetragonal phase in plasma sprayed coatings.

After heat treatment a certain amount of decomposition occurred, similar to that in conventional

YSZ top coats.

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.15 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated DSZ

(SMP6-2444) for 50 h at 1300oC.


114

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.16 XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated DSZ (SMP6-

2444) for 100 h at 1350oC.

6.1.2.3 Yttria-lanthana stabilized zirconia top coats

In the YLaSZ top coat after heat treatment, the tetragonal phases initially present in the as sprayed

coating decompose into a mixture of tetragonal cubic and the cubic pyrochlore (P) phase La2Zr2O7

(Figure 6.17). The percentages of each phase were calculated using Rietveld analysis (see Table

6.2). The presence of the pyrochlore phase La2Zr2O7 in LaSZ has been reported previously [227].

When both lanthana and yttria are present, it appears that decomposition of the lanthana-doped

zirconia and yttria-doped zirconia occurs simultaneously and a mixture of tetragonal, cubic and

pyrochlore phases is present. The formation of the pyrochlore phase P accompanies a volume

increase, which is thought to enhance the tetragonal-to-monoclinic phase transformation [228, 330].

The presence of both lanthana and yttria makes it difficult to estimate the stabilizer content from the

peak position. La2Zr2O7 top coats consisting of the pyrochlore structure have been proposed as a

promising new TBC materials [230, 231, 331, 332], because of their stable structure up to the

melting points and lower thermal conductivity. Although, these top coats have lower cycling life

that YSZ top coats due to their lower CTE. However, in the present study the YLaSZ top coat was

not phase stable, probably due to the presence of both yttria and lanthana and it decomposed after

heat treatment forming a mixture of pyrochlore, cubic and tetragonal phases.


115

27 28 29 30 31 32

Inte

nsi

ty

2 (°)

M (111)

T'(111) & F(111)

M(111)

(a)

La2Zr

2O

7

Figure 6.17. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ (deconvoluted) for heat treated YLaSZ

(AE8321) for 50 h at 1350oC.

Phases (%)

Top Coat / Heat treatment Monoclinic M Cubic F Tetragonal T Tetragonal T1

(low yttria)

Tetragonal T2

(high yttria)

Pyrochlore P

204NS-1 / 100 h at 1300oC

Y2O3 wt%

46 13

- 35 1.7

19 6.3

-

204NS / 50 h at 1300oC

Y2O3 wt%

2.9 29.5 10

- 23.2 1.6

44.4 6.7

-

SPM6-2444 / 50 h at 1300oC 11 89 -

SPM6-2444 /100 h at 1350oC 2 30 68 -

AE8321 / 50h at 1350oC 39.2 37.3 12.5 11

AE8170 / 50 h at 1350oC

Y2O3 wt%

32 14

55 1.8

13 6.6

-

AE9018 / 50 h at 1300oC

Y2O3 wt%

17 14.6

- 35 1.7

48 6.7

-

Table 6.2. Phase constitution of heat treated detached thermal barrier coatings.

6.1.3 Heat Treatment Under a Thermal Gradient

Top coats of YSZ 204NS-1 (attached to the substrate) were heat treated in a thermal gradient for

48 h. The temperature gradient was monitored via thermocouples embedded in the substrate and by

a thermocouple embedded in a laser drilled hole in the top coat - see section 5.5.2. Typical top coat

temperatures ranged from about 1500oC at the free surface to 900

oC at the interface. The XRD

spectra from material near the free surface indicates a mixture of low- and high-yttria tetragonal

phases (T'1 and T'2), having compositions of 2.2wt% Y2O3 and 11.8wt% Y2O3 respectively, with no

residual cubic phase (Figure 6.18). No cubic phase has been retained down to room temperature,

while, as described previously, quite a significant amount is present after holding isothermally at

about 1300oC, even though the phase diagram indicates that more cubic material should be present


116

at the higher temperature. Similar behaviour has been reported [217, 328] previously, particularly

for high cooling rates. The cubic phase forming at higher holding temperatures is expected to have

a lower yttria composition than if it had formed at lower temperatures, which may render it more

liable to transform during rapid cooling to the high-yttria tetragonal T´2, whereas when the high

temperature cubic phase has a high enough yttria content, the cubic polymorph is retained down to

room temperature. On the other hand, the XRD spectrum for the top coat material near the interface

with the bond coat indicates that this material remained predominantly T´ non-transformable and no

phase transformation took place (Figure 6.19). This is not surprising, since this material was being

held at about 900o C.

27 28 29 30 31 32

Inte

nsi

ty

2 (o)

M (111)

T'(111) & F(111)

M (111)

(a)

72 73 74 75

Inte

nsi

ty

2 (o)

(b)

F(400)

T'(400)

T'(004)

Figure 6.18. XRD spectra in 2 ranges of (a) 27-32˚, and (b) 72-75˚ for YSZ (204NS-1) top coats, after

gradient heat treatment, near the interface with the bond coat (which was at about 900oC).

27 28 29 30 31 32

Inte

nsi

ty

2 (o)

M (111)

T'(111) & F(111)

M (111)

(a)

Figure 6.19 XRD spectra, with deconvoluted peaks and differential plots (showing the deviation between

the measured and modelled spectra) in 2 ranges of (a) 27-32o, and (b) 72-75

o for YSZ (204NS-1) top coat

after thermal gradient heat treatment, near the free surface (which was at about 1500oC).


117

6.2 Microstructural development under service conditions


The microstructures of as-sprayed zirconia top coats, while strongly dependent on the spraying

parameters, always show certain characteristics. Figure 6.20 to Figure 6.25 show SEM micrographs

of fracture surfaces of PSZ top coats, before and after heat treatment. As-sprayed coatings exhibit

the characteristic splat structure. Through-thickness microcracks and pores are also present.

Through-thickness microcracks are formed during cooling in order to relieve the quenching stress

experienced by the splats [59]. Pores ranging in sizes can be identified. At higher magnification

(Figure 6.21(b)) adherent splats seem to be in close physical proximity to each other. The grain

structure within individual splats is columnar, due to the directional freezing of the splats after

impact with the substrate.

Compositional differences (i.e. different stabilizers) do not cause changes in observed

microstructural features (Figure 6.22 to Figure 6.25). The dimensions of the splat thickness, inter-

splat pores, large pores and through-thickness microcracks do not appear to vary significantly with

composition, when the same spraying parameters are used. However it is possible that some

systematic differences exist in the microstructure of coatings of different composition, which would

be possible to identify and quantify with other techniques, e.g. small angle neutron scattering. This

was beyond the scope of this study.

No compositional inhomogeneities could be detected in any of the structures, using EDX analysis.

This was also true for the alumina-doped YSZ, where no alumina was detected by EDX.

Isothermal heat treatment resulted in grain growth in all top coats, often bridging across interfaces

between splats in close physical proximity (see Figure 6.21). There is also evidence of the healing

of through-thickness microcracks. It is worth noting that large pores do not appear to heal even

after prolonged heat treatment. No qualitative difference in the microstructure was observed for top

coats heat treated for more that 50 h at temperatures in the range 1300-1450o C (Figure 6.21 to

Figure 6.23). However, quantitative characterization of the microstructure of PS YSZ top coat

using SANS [201] reported significant decreases in the surface area of the void system (up to 33%)

as a result of sintering, at temperatures as low as 1000oC. Cracks appear to sinter first and at lower

temperatures, followed by sintering of intersplat pores at higher temperatures. Thermal exposure

however, had no influence on large (>10μm) pores [135]. This agrees well with the microstructural

features observed here.


118

(a)

(b)

Figure 6.20. SEM micrographs of fracture surfaces of as-sprayed top coat of YSZ (204NS-1) (a) low magnification

(b) higher magnification showing some characteristic microstructural features of PS top coats.

(a)

(b)

Figure 6.21 SEM micrographs of fracture surfaces of detached top coat of YSZ (204NS-1) heat treated for 100 h at

(a)1300oC and (b) 1350

oC.


119

(a)

.(b)

Figure 6.22 SEM micrographs of detached plasma sprayed top coats heat treated at 1450o C for 100 h (a)

YSZ (204NS-1) and (b) DSZ (SPM6-2444).

(a)

(b)

Figure 6.23 SEM micrographs of fracture surface of DSZ (SPM6-2444) detached top coat (a) as-sprayed and

(b) heat treated at 1350oC for 100 h.


120

(a)

(b)

Figure 6.24. SEM micrographs of fracture surface of YLaSZ (AE8321) detached top coat (a) as-sprayed

and (b) heat treated at 1350oC for 50 h (these coatings were sprayed by Sulzer).

(a)

(b)

Figure 6.25 SEM micrographs of fracture surface of Al2O3-doped YSZ (AE9018) detached top coat (a)

as-sprayed and (b) heat treated at 1350oC for 50 h (these coatings were sprayed by Sulzer).

6.2.2 Heat Treatment Under a Thermal Gradient

Under service conditions, TBCs operate with high thermal gradients. It is, thus, of interest to study

the microstructural changes that occur under these conditions. A YSZ top coat attached to a

substrate was heat treated under an imposed thermal gradient, for 17 h, with the top surface

temperature around 1500oC and the temperature near the top coat/substrate interface about 900

oC.

Near the TC/BC interface (Figure 6.26a), there was little or no sintering. Microcracks are still

present and bonding between splats is poor. Grains in individual splats have maintained their

columnar structure. Near the free surface of the TC, on the other hand (Figure 6.26b), pronounced


121

sintering has taken place. There has been extensive healing of microcracks and grain growth has

occurred, with many grains becoming both larger and more equiaxed in morphology.

(a)

(b)

Figure 6.26. SEM micrographs of YSZ top coat (204NS-1) after exposure to a high thermal gradient

for 7 h, (a) near the TC/BC interface and (b) near the TC outer surface.

(a)

)

(b)

Figure 6.27.Higher magnification SEM micrographs of YSZ top coat (204NS-1) after exposure to a

high thermal gradient for 17 h, (a) near the TC/BC interface and (b) near the TC outer surface.

(higher magnification).


122

6.3 Effect of Heat Treatment of Detached Top Coats on sintering

behaviour

6.3.1 Effect of Composition and Morphology on Sintering Characteristics

The effect of heat treatment on the sintering behaviour of detached TBCs was investigated by

monitoring linear dimensional changes during isothermal heat treatment. Dilatometry experiments

were carried out on detached top coats of different compositions. The net linear contraction -

L/Lo(%) in the through-thickness and in-plane direction is plotted in Figure 6.28 and Figure 6.29

against time at temperature for detached top coats of different composition.

Thermal contraction (shrinkage) occurs due to sintering of the coatings. Surface transport

mechanisms do not cause shrinkage [150]. Bulk transport processes which include grain boundary

diffusion, and volume/lattice diffusion, are responsible for the shrinkage observed. For all

specimens, sintering rates change with time. Faster shrinkage rates were observed initially,

followed by a fall to relatively constant rates. The fast shrinkage observed at the early stages, even

under conditions where sintering models [151, 160-163] predict a dominance of surface diffusion,

demonstrates that volume and/or grain-boundary diffusion mechanisms are occurring as well.

Similar trends in the sintering behaviour have been observed by other researchers [182, 197, 198].

However the anisotropy between the in-plane and through-thickness contractions has not been

reported previously. Eaton and Novak [197] performed similar measurements on YSZ and other

coatings, and observed similar behaviour. Sintering of the top coat will result in enhanced intersplat

bonding, lower strain tolerance and higher stiffness of the coating. Previous work [60, 197] has

indicated that a sharp increase in stiffness occurs after a short period of exposure to high

temperature, followed by progressively slower increases. These observations are broadly in

agreement with the current dilatometry results.


123

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 20 40 60 80 100 120 140 160

Time (h)

Net

line

ar

co

ntr

actio

n (

%)

YSZ (204NS-1)

CSZ (205NS)

DSZ (SPM6-2444)

YSZ Al2O3-doped (AE9018)


YLaSZ (AE8321)

YSZ (204NS)

Al2O3-doped

Al2O3-doped

Figure 6.28 Net linear contraction from dilatometry data for detached PS top coats heat treated at 1350

oC in the

through-thickness direction.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0 20 40 60 80 100 120 140 160

Time (h)

Net

linear

contr

action (

%)


YSZ (204NS-1)

CSZ (205NS)

DSZ (SPM6-2444)

YLaSZ (AE8321)

YSZ (204NS)

Al2O3-doped

Figure 6.29 Net linear contraction from dilatometry data for detached PS top coats heat treated at 1350

oC in the in

plane direction.


124

Coatings with different compositions showed significantly different sintering rates. To analyze the

sintering behaviour of differently doped zirconia, the factors affecting bulk diffusion must be

considered. Shrinkage during sintering occurs by grain boundary diffusion and volume/lattice

diffusion. Shrinkage can also occur by dislocation diffusion, but since the contribution from this

mechanism is not likely to be significant in the present case we will not mention this mechanism

any further. At an atomic scale, sintering and shrinkage occur via mass flux of atoms, which is

driven by a gradient of chemical potential. The mass flux for sintering, which is derived from

Fick’s first law is

( )b

J Dk T

(6.1)

where D is the diffusivity or diffusion coefficient, Ω is the volume per lattice site, kb is Boltzmann’s

constant, T is the temperature and () the gradient of hydrostatic stress. The stress is dependent

on the surface free energy, which may be altered by impurities or dopants absorbed at the surface.

Equally, the volume per lattice site Ω is a structural parameter, dependant on the atomic and crystal

structure. The diffusivity D is specific to each diffusion mechanism and it follows an Arrhenius-

type equation. From the above equation, it can be noted that there is a direct dependence of the

atom flux during solid state sintering on structural parameters that can be influenced by dopants and

impurities. In the case of an ionic compound, transport of anions and cations during solid state

sintering by either mechanism occurs with the help of the defects existing in the solid and with the

help of the gradient concentration of these defects existing between the surface and the bulk, or

between the bulk and the grain boundaries [147, 333]. The defects in a solid are produced either as

a result of thermal equilibrium (intrinsic) or as a result of suitable doping (extrinsic) [334]. Intrinsic

defects can be ions occupying an interstitial site (Frenkel defects) or cation and anion vacancies

with no interstitial or misplaced ions (Shottky defects). The number of defects is correctly balanced

as to maintain the stoichiometric formula and preserve electrical neutrality. In sintering of ionic

compounds, the various constituents must diffuse is the stoichiometric ratio of the compound [335].

Thus, the rate of diffusion is regulated by stoichiometric restrictions and it will be determined by

the slowest atomic species [156].

In YSZ, and equally in zirconia stabilized by other trivalent oxides, the introduction of aliovalent

components introduces defects. The majority of defects in YSZ are oxygen vacancies and yttrium

aliovalent dopants on normal cation sites. The slowest and rate controlling diffusional process in

yttria stabilized zirconia is confirmed to be transport of zirconium and yttrium cations [180, 181].

The oxygen vacancies have far lower activation energy for diffusion than the solute cations. The


125

latter are expected to act as the controlling species for grain growth [180-182]. The defect reaction

in YSZ can be written, using the Kröger-Vink notation [334], as

2ZrO ' ..

2 3 Zr o oY O 2Y + 3O + V (6.2)

Other possible defects include zirconium interstitials, oxygen interstitials, yttrium interstitials and

zirconium vacancies. For stabilized zirconia with other trivalent ions, the same principle applies to

generation of defects and controlling diffusant species, since it has been confirmed the dopant

trivalent ions in zirconia occupy the cation sites [204]. Comparing the diffusion coefficients of Zr4+

and O2-

in ZrO2, the Zr4+

is the more stable, slower diffusing species and is therefore expected to be

the rate controlling species in sintering.

Element Diffusion Coefficient (m2 s

-1)

Zr

10-19

O 210-13

Table 6.3 Reported diffusion coefficient of Zr and O

in ZrO2 [333]

The presence of extrinsic defects will affect the transport properties in an ionic crystal. The precise

effect of these defects on diffusion depends on the defect and ion mobility [336]. Sintering rates

should increase if the concentration of slow defects is increased [191, 333]. In the YSZ the slowest

diffusing species is the Zr4+

ion. The mechanism of cation diffusion in YSZ is disputed in the

literature. Kilo et al [183, 184]support that diffusion of cations in YSZ occurs via zirconium

vacancies formed by the creation of Schottky defects. In this case the presence of additional

oxygen vacancies should not affect the cation diffusion. Chien et al [180] support that cation

diffusion occurs by cation vacancies associated with oxygen vacancies. In this case, the addition of

oxygen vacancies might affect cation diffusion. Therefore, it is not clear how the presence of

oxygen vacancies should affect the cation diffusion rate in YSZ. The same mechanism that governs

cation diffusion in YSZ is expected to govern the diffusion of other trivalent lanthanides in zirconia

[337].

For CSZ, the stabilizing species have the same valence as the host ion and so no extrinsic defects

should be introduced. In these cases the intrinsic thermal defects contribute to the diffusion process

[335]. As with the trivalent stabilized zirconias, the cations are expected to be the rate controlling

species for sintering.

Another bulk diffusion mechanism, that also contributes to the shrinkage observed, is grain

boundary diffusion. During grain boundary diffusion, mass removed along the grain boundary is


126

re-deposited at the sinter bond. Shrinkage is realized because of mass transport from the grain

boundary to the free surfaces [147]. As surface energy is consumed by shrinkage of pores, new

grain boundaries emerge at the sinter bonds between splats. If the grain boundary energy is high,

the replacement of free surfaces by grain boundaries might be unfavorable. Segregation of Y, La,

Ca and Ce solute dopants at the grain boundary in zirconia has been observed by various

researchers [183, 186-190]. Segregation of solid solution additives at the grain boundaries lowers

grain boundary energy and aids shrinkage of pores by grain boundary diffusion [149, 155, 191].

In the absence of a solid solution additive, grain growth occurs in order to reduce the total grain

boundary area [149]. The disappearance of grain boundaries due to grain growth will reduce

shrinkage by grain boundary diffusion. Therefore high mobility grain boundaries, which promote

grain growth, are not favorable for high shrinkage. Segregation of solid solution additives at the

grain boundaries, lowers the grain boundary mobility, and contributes to higher shrinkage [149,

155]. The size of the doping cations appears to influence their effect on grain boundary mobility

and grain growth [186]. For stabilized zirconia with different cations, including La3+

, Y3+

and Ce4+

,

larger, low valence cations are more effective in reducing the grain growth, in agreement with the

solute drag models [153, 154]. Hence top coats with, larger, low valance stabilizing cations should

exhibit higher sintering rates. The mobility of the grain boundaries, with respect to the solute drag

mechanism, is given by [154]

2C

D

T (6.3)

where μ is the grain boundary mobility, D is the diffusivity of the solute cation, T is the temperature

and C is the excess solute concentration at the grain boundaries. The highest sintering rates should

be exhibited by coatings in which the doping cations decrease more effectively the grain boundary

mobility. Thus, since the cations are the rate controlling species for sintering, the doping cations

should also have intrinsic low diffusivity. In addition, effective dopants for decreasing the sintering

rate should be soluble within zirconia [338], and not segregate at the grain boundaries. The

solubility limit depends both on radius and valence differences between Zr and the dopant cation.

The solid solubility limits decrease with increase in the ionic misfits (see Table 6.4).

Extensive work by Li and Chen[204, 208, 339, 340] confirmed that a common feature to all the

stabilizing cations is that, when in solution, they maintain the nature of the first neighbor bonds that

they have in the pure stabilizing oxide. Therefore, in the absence of diffusivity data, the melting

temperature of the stabilizing dopant reflects the binding energy [341] and is an indicator of

diffusion rates. Lower melting point materials have weaker atomic bonding and higher molar


127

volume which allows faster diffusion at a given temperature. Hence, for sintering inhibition, oxides

with a high melting temperature, good solubility in zirconia and low ionic misfit would be most

effective in reducing the sintering rate (see Table 6.4).

Dopant oxide or

impurity

Ionic radius

(nm)

Solid solubility

(mol%)

Melting point

(o C)

Reference

ZrO2(host material) 0.084 - 2715

Y2O3 0.102 2.9 2690 [205]

La2O3 0.116 1.15 2305 [227]

CeO2 0.097 18 2400 [307]

Dy2O3 0.103 2408

Al2O3 0.051 0.5 2054 [323] Table 6.4 Ionic radii, solid solubility limits and melting points of dopant oxides or impurities.

The lowest and highest shrinkage rate are exhibited by YSZ (204NS) and YSZ (204NS-1), the top

coat with the lowest and highest amount of impurities, respectively. Impurities which are not in

solid solution, can provide a continuous low diffusivity pathway at the grain boundary which will

restrain grain boundary mobility and reduce grain growth and enhance shrinkage [153], or, provide

a high diffusivity pathway, e.g. liquid phase at the grain boundaries, which will enhance shrinkage

[172]. The effect of impurities on the sintering rate and grain growth, and particularly the role of

SiO2 and Al2O3, has been addressed by several researchers [186, 192, 193, 196, 198, 338, 342].

Vassen et al [198] found an increase in the shrinkage rate with increased silica and alumina content,

but the effect was more pronounced for SiO2. This is in agreement with the results presented here,

where the sintering rate of YSZ (204NS-1) with 0.1 wt% SiO2 is relatively high. Both the Al2O3-

doped YSZ top coats also exhibit higher sintering rates than undoped pure 204NS. Verkerk et al

[192] found that small additions of alumina reduced grain growth by the solute drag effect.

Matsumoto et al [343] reported that the addition of 1 mol% Al2O3 to 4 mol% YSZ resulted in rapid

densification of EBPVD TBCs after sintering for 10 h at 1300o C. However they supported the

argument that the presence of a liquid phase is unlikely. Srdic et al [196] also found that a small

amount of Al2O3 in zirconia suppressed grain growth and enhanced densification.

Thermal contraction is anisotropic for all materials, with greater shrinkage rates in the through

thickness direction (see Figure 6.30 to Figure 6.33). This anisotropic behaviour is attributed to the

relative amount of surface area inter-splat pores and intra-splat microcracks. PS top coats have

greater surface area of inter-splat pores, compared to intra-splat cracks [131, 136]. According to

Wang et al [131], shrinkage of the inter-splat interfaces may have greater influence on the thermal


128

conductivity than on the elastic modulus. Their role as a barrier to heat transfer may be more

significant than as a barrier to force transfer [131].

0

0.5

1

1.5

2

0 20 40 60 80 100 120 140 160

in plane at 1350o C

through thickness at 1350o C

in plane at 1450o C


Net

Lin

ear

Contr

acti

on (

%)

Time (hours)

0

0.1

0.2

0.3

0.4

0.5

0 20 40 60 80 100

in plane at 1300o C


in plane at 1350o C


Net

Lin

ear

Contr

acti

on (

%)

Time (hours)

Figure 6.30 Net linear contraction for detached top

coats of YSZ (204NS-1) heat treated at 1350oC and

1450o C, showing anisotropy of contraction.


coats of YSZ (204NS) heat treated at 1300oC and


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 20 40 60 80 100 120

in plane at 1350o C


in plane at 1450o C


Net

Lin

ear

Con

trac

tion

(%

)

Time (hours)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 20 40 60 80 100

in plane at 1350o C


in plane at 1450o C


Net

Lin

ear

Con

trac

tion

(%

)

Time (hours)

Figure 6.32Net linear contraction for detached top coats

of DSZ (SPM6-2444) heat treated at 1350oC and



coats of CSZ (205NS) heat treated at 1350oC and



129

6.3.2 Effect of Phase Transformation on Measured Volume Changes

An estimate can be made of the magnitude of volume change associated with the relevant phase

changes, or at least an upper bound can be placed on them. There is a particular interest in the

tetragonal-to-cubic transformation, since this is likely to occur during extended heating at high

temperature. This can be seen from the phase diagram (Figure 3.12a) [306]. The volume change

associated with the tetragonal-monoclinic transformation is well established [212] in literature, to

be about 4%. For the tetragonal to cubic transformation, since the number of atoms per unit cell is

the same for (all of the possible) tetragonal and cubic phases, to estimate the volume change it is

only necessary to establish the volume of the unit cell in each case. The volume of the unit cell in

each case can be obtained using lattice parameter values inferred from measured X-ray peak

positions†. The composition of the cubic and tetragonal phases can also be deduced from these

measured lattice parameters using equations 5.3 and 5.4. Measured peak positions, lattice

parameters and unit cell volumes for YSZ (204NS-1) and YSZ (204NS) after various heat

treatments are given in Table 6.5 and Table 6.6.

The volume contraction associated with the phase changes in YSZ (204NS-1) after heat treatment at

1300o C for 100 h is about 0.5 %. The actual shrinkage observed after cooling to room temperature,

derived from the linear contractions, amounts to about 1.9 %. Hence microstructural effects due to

sintering, and not phase changes, must be primarily responsible for the contraction observed.

For the YSZ 204NS-1 top coat, no transformation of the monoclinic phase was observed and, since

this transformation has the most significant volume change, the volume changes due to phase

transformations associated with the YSZ (204NS) top coat, which contained 2.9 (%) monoclinic

phase after 50 h at 1300oC, were investigated. In this case, the volume change associated with

phase transformation after heat treatment is ΔV/Vo = -0.35%. The actual shrinkage observed after

cooling to room temperature is ΔL/Lo = -0.26% in the in-plane direction and ΔL/Lo = -0.73% in the

through-thickness direction, which amounts to a total volume change of ΔV/Vo = -1.2%. Hence,

the shrinkage observed can be attributed primarily to microstructural changes associated with

sintering effects.

Phase /

Treatment

tetragonal T' /

as-sprayed

tetragonal T'1 /

100 h @ 1300˚C

tetragonal T'2 /

100 h @ 1300˚C

cubic F /

100 h @ 1300˚C

†Of course, the X-ray data were obtained at room temperature, whereas interest centres on volume at high

temperature. However, the difference in thermal expansivities between cubic and tetragonal phases is apparently [366]

less than 1 10-6

K-1

, so the possible error from this source must be less than about 0.1% in linear dimensions.


130

2 for (004) K1 73.139˚ 73.030˚ 73.236˚ -

2 for (400) K1 73.981˚ 74.447˚ 74.217˚ 73.738˚

c (nm) 0.51737 0.51803 0.51678 -

a (nm) 0.51231 0.50956 0.51091 0.51375

wt% Y2O3 7.8 1.7 6.3 13

vol. of unit cell (nm3) 0.13579 0.13451 0.13489 0.13561

Composition (%) 100 35 19 46

V/V (%) 0 -0.94 -0.66 -0.13

Table 6.5. Measured X-ray peak positions, deduced lattice parameters and associated unit cell volumes for

phases within YSZ (204NS-1) coating, with and without a prior heat treatment. Also shown are percentages

of phases present and phase compositions from the lattice parameters, using eqns.5.3 and 5.4.

An upper bound on the expected volume change can be obtained by assuming that thermodynamic

equilibrium is attained at the heat treatment temperature. For example, at ~1300˚C, up to ~50% of

the tetragonal phase could be transformed to the cubic phase, for material with an overall

composition of about ZrO2-7wt%Y2O3. It can be seen that, a volume contraction of up to about

0.5–0.6% might be expected when tetragonal material containing about 7wt% Y2O3 transforms to

50% of F phase containing 13wt% Y2O3 and 50% of T'1 phase containing 2wt% Y2O3. The

corresponding linear contraction is thus about 0.2%. It can be concluded that phase changes are

expected to make only relatively small contributions to dilatometry data, indicating linear

contractions of the order of 1% or more. This calculation thus also confirms that the contraction

can be attributed primarily to microstructural changes associated with sintering effects. If anything,

the effect of taking phase transformation-induced linear contraction into account would be to

increase the deduced anisotropy of the sintering contraction.

Another interesting observation on the measured linear shrinkage is the difference observed

between the total linear shrinkage observed at temperature and the total linear shrinkage observed

after cooling at room temperature. For the YSZ (204NS) top coat heat treated at 1300o C, in the in-

plane direction, there is a difference of about 0.15 (%) and in the through-thickness direction, of

about 0.47 (%) (see Figure 6.35). In order to confirm that these differences were genuine,

calibration with a blank alumina sample was performed, which, as expected, showed no difference

between the shrinkage observed at high temperature and after cooling (the shrinkage in both cases

was zero) (see Figure 6.34). This difference is attributed to phase changes occurring during cooling

down, particularly the tetragonal to monoclinic transformation, which is known to have a 4 %

volume change.


131

Phase /

Treatment

tetragonal T' /

as-sprayed

tetragonal T'1 /

50 h @ 1300˚C

tetragonal T'2 /

50 h @ 1300˚C

cubic F /

50 h @ 1300˚C

Monoclinic/

50 h @ 1300˚C

2 for (004) K1 73.1060 73.3219 - -

2 for (400) K1 74.5336 74.2704 73.8420 -

c (nm) 0.51696 0.51757 0.51626 - -

a (nm) 0.51156 0.50905 0.51060 0.51313 -

wt% Y2O3 7.3 1.6 6.7 10 -

vol. of unit cell

(nm3)

0.13528 0.13412 0.13459 0.13511 -

Composition (%) 100 23.2 44.4 29.5 2.9

V/V (%) 0 -0.86 -0.51 -0.13 4

Table 6.6 Measured X-ray peak positions, deduced lattice parameters and associated unit cell volumes for phases

within YSZ (204NS) coating, with and without a prior heat treatment. Also shown are percentages of phases present

and phase compositions from the lattice parameters, using eqns.5.3 and 5.4.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0

200

400

600

800

1000

1200

1400

0 1 2 3 4 5

L/Lo for T= 600

o C

L/Lo for T= 1300

o C

T = 600o C

T = 1300o C

L

/Lo (

%)

Tem

peratu

re (o C

)

Time (h)

-1

-0.5

0

0.5

1

1.5

2

2.5

3

0

200

400

600

800

1000

1200

1400

0 10 20 30 40 50 60

through-thicknessin-plane

Temperature

L

/Lo (

%)

Tem

peratu

re (oC

)

Time (hours)

YSZ (204NS)

Figure 6.34. Dilatometry calibration plots for alumina,

showing no difference in the shrinkage observed at high

temperature and after cooling to room temperature.

Figure 6.35. Dilatometry plots for YSZ (204NS) in-plane

and through-thickness, showing the difference in the total

linear shrinkage observed at high temperature and after

cooling to room temperature.

6.3.3 Effect of Thermal Cycling on Sintering Behaviour

Thermal barrier coatings under service conditions, particularly in the aerospace industry are

thermally cycled routinely. The effect of thermal cycling on the sintering behaviour was examined


132

using the dilatometry technique. Detached top coats of YSZ (204NS) were thermally cycled and

the shrinkage was measured in-situ in the in-plane direction. The thermal cycle performed and the

shrinkage observed is shown in Figure 6.36. The individual high temperature segments are plotted

cumulatively in Figure 6.37 and compared with the shrinkage observed during continuous heat

treatment for the same time. During thermal cycling, faster shrinkage is observed during the initial

cycles, followed by slower shrinkage rates at later cycles, as is the case for continnous heat

treatment. Thus the tendency for high initial sintering rates is not altered by thermal cycling.

However, surprisingly, the shrinkage observed during thermal cycling is less than that observed

when heat treatment happens continuously, for the same amount of time at high temperature.

Cooling down and re-heating appears to slow down the shrinkage, possible due to phase changes

occurring during repeated cool down-heat up cycles which might affect the diffusion processes at

high temperature, which do not resume at the same rate.

-0.5

0

0.5

1

1.5

2

0 10 20 30 40 50 60 70 80

Time (h)

L

/Lo (

%)

0

200

400

600

800

1000

1200

1400

Te

mp

era

ture

(oC

)

dL/Lo (%)

Temperature (oC)

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0 10 20 30 40 50 60Time (hours )

- L

/Lo

(%

)

50h continuous

0-1 hours

1-2 hours

2-5 hours

5-10 hours

10-20 hours

20-50 hours

1300oCYSZ (204NS)

Figure 6.36. Linear shrinkage observed during thermal

cycling of detached YSZ (204NS) top coat in the in-plane

direction.

Figure 6.37. Comparison between linear shrinkage observed

during heat treatment of detached YSZ (204NS) top coat at

1300oC, carried out continuously and during thermal cycling.

6.3.4 Effect of Sintering on Coating Porosity

Porosity measurements on as-sprayed and isothermally heat treated PSZ samples were conducted to

complement the dilatometry results, since it is well known that relative shrinkage during sintering is

related to the initial porosity of the material, and the higher the initial porosity, the higher the

shrinkage observed [147]. The theoretical density was calculated from the unit cell volume and the


133

atomic weight of the elements present, as described in section 5.3.5. The theoretical density was

calculated for the as-sprayed coatings. Where more than one phase was present in the as-sprayed

coating (as in the YLaSZ coating), the unit cell volume of each of these phases and the relative

percentage of each phase, was calculated in order to derive the overall density. The theoretical

densities after heat treatment, taking into account the phase changes, were not calculated. Since the

error involved in the measurement of porosity is of the order of the change in porosity due to heat

treatment, no additional information could be obtained by more detailed evaluation of the

theoretical density. The porosity levels of as-sprayed top coats of all materials are presented in

Figure 6.38. The porosity of all coatings is similar with a mean between 10-12 % and a standard

deviation of about 3-4 %. Since the porosity of plasma sprayed coatings does not differ much with

different materials, it can be concluded that the trends observed during dilatometry are not due to

difference in initial porosity levels of the samples, but probably represent inherent difference in

their sintering characteristics.

Top Coat Material YSZ

(204NS-1)

YSZ

(204NS)

DSZ

(SMP6-2444)

CSZ

(205NS)

YLaSZ

(AE8321)

doped-YSZ

(AE8170)

doped-YSZ

(AE9018)

Theoretical density

(gr cm-3

) 6.0429 6.0406 6.2953 6.3658 6.1136 6.04980 6.0632

Table 6.7. Theoretical density of as-sprayed top coats deduced from X-ray data and top coat composition.

Figure 6.39 is a plot of the porosity as a function of heat treatment time at 1350o C and at 1450

o C,

for YSZ(204NS-1) and DSZ detached top coats. Porosity decreases slightly with heat treatment, and

the change is the order of 3-4%. This observation is more or less consistent with the linear changes

observed during dilatometry for these coatings after the same heat treatment time at 1350 oC and

1450 oC, where linear contraction in the through-thickness direction is of the order of 1.2-1.7% and

in the-plane direction, of the order 0.3-0.5%, which corresponds to a volume contraction of about

2-3%. It is not expected that it would be possible to differentiate between the changes in porosity

for different top coats (YSZ and DSZ), since the errors involved in this measurement are larger than

the difference in the shrinkage rate between the top coats. For the same reason, no significant

difference between the changes in porosity after heat treatment at 1350o C and 1450

oC is apparent.

The small changes in porosity are consistent with the microstructural features observed after heat

treatment (see section 6.3). Most of the porosity is in the form of large pores that do not appear to

sinter with heat treatment. Changes in porosity are, therefore, due to sintering of finer pores.


134

5

7

9

11

13

15

17

19

Po

rosity (

%)

YSZ (204NS-1)

YSZ(204NS)

CSZ (205NS)

DSZ (SPM6-2444)

YLaSZ (AE321)

doped-YSZ (AE8170)

doped YSZ (AE9018)

Average

Figure 6.38. Porosity measurements for as-sprayed top coats.

4

5

6

7

8

9

10

11

12

13

14

0 20 40 60 80 100 120 140 160

Heat treatment time (h)

Poro

sity (

%)

YSZ heat treated at 1350°C

DSZ heat treated at 1350°C

YSZ heat treated at 1450°C

DSZ heat treated at 1450°C

Figure 6.39 Porosity measurement of YSZ and DSZ top coats after heat treatment at 1350oC and 1450

oC.


135

6.4 Conclusions

The following conclusions can be drawn from this work.

1 Exposure to high temperatures induced phase changes in all top coats studied, which may affect

the microstructure and perhaps the stress state in TBCs. The YSZ as-sprayed top coats are

entirely the so-called “non-transformable tetragonal” T' phase. At temperatures above 1300oC

they decompose to their equilibrium tetragonal (T) and cubic (F) phases. On cooling to room

temperature the coatings consist of mixtures of cubic (F) phases, and high-yttria tetragonal

phases (T'1 and T'2). The presence of SiO2 and Al2O3 impurities appears to decrease the rate of

decomposition.

2 Dysprosia, in the amount present in the current study (3.58 mol%), successfully stabilized the

tetragonal phase in plasma sprayed zirconia coatings. After heat treatment, a certain amount of

decomposition occurred, similar to that in conventional YSZ top coats.

3 YLaSZ as-sprayed top coats consisted of two tetragonal phases. After heat treatment, the

YLaSZ top coat was not phase-stable, probably due to the presence of both yttria and lanthana

and it decomposed to a mixture of cubic (F), high-yttria tetragonal phases (T'1 and T'2), and the

pyrochlore (P) phase.

4 The Ce4+

ion in the CSZ top coat was reduced to Ce3+

during annealing at 500oC in vacuum. As

a result, the non-transformable tetragonal T' phase decomposed into the cubic (F) phase. The

cerium valence state has a great influence on the phase composition, and it is believed to affect

the stress state in the coating.

5 The high thermal gradients commonly present in TBCs under service conditions lead to widely

varying conditions within the top coat. Sintering is likely to occur much more rapidly near the

free surface, not only because the temperatures are higher there, but because the residual

stresses tend to be more compressive. After heat treatment in a high thermal gradient, material

near the bond coat remained entirely T', whereas material near the free surface was composed of

a mixture of low- and high-yttria tetragonal phases (T'1 and T'2), which formed on cooling from

the high temperature tetragonal (T’) and cubic (F) phases, respectively.

6 Detached top coats exhibit substantial reductions in linear dimensions following exposure to

elevated temperatures. These linear contractions are greater in the through-thickness direction

than the in-plane directions. This is thought to be at least primarily associated with

microstructural changes induced by sintering processes and the anisotropy is a consequence of


136

the splat and pore geometry in these coatings. Any contributions arising from phase changes

occurring at high temperature are expected to be relatively small.

7 Stabilizers affect the shrinkage rate of top coats. It is thought that this is primarily due to

changes in the diffusion kinetics, since the microstructural characteristics of all the top coats

studied where qualitatively similar. Stabilizers may affect the volume diffusion coefficient,

grain boundary diffusion coefficients, grain boundary mobility or alter the ratio of grain

boundary to surface energy. Any of these could lead to differences in the shrinkage observed.

8 The impurity level, specifically that of SiO2, significantly increases the shrinkage observed. It

is thought that impurities which are not in solid solution provide a continuous low diffusivity

pathway at the grain boundary, which will restrain grain boundary mobility, reduce grain

growth and enhance shrinkage. Alternatively, impurities may also do so by providing a high

diffusivity pathway, e.g. liquid phase at the grain boundaries, which will enhance shrinkage.

9 Thermal cycling, commonly experienced by TBCs under service conditions, affects their

shrinkage. The shrinkage observed during thermal cycling is less than that observed under

continuous heat treatment. Cooling down and re-heating appears to slow down the shrinkage,

possibly due to phase changes occurring during repeated cool down-heat up cycles which might

affect the diffusion processes at high temperature, which do not resume at the same rate.

Chapter 7: Thermomechanical Behaviour of PS TBCs

137

7 Thermomechanical Behaviour of Plasma Sprayed

TBCs

7.1 Effect of Top Coat Sintering on Mechanical Properties

7.1.1 Cantilever Bending

This technique, which measures the global in-plane stiffness of the coating, was applied to detached,

as-sprayed top coats. A typical loading/unloading plot is presented in Figure 7.1. The hysteresis

observed could be attributed to the opening of cracks during initial loading, that fail to close when

the load is removed. Diminishing hysteresis was observed after loading and unloading several

times. The average value for the gradient of a load/unload/load/unload sequence was calculated.

This value was used to determine the stiffness, by applying equation 5.8. The stiffness of as-

sprayed zirconia-based top coats was in the range of 10-15 GPa (see Figure 7.2). It is worth noting

that elastic moduli of PS top coats depend on microstructure, which in turn depend on the

processing parameters, so values reported in the literature vary over a wide range corresponding to

the different parameters used (see section 3.2.1). However, the global stiffness of PS top coats is

expected to be considerably lower than that of the corresponding dense ceramic, due to the

microcracks, interlamella pores and the poor intersplat bonding of as-sprayed coatings. After

isothermal heat treatment for 88 h at 1350o C in air, the elastic modulus increases significantly due

to sintering effects (see Figure 7.2). Similar effects of heat treatment on stiffening have been

previously reported [60, 112].

It is known that the apparent stiffness of these materials varies with applied stress level, tending to

be higher under compressive loading, which closes the microcracks. During cantilever bending, the

detached top coat is placed both under tension and compression. In the results presented here, this

effect is neglected, but it may be noted that the resultant error in stiffness is probably [344] of the

order of ±5 GPa. This is a significant uncertainty, but the observed trend of increasing stiffness

after heat treatment is nevertheless quite clear. No significant difference is observed for coatings of

different compositions. However, due to the errors involved, the differences in the sintering and

stiffening behaviour between top coats of different compositions can not be precisely evaluated

with this technique.


138

7.1.2 Nanoindentation

7.1.2.1 As-sprayed and After Isothermal Heat Treatment

Nanoindentation measurements were performed on as-sprayed detached top coats of YSZ (204NS-1)

in the through thickness direction. The indents were located in positions remote from visible cracks

in the polished surface. The results obtained give an average stiffness value of 130 ± 20 GPa. It is

expected that the value of Young’s Modulus given by indentation would approach the value of the

monolithic material [61] since the region undergoing deformation is small and thus expected to be

free of significant defects. The slightly lower values measured are possibly due to some

accommodation of the indent by intersplat porosity or cracks. In other techniques, such as cantilever

bending (see section 7.1.1), the volume of material tested is more representative of the material as a

whole. Similar values for the Young’s modulus by nanoindentation of PS YSZ top coats have been

found in-plane (144 GPa) [109].

After heat treatment at 1300oC for 100 h, the stiffness was found to be 205 ± 10 GPa, which is close

to the value for bulk zirconia. Similar values have also been found in the in-plane direction [109].

This suggests that sintering has caused healing of microcracks and interfaces between splats and

thus, the effect of these features in the measurement is reduced.

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0 0.2 0.4 0.6

Displacement (mm)

Fo

rce

(N

)

Load

Unload

0

10

20

30

40

50

60

70

80

alumina-doped

YSZ (AE8170)

DSZ

(SPM6-2444)

YSZ

(204NS)

YSZ

(204NS-1)

Yo

un

g's

Mo

du

lus E

(G

Pa

)

As sprayed Heat treat for 88 h at 1350° C

Figure 7.1. Typical load/unload plot for as as-sprayed

YPSZ top coat during cantilever bend testing.

Figure 7.2. Young’s modulus of zirconia-based detached top

coats as sprayed and after 88 h at 1350oC.


139

7.1.2.2 Heat Treatment Under a Thermal gradient

Measurements were made on a top coat that had been heat treated for 17 h with a high through-

thickness thermal gradient. Data are shown in Figure 7.3. A through-thickness gradient in the

stiffness is apparent. Near the TC/BC interface, values were similar to those obtained with the as-

sprayed top coat, but near the outer surface the average value increases, approaching that of

monolithic dense YSZ. There is, of course, quite a lot of scatter in the data, since the value

obtained is sensitive to the presence or absence of neighbouring fine scale flaws. These results are

consistent with the microstructural observations, indicating pronounced sintering near the outer

surface. The absolute values of Young’s modulus obtained using this technique are, as expected,

much higher than those given by methods that measure the global stiffness, since gross flaws affect

the latter, but not the former. Also, the indentation method senses the compressive stiffness only,

which is known to be larger than that under tension in these materials, particularly at relatively high

strains.

0

50

100

150

200

250

0 0.5 1 1.5 2 2.5 3

Sti

ffnes

s (G

Pa)

Distance from TC/BC interface (mm)

Figure 7.3. Stiffness data obtained by nanoindentation on a polished transverse section of a top coat

after heat treatment in thermal gradient (~900˚C at the interface, ~1500˚C at the free surface) for 17 h.

7.2 Microstructure and Properties of Thick As-sprayed TBCs

Thick TBCs provide greater temperature drops across the coating, which offers advantages with

respect to reduction of the cooling air required by hot components or allows increases in gas turbine

inlet temperature, thus increasing the efficiency and performance of the engine. However thick


140

TBCs show low thermal shock resistance [345]. With increased thickness, the elastic strain energy

stored in the coating increases and hence the driving force for spallation increases. Microstructural

modifications that induce microcracks and porosity increase the thermal shock resistance [68, 346].

7.2.1 The Effect of Substrate Temperature on Microstructure and Stiffness

Thick YSZ top coats (about 1.5 mm) were produced by PS. Coatings sprayed at four different

average substrate temperatures were examined. The substrate temperature was controlled by

cooling the back surface of the substrate during spraying with a jet of compressed air and/or by

waiting 35 s in-between each gun pass. The average backsurface temperature during spraying for

the different spraying conditions is shown in Figure 7.8. The temperature of the substrate changes

continuously during spraying. The error bars in Figure 7.8 indicate the maximum and minimum

temperatures of the backsurface of the substrate during spraying. The four samples produced have

been labelled as A, B, C and D with decreasing average substrate temperature. The sprayed

specimens were vacuum impregnated with epoxy, ground and finely polished. Cross sections of the

samples were examined by optical microscopy, and fracture cross sections were also prepared for

microstructure characterization. The polished cross sections are presented in Figure 7.4 to Figure

7.7. The microstructural features observed can be defined as follows: (i) segmentation cracks

(cracks running perpendicular to the coating surface and penetrating at least half the coating

thickness), (ii) branching cracks (cracks parallel to the coating plane, starting from segmentation

cracks), (iii) microcracks (other cracks in the coating not fulfilling the criteria of segmentation or

branching cracks) (iv) porosity (spherical porosity in the cross section of the coating). The

segmentation crack density was calculated by dividing the number of segmentation cracks present

in each cross section with the cross section length.

Coating A, sprayed with the highest average surface temperature, had the highest levels of

segmentation cracks (1.1 mm-1

), and significant branching cracks originating from them. Coating D,

with the lowest average back surface temperature, produced by cooling the substrate and waiting in-

between gun cycles, had a segmentation crack density of 0.7 mm-1

. Also, it’s worth noting that in

this coating no branching cracks were present. Coatings B and C had a similar average backsurface

temperature. However, the difference between the minimum and maximum temperatures in coating

B was about 100o C more than in coating C. In coating B, the crack segmentation density is

relatively low (0.72 mm-1

) and branching cracks are present. Coating C has a segmentation crack

density of 0.91 mm-1

and very few branching cracks are observed


141

Figure 7.4. Polished cross section of as-sprayed YSZ

top coat sprayed without cooling and without waiting in-

between each gun cycle. [average backsurface

substrate temperature : 507oC].


top coat sprayed with 35 sec wait in-between each gun

cycle. [average backsurface substrate temperature :

445oC].

Figure 7.6 Polished cross section of as-sprayed YSZ

top coat sprayed with air cooling at the back substrate

surface. [average backsurface substrate temperature :

429oC].


top coat sprayed with air cooling at the substrate back

surface and with 35 sec. wait in-between each gun

cycle. [average backsurface substrate temperature :

212oC].

These observations suggest that higher average substrate temperatures and lower differences

between the maximum and minimum substrate temperature give rise to coatings with higher density

of segmentation cracks. Segmentation cracks are generated when the sample is cooled to room

temperature, after the deposition process is complete, as a result of relaxation of bi-axial stresses

[347]. The segmentation crack initiates at the free surface and propagates through the thickness of

the coating. When it reaches the weak interfaces that exist in-between layers due to poor splat

bonding, the stresses at the crack tips are relieved by opening up these cracks. At high average


142

substrate temperature and low differences between the maximum and minimum substrate

temperature, there will be improved contact between splats [348-350], and stresses in the top coat

cannot relax by generation of cracks in-between splats or microcracking. Thus relaxation occurs by

the generation of segmentation cracks. When the difference between the minimum and maximum

spraying temperature is high, even for a high average substrate temperature, bonding between splats

is poor and the segmentation crack density is relatively low. Similar results have been observed by

other researchers [68, 346]. Coatings with higher densities of segmentation cracks density perform

better under thermal cycling [68, 346].

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

0 1 2 3 4 5

Ave

rag

e B

acksu

rfa

ce

Te

mp

era

ture

(oC

)

continuous

spraying

A

waiting in-

between gun

cycles

B

continuous

spraying with

substrate

cooling

C

waiting in-

between cycles

with substrate

cooling

D

Figure 7.8 Average substrate back-surface temperature during spraying for different

cooling conditions. The error bars indicate the temperature range of the substrate during

spraying.

The numerical model described in section 3.1.3 and developed by Clyne and co-workers [95, 99,

351-353] was used to calculate the residual stresses in the top coat and confirm the segmentation

and branching crack initiation process. The measured thermal history was compared with that

obtained using the numerical model. Comparisons of experimental and modelled backsurface

thermal histories for each of the different spraying conditions are shown in Figure 7.9 to Figure 7.12.

The agreement for all cases is good. The residual stress distribution through the coating was then

obtained from the numerical model. Figure 7.14 shows the residual stress distribution through the

thickness of the coating produced by waiting in-between each gun cycle. There is a stress gradient

through the thickness of the coating, with compressive stresses at the TC/BC interface and tensile

stresses in the outer surface. The results from the model confirm that near the free surface of the


143

top coat there are high tensile stresses that probably lead to initiation of the segmentation cracks.

Branching cracks initiate from the segmentation cracks at places of weak splat bonding and where

stresses are tensile (see Figure 7.5).

0

100

200

300

400

500

600

700

800

0 50 100 150 200 250

Time (sec)

Tem

pera

ture

(oC

)

experiment

model

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Time (sec)T

em

pera

ture

(oC

)

experimentmodel

Figure 7.9 Experimental and modelled back surface

thermal histories for YPSZ-TC sprayed without cooling

and without waiting in-between each gun cycle.

[average backsurface substrate temperature : 507oC]


thermal histories for YPSZ-TC sprayed with 35 sec wait

in-between each gun cycle. [average backsurface

substrate temperature : 445oC]

0

100

200

300

400

500

600

700

0 50 100 150 200

Time (sec)

Tem

pera

ture

(oC

)

experiment

model

0

100

200

300

400

500

600

700

800

0 100 200 300 400

Time (sec)

Tem

pera

ture

(oC

)

experiment

model


thermal histories for YPSZ-TC sprayed with air cooling

at the back substrate surface. [average backsurface



thermal histories for YPSZ-TC sprayed with air cooling

at the substrate back surface and with 35 sec. wait in-

between each gun cycle. [average backsurface



144

Segmentation and branching cracks are characteristic of thick TBCs and are not present in coatings

less than about 1 mm thick. The average stress state in the top coat after spraying coatings of

different thickness was estimated using the numerical model by matching the measured and

modelled thermal histories. The results show that the average residual stress in the top coat changes

from compressive to tensile with increasing coating thickness (see Figure 7.13). Stresses in the

outer surface of thicker coatings are higher (more tensile) than thinner coatings. Even when the

average stress in the top coat is compressive, the stresses near the free surface of thick coatings are

tensile. For thinner coatings the stresses near the free surface are possibly not high enough for

initiation of the segmentation cracks.

-16

-14

-12

-10

-8

-6

-4

-2

0

2

4

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Coating thickness (mm)

Ave

rag

e S

tre

ss in

Co

atin

g (

MP

a))

Figure 7.13 Averages residual stress in APS-YPSZ top

coats of different thickness.

Figure 7.14 Residual stress through thickness of as-

sprayed YSZ top coat sprayed by waiting in-between

each gun cycle.

The Young’s modulus of coating A, B, C, and D was determined by cantilever bending, with the

free surface of the coating in tension. The results are presented in Figure 7.15 as a function of

segmentation crack density. The Young’s moduli are in the range of 4.5-10 GPa. Segmentation

cracks are expected to have a strong influence on the Young’s Modulus measured by this technique.

Coating A, with the highest segmentation crack density, has the lowest Young’s modulus. Due to

the fact that segmentation cracks do not always penetrate entirely the thickness of the coating, lower

values for the stiffness of the coating are expected when the free surface is loaded under tension.

-60

-40

-20

0

20

40

60

80

1 1.2 1.4 1.6 1.8 2 2.2

Distance (mm)

Str

ess (

MP

a)

maximum stress

(segmentation

crack initiation)

TC/BC interface

initiation of

branching cracks


145

A

CD

B

0

2

4

6

8

10

12

14

0.6 0.7 0.8 0.9 1 1.1 1.2

Segmentation crack density (mm-1

)

Yo

un

g's

m

od

ulu

s (

GP

a)

-8

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

100 200 300 400 500 600

Average Substrate Temperature (oC)

Avera

ge S

tress in

Coatin

g (

MP

a)

Figure 7.15. Influence of segmentation crack density on

Young’s modulus of thick PS TBCs. Error bars indicate the

standard deviation and the letters A, B, C, D refer to the

corresponding coatings sprayed with conditions specified in

the text.

Figure 7.16 Average residual stress in APS YPSZ top

coat against average temperature at the substrate back

surface during spraying.

7.2.2 The Effect of Substrate Temperature on Residual Stresses

Figure 7.16 shows that the average TC residual stress changes from tensile to compressive with

increasing average substrate temperature. There is a stress gradient through the thickness of the TC.

It can be seen in Figure 7.14 that stresses at the outer surface are tensile, with higher stresses

(reaching 68 MPa) for the lower substrate temperature. At the TC/BC interface, stresses are

compressive. Coatings deposited at higher temperatures showed the highest compressive stresses at

the interface. With increasing substrate temperature, stresses become more compressive. The

reason for this is that, as the substrate temperature is increased, the stresses due to differential

thermal contraction generated on cooling increase, producing greater compressive residual stresses

in the top coat. Similar results on the effect of substrate temperature on the residual stress state of

the top coat have been reported [77, 348, 354, 355].

7.3 Stresses in TBCs after heat treatment

The model described in section 3.1.3 and developed by Clyne and co-workers [95, 99, 351-353] has

been used in the present work for prediction of residual stresses after spraying and subsequent heat

treatments. The predicted effect of imposing a high through-thickness thermal gradient is shown

Figure 7.17b. It can be seen that the stress level in the coating is moved towards more compressive


146

values when the thermal gradient is imposed, particularly near the free surface, where also the

temperature is the highest. This is expected to accelerate the rates of sintering and stiffening.

Figure 7.17a shows the predicted effect on the residual stresses of heat treatment isothermally at

1100o C. In this case, the stress in the top coat is approximately uniform and tensile and it is

expected that sintering processes will be retarded by the presence of tensile stresses – which hold

open the microcracks and pores [60]. The difference in the stress state when heat treated under a

thermal gradient and heat treated isothermally is quite significant. This must be considered if

realistic conclusions about the thermomechanical behaviour of TBCs are to be drawn.

-10

-5

0

5

10

15

20

25

30

0

200

400

600

800

1000

1200

1400

1600

0 10 20 30 40 50

Distance from base (mm)

stress

temperature

Str

ess

(MP

a)

Tem

pera

ture (

o C)

Substrate Coating

(a)

-200

-100

0

100

200

300

0

200

400

600

800

1000

1200

1400

1600

0 10 20 30 40 50

stress

temperature

Str

ess

(M

Pa)

Tem

per

atu

re (

oC)

Distance from base (mm)

Substrate Coating

(b)

Figure 7.17. Predicted through-thickness distributions of stress and temperature in a

TBC system (a) when isothermal at 1100˚C and (b) when subjected to a high thermal

gradient, similar to that employed in the experimental work.


147

7.4 Conclusions


1. The Young’s modulus of as-sprayed zirconia-based top coats, measured by cantilever bending,

is in the range of 10-15 GPa. After heat treatment the elastic modulus increases substantially

due to sintering. Due to the errors involved, the differences in the sintering and stiffening

behaviour between top coats of different compositions cannot be precisely evaluated with this

technique. The observed trend of increasing stiffness after heat treatment is nevertheless quite

clear. The Young’s moduli of YSZ top coats measured by indentation approach the value of the

monolithic material and are not representative of the global stiffness of the coating due to the

small volume of material undergoing deformation.

2. A through thickness gradient in the stiffness of the top coat is obtained when the coating is heat

treated in a thermal gradient. This is because sintering progresses more rapidly near the free

surface where the temperature is highest and the stresses are more compressive.

3. Thick thermal barrier coatings exhibit microstructural features, such as segmentation and

branching cracks, which affect the elastic properties and which are not present in thinner TBCs.

These cracks are generated as a result of relaxation of bi-axial tensile residual stresses near the

free surface of thick TBCs.

4. Crack segmentation density in thick thermal barrier coatings is strongly dependant on the

substrate temperature during deposition. Higher average substrate deposition temperatures,

combined with low variations between the minimum and maximum substrate temperature, give

higher crack segmentation densities. Coatings with higher crack segmentation densities have a

lower Young’s modulus.

5. The average top coat residual stress changes from tensile to compressive with increasing

average substrate temperature.

6. The high thermal gradients commonly present in TBCs under service conditions lead to widely

varying conditions within the top coat. Sintering is likely to occur much more rapidly near the

free surface, not only because the temperatures are higher there, but because the residual

stresses tend to be more compressive. It is known that tensile stresses retard the progression of

sintering.

Chapter 8: Thermal Conduction in PS TBCs

148

8 Thermal Conduction in Plasma-Sprayed TBCs

8.1 Thermal Conductivity of Plasma sprayed TBCs

8.1.1 Steady state Rig

Figure 8.1 and Figure 8.2 show experimental data, and corresponding best linear fits from eqn 5.19,

for samples tested using both silicone-based thermal compound and conductive pads. By

employing a thermal compound, ktrue and h were found to be 1.38 ± 0.27 W m–1

K–1

and

16.7 kW m–2

K–1

, respectively (Figure 8.1). Using conductive pads, on the other hand, measured

ktrue and h values were relatively lower, i.e. 1.1 ± 0.10 W m–1

K–1

and 3.77 kW m–2

K–1

, respectively

(Figure 8.2). The silicone-based compound appears to be more effective than conductive pads in

raising the interfacial thermal conductance (h), by filling more gaps between the matching surfaces.

However, the use of such compound also raises the true conductivity, by as much as ~30%. With

porous materials, there is a serious danger of liquids or soft solids entering the pores during

specimen preparation or during measurement, which will tend to raise the measured conductivity.

Therefore, the use of thermal compounds should be avoided for coatings containing relatively large

surface-connected pores, such as those typically found in sprayed TBCs.

It is worth pointing out that some thermophysical properties of PS coatings are often sensitive to

powder composition and spraying parameters. Nevertheless, the measured value of thermal

conductivity of detached YSZ coatings (1.1 ± 0.10 W m–1

K–1

) is in relatively good agreement with

literature data for PS coatings of similar composition and porosity (see Figure 4.6).

It can be seen that both cases exhibited larger scatter than the fused silica samples that were used to

validate the technique - see figure 5.18. The detached YPSZ coatings were found to be slightly

curved, as a result of residual stresses, making preparation of samples with flat surfaces difficult.

Also, the inherent variability in the microstructure of such material is much greater than that of

fused silica. Thin conductive pads are recommended for measuring the conductivity of YPSZ. Its

application does not present the danger of soft solids (e.g. silicone compound) infiltrating the larger

pores and altering the conductivity. Also, conductive pads are sufficiently compliant to

accommodate the slightly curved surface of detached coatings, hence raising the interfacial thermal

conductance.


149

0.0

0.5

1.0

1.5

2.0

0.0 0.5 1.0 1.5 2.0 2.5

YPSZ coating

(with silicone thermal compound)

T

/Q (

K m

-2 k

W-1

)

x (mm)

ktrue

= 1.38±0.27 W m-1

K-1

h = 16.7 kW m-2

K-1

1/ktrue

= 0.7235

2/h = 0.12

Figure 8.1. Plot of TQ against specimen thickness for detached as-sprayed YSZ top coats, obtained using

the conductive compound. Deduced values of ktrue and h are shown.

0.0

0.5

1.0

1.5

2.0

2.5

0.0 0.5 1.0 1.5 2.0 2.5

YPSZ coating

(with conductive pads)

T

/Q (

K m

-2 k

W-1

)

x (mm)

ktrue

= 1.1±0.1 W m-1

K-1

h = 3.77 kW m-2

K-1

1/ktrue

= 0.9083

2/h = 0.53

Figure 8.2. Plot of T/Q against specimen thickness for detached as-sprayed YSZ top coats, obtained using

the conductive pads. Deduced values of ktrue and h are shown.


150

8.1.2 Laser Flash Measurements

The thermal conductivity of detached as-sprayed YSZ (204NS) and DSZ top coats was measured

using laser flash technique. Details on the experimental procedure are given in section 5.6.3. The

theoretical background of this technique is given in 4.2.2.3. Measured thermal diffusivity data for

detached YSZ (204NS) top coats are presented in Figure 8.3 together with measured specific heat

data and deduced thermal conductivity. The density of the YSZ detached top coat was measured by

the Archimedes method (see section 5.3.5) to be about 5.35 g cm-3

which corresponds to about 11%

porosity, (typical of as-sprayed coatings -see section 6.4.5).

Figure 8.3. Thermal conductivity, thermal diffusivity and specific heat measurements from 25-1400

o C of

detached YSZ as-sprayed top coat (204NS) The thermal diffusivity was measured by the laser flash

technique.

The measured thermal conductivity decreases slightly with increasing temperature, up to about

1100oC, indicating a phonon conduction mechanism (see section 4.1.1) and then increases slowly

above about 1100oC. The increase at higher temperature is probably due to radiative transport (see

section 4.1.2). At 100oC the thermal conductivity is 1.13 W m

-1 K

-1 which is close to the value

measured using the steady state bi-substrate technique (1.1 W m-1

K-1

). The data for DSZ follow a

similar trend (Figure 8.4). The porosity of the DSZ as-sprayed sample was also about 11 %

(density ~ 5.8 g cm-3

). The measured thermal conductivity at 100oC is about 1.38 W m

-1 K

-1. The


151

error expected in the values measured here is about 10%. The thermal conductivity of as sprayed

DSZ is slightly greater than YSZ (204NS). This will be discussed in more detail in section 8.2.

Figure 8.4. Laser flash data for the thermal diffusivity of detached DSZ as-sprayed top coats (SPM6-2444)

together with measured specific heat data and deduced thermal conductivity.

Laser flash has been extensively used for the measurement of the thermal diffusivity of the thermal

barrier coatings because of its rapid measurement and high temperature capabilities- see sections

4.2.2.3 and 4.3. It is important to consider several factors that could be significant in the

determination of the thermal diffusivity of TBCs. One of these factors is the gas content and gas

pressure of the pores in the TBC during the measurement, since this will affect the measured value

of the thermal diffusivity. In addition, TBCs are translucent in the infrared region and hence the

energy pulse is not absorbed at the surface of the coating. To overcome this problem the surface of

translucent specimens is routinely treated for laser flash measurements by applying an opaque

coating of graphite, Au or Pt. However, high temperatures can cause partial evaporation of these

layers and possible reaction with the sample is another cause of concern [356]. Also, since the laser

flash method measures the thermal diffusivity of the sample, calculation of conductivity requires

prior knowledge of the volume specific heat at different temperatures. Since TBCs show large

variations in porosity and crystallite size [14, 357], it can be difficult to measure the volume specific

heat accurately. Another problem lies in determining suitable laser pulse power and test duration


152

values [358]. In general, these are inherent drawbacks with all transient methods. The reliability of

laser flash thermal diffusivity measurements for thermal barrier coatings was addressed by Wang at

el [358]. Their work suggests that due to the inherent low thermal conductivity and low sample

thickness of TBC, large temperatures can develop at the front and rear surface of the sample. This

could introduce errors in the temperature measurement at the rear surface due to non-linear

temperature-voltage response of IR detectors, and yield lower values for the thermal diffusivity of

the TBCs.

8.1.3 Hot Disk

The thermal conductivity of as-sprayed detached YSZ (204NS) top coats was measured using the

hot disk method. For the theoretical background of the technique see section 4.2.2.2. and for the

experimental procedure see section 5.5.2. The thermal conductivity was estimated to be

1.2 ± 0.1 W m-1

K-1

and the thermal diffusivity 0.48 mm s-1

, which is in reasonable agreement with

the values measured by the bi-substrate steady state technique and laser flash. Figure 8.5 is a

comparative plot of the thermal conductivity values of as-sprayed YSZ top coats measured with the

techniques mentioned above. The error bars indicate the errors associated with the technique (for

laser flash and hot disk techniques) or the experimental scattering (for steady state technique).

Most of the techniques currently employed for coatings are either steady-state [258, 263, 359] or

laser flash [358, 360] (transient) methods- see section 4.3. However, measured values obtained

using different techniques often show significant variations. This is largely because the

assumptions employed are not strictly valid [361]. Laser flash is often considered to be the most

convenient and accurate method over a wide range of temperatures. It does not require large

samples and can be used for remote measurement, both under vacuum and in gas atmospheres.

However, there are several issues surrounding the application of this method to thermal barrier

coatings (TBCs) [358, 360]. The Hot Disk method has not been extensively used for the

measurement of thermal conductivity of TBC coatings. This technique is very sensitive to sample

dimensions and two identical samples of the same dimensions are needed for a single measurement.

In addition, there are problems in determining suitable power and test duration values as well as a

“time window” for analysis -see section 5.5.2 and 4.2.2.2. On the other hand, in the steady-state

method presented here, only the temperature drop across the sample and the heat flux are required

to calculate the thermal conductivity. For relatively low temperatures (from room temperature to

about 500ºC), the steady state method presented is a simple set up which can reliably be used to


153

measure the thermal conductivity of TBCs and ceramic coatings of different thicknesses and

porosity levels.

Figure 8.5. Thermal conductivity of YSZ (204NS) top coats

measured by different techniques.

8.2 Effect of Powder Composition on Thermal Conduction

The thermal conductivity of DSZ top coats has been found to be higher than that of YSZ, while that

of YLaSZ is lower than YSZ (see Figure 8.6 and Figure 8.7). The thermal conductivity of

stabilized zirconia is believed to be dominated by oxygen vacancy scattering [262, 272]-see section

4.3. Scattering by point defects, such as vacancies and foreign ions, increases with defect

concentration and difference in the mass between the host ion and the defect (see equation 4.3).

Larger misfits between the ionic radii of the host and dopant ions are also expected to introduce

lattice strain which will promote phonon scattering [18].

In a DSZ top coat the dopant ion (Dy3+

) has a high atomic weight (162.5) and the ionic radius is

0.103 nm, which is larger than the ionic radius of the host ion Zr4+

(0.084 nm) and about the same

as Y3+

(0.102 nm) (see table Table 8.1). However, the DSZ top coat contains 3.58 mol% of dopant,

whereas YSZ contains 4.3 mol% of dopant, and YLaSZ has a total of 6.25 mol% (2.6 mol% La2O3

and 3.65 mol% Y2O3). The thermal conductivity of YSZ, is slightly lower than the thermal

conductivity of DSZ which has a dopant ion with a large atomic weight and slightly less stabilizer.

In the YLaSZ while the dopants have lower atomic mass than Dy3+

(atomic weight of La3+

is 138.9

and Y3+

88.9) the total amount of stabilizer is greater and hence the level of oxygen vacancies is

higher, which promotes scattering of phonons. Additionally, the ionic radii misfit between the Zr4+

host ion (0.084 nm), the La3+

(0.116 nm) and Y3+

(0.102 nm) ions create lattice strains, again

promoting phonon scattering. This effect is expected to be more pronounced for the La3+

ion, since


154

it has the largest misfit. The lower thermal conductivity of YLaSZ and YSZ top coats confirms that

oxygen vacancies play a dominant role in phonon scattering and hence in reducing the thermal

conductivity. Similar effects of La2O3 additions have been found by other researchers [274].

It must be pointed out that, even though the coatings have similar porosity (about 10%), there might

be differences in the microstructure, such as many interfaces with poor bonding between splats,

with little effect on the total porosity level, but a strong effect on thermal conductivity.

Material Ion/

Stabilzing

Ion

Ionic

radius

(nm)

Ionic

mass

Valence

difference

from Zr4+

Atomic

concentration

in powder

(mol%)

Measured k

at 100oC

(W m-1

K-1

)

ZrO2 Zr4+

0.084 91.2 0 -

YSZ Y3+

0.102 88.9 -1 4.3 1.1

DSZ Dy3+

0.103 162.5 -1 3.58 1.29

YLaSZ La3+

/

Y3+

0.116 /

0.102

138.9 /

88.9

-1 /

-1

2.6 /

3.65

0.68

Table 8.1 Ionic radius, ionic mass, valance and atomic concentration of stabilizing ion in YSZ, DSZ

and YLaSZ top coats together with measured values of thermal conductivity at 100oC by the bi-

substrate steady state technique.

(a)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 2 4 6 8 10 12

YSZ

DSZ

YLaSZ

Th

erm

al co

ndu

ctivity (

W m

-1 K

-1)

Time (hours)

1400oC

(b)

0.0

0.4

0.8

1.2

1.6

2.0

2.4

0 2 4 6 8 10 12

YSZ

DSZ

Th

erm

al co

ndu

ctivity (

W m

-1 K

-1)

Time (hours)

1200oC

Figure 8.6. Thermal conductivity of detached YSZ, DSZ and YLaSZ top coats at 100oC as a function of heat treatment time

(a) at 1400oC. and (b) at 1200

oC. Measurements where performed using the steady state technique.


155

(a)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 2 4 6 8 10 12

YSZ

DSZ

Th

erm

al co

ndu

ctivity (

W m

-1 K

-1)

Time (hours)

1400oC

(b)

0.0

0.4

0.8

1.2

1.6

2.0

2.4

0 2 4 6 8 10 12

YSZ

DSZ

Th

erm

al co

ndu

ctivity (

W m

-1 K

-1)

Time (hours)

1200oC

Figure 8.7. Thermal conductivity of detached YSZ and DSZ top coats at 100

oC as a function of heat treatment time. (a) at

1400oC and (b) at 1200

oC. Measurements were performed using the laser flash technique.

8.3 The Effect of Heat Treatment on the Thermal conductivity of TBCs

The thermal conductivity of detached plasma sprayed top coats after isothermal heat treatment at

1200oC and 1400

oC was measured using the steady state and the laser-flash techniques. Figure 8.6

shows the evolution of the thermal conductivity for YSZ, DSZ and YLaSZ top coats after heat

treatment at 1200oC and 1400

oC measured by the steady state technique. For each time and

temperature, at least two samples of different thickness were used for the measurement. Conductive

pads between the sample and substrate were used in order to minimize the contact resistance. For

calculating the thermal conductivity, the interface conductance h was assumed to be 3.77 kW m-2

K-

1, as previously calculated for as-sprayed YSZ top coats of different thicknesses (see section 8.1.1).

The surface finish in all samples was the same and the same pressure was applied, so the

assumption that the interfacial conductance was similar should be valid.

The thermal conductivity increases after heat treatment, which is more pronounced for the sample

heat treated at 1400oC. Very little increase has occurred after heat treatment at 1200

oC. This

increase is due to sintering of the top coat which causes the disappearance of small pores and

interphases between splats as well as grain growth. Small pores, splat interphases and grain

boundaries inhibit heat transfer and cause scattering of phonon and photons. Therefore the increase

of the thermal conductivity observed is due to sintering. At 1200oC not much sintering takes place


156

and hence a smaller increase in the thermal conductivity is measured. Similar trends have been

observed by other researchers [112, 254, 257, 259, 262, 264, 265](see section 4.3). Increase in the

thermal conductivity is due to sintering of cracks, pores and interfaces during heat treatment. In PS

coatings, poor contact between lamellae in close physical proximity bond together assisted by

diffusion processes at high temperatures. It should be pointed out that even though no significant

increase in the thermal conductivity is observed after 10 h at 1200

oC, there are significant increases

in the Young’s modulus at this temperature and heat treatment times [362].

The thermal conductivity of the detached YSZ (204NS) and DSZ (SPM6-2444) coatings after

isothermal heat treatment was also measured by the laser flash technique. Results are presented in

Figure 8.7. Both techniques confirm similar trends: there is an increase in the thermal conductivity

after heat treatment for both coatings and the increase is more pronounced for the coating that has

been heat-treated at 1400oC. The thermal conductivity of the heat treated coatings follows the same

trend as the as-sprayed coatings with temperature; firstly decreases and then at above 1100oC it

slightly increases due to the effect of radiation. Both the bi-substrate and laser flash techniques give

comparable results and trends.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0 2 4 6 8 10 12

Heat Treatment Time (hours)

No

rmilis

ed

Th

erm

al C

on

du

ctivity k

/ko

YSZ

DSZ

YLaSZ

Figure 8.8. Relative change in the thermal conductivity of detached top coats after heat treatment at 1400

oC

measured by the steady state technique. k/ko is defined as the thermal conductivity of coating after heat

treatment (k) over the thermal conductivity of the as-sprayed coating (ko).

The relative increase in the thermal conductivity after heat treatment at 1400oC is presented in

Figure 8.8. It is apparent than even though the initial thermal conductivity of YLaSZ is

considerably lower than those of YSZ and DSZ top coats, it quickly increases with heat treatment

due to sintering. This implies that the microstructural features, such as interfaces and poor bonding


157

between splats that cause a reduction in thermal conductivity, are not stable and hence some of the

benefits of such features quickly disappear after a few hours of exposure to high temperature.

Nevertheless, the absolute value of the thermal conductivity of YLaSZ after 10 h at 1400oC is still

lower than that of YSZ and DSZ top coats. The relative increase of DSZ and YSZ top coats is very

similar, however, the initial thermal conductivity of DSZ is higher than YSZ.

8.4 Effect of Pore Conductivity on Overall Thermal Conductivity

The effect of pore conduction on the overall thermal conductivity of PS YSZ was explored by

experimental measurements in air and in vacuum. Measurements were performed at atmospheric

pressure in air with the steady state technique and compared to corresponding measurements in

vacuum. The thermal conductivity of YSZ plasma sprayed TBCs in vacuum was found to be ~15%

lower than that in air (Figure 8.9a). Similar results have been found by other researchers [363].

The analytical TFR model described in section 4.4.4 was used to interpret the results (Figure 8.9b).

For comparative purposes, some measured data available in literature on the thermal conductivity of

7-8 wt% YSZ with about 10% porosity are also plotted on the same graph. Open symbols indicate

measurements performed in vacuum and filled symbols measurements performed in air or another

gas at atmospheric pressure. For the model, the values used for the splat thickness (=Lv-dv) was

1.9 µm, the inter-splat pore thickness, (dv), 0.1 µm and inter-bridge spacing, Lv, 6 µm (see section

4.4.3 for more detailed explanation for the values used). These values correspond well to the

microstructural feature observed in our PS top coat (see Figure 8.10). For this input data there is

good agreement between the measured and predicted effective thermal conductivity. It can be

concluded that thermal conduction through the pores affects the overall thermal conductivity of

TBCs and thus cannot be neglected.

In noting the effect of gas pressure, it may be recalled that in the hot end of a typical gas turbine

engine, the pressure varies between about 40 atmospheres and one atmosphere. Furthermore the gas

permeability of TBC top coats is known to be very high [21]. Therefore, such pressures will

quickly become established throughout most of the pores within such coatings. At atmospheric and

reduced pressure, convection in pores can be neglected, since the mean free path of air molecules is

comparable with the pore dimensions. The pore conductivity is low and temperature independent.

Higher gas pressures can significantly raise the pore conductivity, particularly at high temperature

(see Figure 8.9b). However experimental measurements of the thermal conductivity are normally

carried out either at one atmosphere or under reduced pressure, in which case the pore conductivity


158

will be appreciably lower. Furthermore, measurements are often made at room temperature, which

will also have the effect of reducing the pore conductivity.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

air vacuum

Therm

al C

onductivity (

W m

-1 K

-1)

0.5

1

1.5

2

2.5

500 1000 1500 2000

model (current work) in air 40atmmodel (current work) in air 1atmmodel (current work) in vacuummeasurement (from literature) in airmeasurement (from literature) in vacuummeasurement (current work) in airmeasurement (current work) in vacuum

Therm

al C

on

ductivity (

W m

-1 K

-1)

Temperature (K) Figure 8.9. (a) Thermal conductivity of YSZ top coat measured by the steady state technique in air and in vacuum

(b) Comparison of measured values of the thermal conductivity of 7-8wt%YSZ with prediction from the model

described in section 4.4. The literature values were taken from references [67, 112, 259, 264, 364, 365].

8.5 Experimental and Predicted Thermal Conductivity

Microstructural features from YSZ top coats and measured thermal conductivity by the steady state

method were compared with the predictions of the microstructure-based model described in section

4.4.4. There are difficulties in making reliable measurements of microstructural features such as

inter-splat bridge areas and spacings. These are often more clearly seen on fracture surfaces, such

as those shown in Figure 8.10, rather than on polished sections, but it is possible that the

microstructure could be affected by the fracturing process. The input parameters of the model were

inferred from SEM micrographs (Figure 8.10) of the YSZ top coats and densitometry.

For the particular input data employed, the TFR model gives predictions for the thermal

conductivity which are consistent with these measurements. While this procedure is clearly an

approximate one, it shows the potential of using mictrostructural parameters which can be identified

from various types of experimental measurement including inspection of micrographs, densitometry

and small angle neutron scattering (SANS), to predict the thermal conductivity of PS top coats.

Furthermore, the model can give an idea of how the sintering mechanisms generate microstructural

changes, which affect the coating conductivity.


159

Figure 8.10. SEM micrographs of fracture surfaces of as-sprayed YSZ top coats showing typical microstructural

features.

In addition, the model predictions were also compared with the experimental data from literature as

well as other microstructure-based models for the thermal conductivity [288, 291] - see Figure 8.9b).

Unfortunately, while there are many reported thermal conductivity measurements for plasma

sprayed TBCs, there are certain problems in making comparisons with predictions from

microstructure-based models. In particular, the properties of plasma sprayed TBCs are often very

sensitive to powder composition and spraying conditions [137, 302], so it is important that both

thermal properties and microstructural features should be measured on the same specimens.

However, there have been a few studies in which systematic measurements have been made of both

conductivity and microstructural parameters. Data from one such study [119] are presented in

Figure 8.11. It can be seen that substitution of appropriate input data into the TFR model gives

predictions for the thermal conductivity which are consistent with these measurements. For these

particular input data, the predictions of the McPherson [288] model look rather unreliable, while

those of Lu et al [291] are closer to experiment and to the TFR model.


160

0

0.4

0.8

1.2

1.6

2

2.4

0 0.2 0.4 0.6 0.8 1

TFR model (current work)measured (Kulkarni et al)measured (current work)modified shear lag model (Lu et al)contact resistance model (McPherson)

Effe

ctive T

he

rma

l C

on

du

ctivity,

keff (

W m

-1 K

-1)


/ Stot

Splat thickness Lv = 1.9 m

Bridge thickness dv = 0.1 m

Bridge spacing, Lh

= 6 m

0

0.5

1

1.5

2

0 0.2 0.4 0.6 0.8 1

TFR model (current work)

measured (current work)

modified shear lag model (Lu et al)

contact resistance model (McPhearson)

Effe

ctive t

he

rma

l co

nd

uctivity, k

eff (

W m

-1 K

-1)


/ Stot

Splat thickness Lv = 1.9 m

Bridge thickness dv = 0.1 m

Bridge spacing, Lh = 6 m

Figure 8.11. Comparison between predicted conductivities obtained from TFR (see section 4.4.4.), contact resistance [288]

and the modified shear lag [291] models and experimental measurements [119] (a) in vacuum and (b) in air (1 atm).

8.6 Conclusions


1. A novel thermal conductivity measurement technique is described. The setup is simple and can

be applied to ceramic coatings of different thickness and porosity levels. The setup is designed

to generate steady one-dimensional heat flow through the coating. The heat flux is calculated as

the average through upper and lower substrates, taking into account their thermal conductivities,

including any temperature dependence. The method involves testing samples of different

thickness, so that a plot can be constructed of temperature drop over heat flux against specimen

thickness. This allows the true thermal conductivity of the sample, ktrue, and the interfacial

thermal conductance, h, to be evaluated. The technique has been validated using fused silica

samples of known thermal conductivity-see section 5.6.1.3, employing conductive paste at the

interface. The measured values were consistent with this figure, within the expected

experimental error.

2. The thermal conductivity of as-sprayed YSZ TBCs was measured using the bi-substrate

technique and was found to be 1.1 ± 0.10 W m-1

K-1

, using conductive pads and specimens

about 0.5-2 mm in thickness. However, a higher value of 1.38 ± 0.27 W m-1

K-1

was obtained

when the conductive paste was employed. This is attributed to entry of the paste into the

extensive surface-connected porosity which is known to be present in these TBCs. The use of


161

such conductive pastes is not recommended when relatively large surface-connected pores are

known to be present. Laser flash and Hot disk techniques were also used to measure the thermal

conductivity of as-sprayed TBCs, with reasonably good agreement, within experimental error.

The drawbacks of each of the techniques were briefly discussed.

3. The thermal conductivity of YSZ and DSZ as-sprayed and heat treated top coats was measured

using laser flash technique up to 1400oC. The thermal conductivity decreases slightly with

increasing temperature, up to about 1100oC, indicating a phonon conduction mechanism. It then

increases slowly above about 1100oC. The increase at higher temperature is probably due to

radiative transport.

4. The thermal conductivity of DSZ top coats has been found to be higher than that of YSZ, while

that of YLaSZ is lower than YSZ. The results suggest that oxygen vacancies play a dominant

role in phonon scattering and hence in reducing the thermal conductivity, whereas phonon

scattering due to the mass difference between dopant and host ion has a weaker effect.

5. Heat treatment results in an increase of the thermal conductivity of YSZ, DSZ and YLaSZ

coatings due to sintering. The increase is more pronounced at 1400oC, whereas after 10 h

at1200oC hardly any increase is observed. The relative increase in the thermal conductivity is

higher for the YLaSZ top coats. This suggests that some microstructural features in YLaSZ

responsible for a reduction in thermal conductivity are not stable, and hence some of the

benefits of such features quickly disappear after a few hours at high temperature. Nevertheless,

the absolute value of the thermal conductivity of YLaSZ top coats after 10 h at 1400oC is still

lower than YSZ and DSZ top coats.

6. The effect of thermal conduction through the gas in the pores on the overall thermal

conductivity of TBCs was explored. The thermal conductivity of YSZ plasma sprayed TBCs in

vacuum was found to be ~15% lower than in air. Predictions from the TFR model show that

higher gas pressures in the pores, up to the range typical of gas turbine service conditions, can

significantly raise the pore conductivity, particularly at relatively high temperatures.

Conversely, the pore conductivity is predicted to be low under reduced pressure, which is

sometimes used for experimental measurements. Depending on temperature and microstructure,

there may be a significant effect on the overall conductivity.

7. Some limited comparisons have been made between experimentally measured conductivities

and predictions from the TFR model. These suggest that it should be possible to predict the

conductivity from measurable microstructural parameters.

Chapter 9: Conclusions and Future Work

162

9 Conclusions and Future Work

9.1 General Conclusions

In the present work plasma sprayed zirconia-based top coats were characterized with respect to their

sintering behaviour, microstructural features, phase constitution, thermal conductivity and residual

stress state.

Exposure to high temperatures induced phase changes in all the zirconia-based top coats studied,

and this may affect the microstructure and perhaps the stress state in TBCs. The presence of SiO2

and Al2O3 impurities appears to decrease the rate of decomposition. Dysprosia stabilizer, in the

amount present in the current study (3.58 mol%), successfully stabilized the tetragonal phase in as-

sprayed zirconia coatings. The oxidation state of cerium in ceria-stabilized-zirconia has a great

influence on the phase composition, and it is believed to affect the stress state in the coating.

The addition of La2O3 significantly lowered the thermal conductivity. However La2O3 coatings

decomposed after 10 h and the pyrochlore phase was formed. The formation of the pyrochlore

phase accompanies volume increase, which is thought to enhance the tetragonal-to-monoclinic

phase transformation above 1 mol% La2O3. Hence the amount of La2O3 added must be tailored in

order to avoid degradation of the thermal barrier coating.

Due to sintering processes, detached top coats exhibit substantial reductions in linear dimensions

during exposure to at elevated temperatures. These linear contractions are greater in the through-

thickness direction than in the in-plane directions. Any contributions to the observed shrinkage

arising from phase changes are expected to be relatively small. Top coat compositon affects the

shrinkage rate of top coats, due to changes in the diffusion kinetics. The presence of SiO2 and

Al2O3 impurities enhances the sintering kinetics by providing a continuous low diffusivity pathway

at the grain boundary which will restrain grain boundary mobility, reduce grain growth and enhance

shrinkage. It may also do so by providing a high diffusivity pathway, e.g. liquid phase at the grain

boundaries, which will enhance shrinkage. Thermal cycling, commonly experienced by TBCs

under service conditions, affects their shrinkage. The shrinkage observed during thermal cycling is

less than that observed when heat treatment is continuous.

The high thermal gradients, commonly present in TBCs in-service, lead to widely varying

conditions within the top coat. Sintering is likely to occur much more rapidly near the free surface,

not only because the temperatures are higher there, but because the residual stresses tend to be more

compressive. The phase constitution near the bond coat interface is different from that near the free


163

surface. In addition, a through thickness gradient in the stiffness of the top coat is obtained when

the coating is heat treated in a thermal gradient.

The Youngs modulus of as-sprayed zirconia-based top coats, measured by cantilever bending, is in

the range of 10-15 GPa. After heat treatment, the elastic modulus increases substantially due to

sintering. Measurements by the nanoindentation technique gave Young’s Modulus values

approaching that of the monolithic material and are not representative of the global stiffness of the

coating due to the small volume of material undergoing deformation.

Thick thermal barrier coatings exhibit microstructural features such as segmentation and branching

cracks, which affect the elastic properties, and are not present in thinner TBCs. Crack segmentation

density is strongly dependent on the substrate temperature during deposition. Coatings with higher

crack segmentation densities have lower Young’s modulus. The presence of branching-cracks

could be significant when considering spallation of the top coat.

The thermal conductivity of as-sprayed YSZ TBCs was measured using a novel steady-state

thermal conductivity measurement technique and was found to be 1.1 ± 0.10 W m-1

K-1

. Laser flash

and hot disk techniques were also used to measure the thermal conductivity of as-sprayed TBCs,

with reasonably good agreement.

The thermal conductivity of DSZ top coats has been found to be higher than that of YSZ, while that

of YLaSZ is lower than YSZ. The results suggest that oxygen vacancies play a dominant role in

phonon scattering and hence in reducing the thermal conductivity, whereas phonon scattering due to

mass difference between dopant and host ion has a weaker effect. Heat treatment results in an

increase of the thermal conductivity due to sintering. This increase is more pronounced at 1400oC,

whereas at 1200oC hardly any increase is observed. The relative increase in the thermal

conductivity is higher for the YLaSZ top coats.

Thermal conduction through the pores affects the overall thermal conductivity of TBCs. Thermal

conductivity of YSZ plasma sprayed TBCs in vacuum was found to be ~15% lower than in air.

Predictions from the TFR model show that higher gas pressures in the pores, up to the range typical

of gas turbine service conditions, can significantly raise the pore conductivity, particularly at

relatively high temperatures.

Comparisons between experimentally measured conductivities, and predictions from the TFR

model, suggest that it should be possible to predict the conductivity from measurable

microstructural parameters.


164

9.2 Future work

In light of the work carried out in this study, future work should focus on the optimization of the

performance of the top coat. Better understanding of the sintering mechanisms is required. For

example, the relative contributions of the different sintering mechanisms to the sintering behaviour

of the top coats and the factors that influence them must be better understood. To achieve this, the

diffusion kinetics of the top coat materials must be studied in more detail and independently of the

microstructural features of PS top coats. It is suggested that simpler powder geometries, e.g.

spheres, should be used to study the diffusion kinetics of top coat materials. Also, it has been

shown that the role of impurities in sintering kinetics is of considerable importance, and studies

need to be undertaken to explore this further, for new coatings with differing impurity levels.

It has been shown that sintering of the top coat causes healing of small defects within the

microstructure and leads to an increase in stiffness and thermal conductivity. In parallel with

understanding the diffusion kinetics independently of microstructure, the microstructural features of

PS top coats must be investigated quantitatively and their role in sintering must be explored. It is

suggested that techniques such as small angle neutron scattering, measurement of surface area,

mercury intrusion porosimetry, and helium pycnometry, can be used to characterize the

microstructure of PS top coats. In addition the effect, if any, of top coat composition and impurity

levels on these microstructural features as well as the evolution of the microstructure after heat

treatment should be explored. In parallel, characterization of the microstructural features, and

understanding of their role in sintering, can be used in order to optimize deposition processes, so

that coatings may be deposited with larger flaw sizes, such that locking up of the structure by

sintering of such defects cannot occur.

The mechanisms of generation of segregation and branching cracks in thick TBCs, and their

significance on the thermomechanical stability of the top coat must be investigated. The presence

of branching cracks could be significant with respect to the spallation of part of top coat as these

interfaces are weak. Study of their evolution during isothermal heat treatment and heat treatment

under a thermal gradient is of interest.

An immediate aim must be to continue the investigation of the effect of heat treatment of TBC

under a strong thermal gradient on the stiffness of the top coat. The presence of the observed

layered microstructure, with a sintered outer surface and a porous structure near the BC/TC

interface, affects the strain tolerance of the top coat. Different coating elements will experience

different thermal histories and thus attain different stiffnesses. A more detailed analysis of the


165

effects of top coat stress at high temperatures on the rate of sintering is required. Measurement of

corresponding changes in porosity is necessary.

The potential of using different top coat materials to lower the thermal conductivity has been

demonstrated. In order to achieve the maximum reduction in thermal conductivity, studies should

be performed to investigate further the role of microstructure as well as the role of intrinsic thermal

conductivity of the top coat materials. Guidelines for the selection of top coat materials as well as

optimum microstructures to reduce the thermal conductivity need to be established.

This work could lead to significantly more sinter-resistant, low thermal conductivity plasma

sprayed top coats, which could provide a real alternative to EB-PVD coating in some applications.

References

166

References

1. Rolls-Royce, The Jet Engine. 1996, Birmingham: Renault Printing Co Ltd.

2. Sims, C.T., Stoloff N.S., Hagel W.C., Superalloys II. 1987: John Wiley & Sons.

3. Becker, B. Potential Benefits of Advanced Materials for Performance and Life Cycle Cost of

Large Industrial Gas Turbines. in 5th International Charles Parson Turbine Conference:

Advanced Materials for 21st Century Turbines and Power Plant. 2000, edited by Strang, A.,

Banks, W.M., Conroy R.D., McColvin, G.M., Neal, J.C., and Simpson, S., Cambridge, UK,

p. 658-670.

4. Siegel, R. and Spuckler, C. M., Analysis of Thermal Radiation Effects on Temperatures in

Turbine Engine Thermal Barrier Coatings, Materials Science & Engineering A, 1998,

245(2), p. 150-159.

5. Endres, W. Design Principles of Gas Turbines. in Proceedings of the symposium on High

Temperature Materials in Gas Turbines. 1973, edited by Sahm PR, S.M., Baden,

Switzerland: Elsevier Scientific Publishing Company.

6. Kaysser W.A., Bartsch M., Krell T., Fritscher K., Leyens C., Schulz U., and M., Peters,

Ceramic Thermal Barriers for Demanding Turbine Applications, CFI-Ceramic Forum

International, 2000, 77(6), p. 32-35.

7. Beele, W., Marijnissen, G., and van Lieshout, A., The Evolution of Thermal Barrier

Coatings - Status and Upcoming Solutions for Today's Key Issues, Surface & Coatings

Technology, 1999, 121, p. 61-67.

8. Stover, D. and Funke, C., Directions of the Development of Thermal Barrier Coatings in

Energy Applications, Journal of Materials Processing Technology, 1999, 93, p. 195-202.

9. Winstone, M.R., Partridge, A., and Brooks, J.W. The Contribution of Advanced Materials to

Future Aeroengines. in 5th International Charles Parson Turbine Conference: Advanced

Materials for 21st Century Turbines and Power Plant. 2000, edited by Strang, A., Banks,

W.M., Conroy R.D., McColvin, G.M., Neal, J.C., and Simpson, S., Cambridge, UK, p. 780-

797.

10. Kimmel, J., Mutasim, Z., and Brentnall, W., Effects of Alloy Composition on the

Performance of Yttria Stabilized Zirconia-Thermal Barrier Coatings, Journal of

Engineering for Gas Turbines and Power - Transactions of the ASME, 2000, 122, p. 393-

400.

11. Price, F.C., Stiles, W.R., and Singh, C. Materials for Marine, Light Industrial and

Aeroderivative Gas Turbines. in 5th International Charles Parson Turbine Conference:

Advanced Materials for 21st Century Turbines and Power Plant. 2000, Cambridge, UK.

12. McLean, M., Nickel-base superalloys: current status and potential, Philosophical

Transactions of the Royal Society of London, 1995, A(351), p. 419-433.

13. Backmann, D.G. and Williams, J.C., Advanced Materials for Aircraft Engine Applications,

Science, 1992, 255(February), p. 1082-1087.

14. Miller, R. A., Thermal Barrier Coatings for Aircraft Engines: History and Directions,

Journal of Thermal Spray Technology, 1997, 6(1), p. 35-42.

15. Taylor, R., Brandon, J.R., and Morell, P., Microstructure, Composition and Property

Relationships of Plasma Sprayed Thermal Barrier Coatings, Surface & Coatings

Technology, 1992, 50, p. 141-149.

References

167

16. DeMasi-Marcin, J. T. and Gupta, D. K., Protective coatings in the gas turbine engine,

Surface & Coatings Technology, 1994, 68/69, p. 1-9.

17. Sims, C.T. and Hagel, W. C., The Superalloys. 1972: John Wiley & Sons.

18. Schulz, U., Leyens, C., Fritscher, K., Peters, M., Saruhan-Brings, B., Lavigne, O., Dorvaux,

J-M., Poulain, M., Mevrel, R., and Caliez, M., Some Recent Trends in Research and

Technology of Advanced Thermal Barrier Coatings, Aerospace Science and Technology,

2003, 7, p. 73-80.

19. Meetham, GW, Superalloys in gas turbine engines, in The Metallurgist and Materials

Technologist, 1982, p. 387-392.

20. Caron, P. and Khan, T., Evolution of Ni-based superalloys for single crystal gas turbine

blade applications, Aerospace Science and Technology, 1999, 3, p. 513-523.

21. Fox, A.C. and Clyne, T.W., Oxygen Transport through the Zirconia Layer in Plasma

Sprayed Thermal Barrier Coatings, Surface & Coatings Technology, 2004, 184, p. 311-321.

22. Wood J.H., Goldman E.H., Protective Coatings, in Superalloys II, edited by Sims, C.T.,

Stoloff N.S., Hagel W.C., 1987, John Wiley & Sons. p. 359-384.

23. Evans, A.G, Mumm, D.R., Hutchinson, J.W., Meier, G.H., and Pettit, F.S., Mechanisms

controlling the Durability of Thermal Barrier Coatings, Progress in Materials Science, 2001,

46, p. 505-553.

24. Tawancy, H. M., Abbas, N. M., and Bennett, A., Role of Y during High Temperature

Oxidation of an M-Cr-Al-Y Coating on an Ni-base Superalloy, Surface & Coatings

Technology, 1994, 68/69, p. 10-16.

25. Choquet, P. and Mevrel, R., Microstructure of Alumina Scales Formed on NiCoCrAl Alloys

with and without Yttrium, Materials Science & Engineering, 1989, A120, p. 153-159.

26. Taylor, M. P. and Evans, H. E., The Influence of Bond Coat Surface Roughness and

Structure on the Oxidation of a Thermal Barrier Coating System, Materials Science Forum,

2001, 369-3, p. 711-717.

27. Busso, E.P., Lin, J., Sakurai, S., and Nakayama, M., A Mechanistic Study of Oxidation-

Induced Degradation in a Plasma-Sprayed Thermal Barrier Coating System. Part I: Model

Formulation, Acta Materialia, 2001, 49, p. 1515-1528.

28. Thompson, J.A, Tsui, Y.C., Reed, R.C., Rickerby, D.S., and Clyne, T.W., Creep of Plasma

Sprayed CoNiCrAlY and NiCrAlY Bond Coats and its Effects on Residual Stresses During

Thermal Cycling of Thermal Barrier Coating Systems, in High Temperature Surface

Engineering, edited by Nicholls, J. and Rickerby, D.S., 2000, IOM: Edinburgh, UK. p. 199-

212.

29. Evans, H. E. , Strawbridge, A. , Carolan, R. A. , and Ponton, C. B., Creep Effects on the

Spallation of an Alumina Layer from a NiCrAlY Coating, Materials Science & Engineering,

1997, A225, p. 1-8.

30. Rosler, J., Baker, M., and Volgmann, M., Stress state and failure mechanisms of thermal

barrier coatings: Role of creep in thermally grown oxide, Acta Materialia, 2001, 49(18), p.

3659-3670.

31. Herman, H. and Shankar, N. R., Survivability of Thermal Barrier Coatings, Materials

Science & Engineering, 1987, 88, p. 69-74.

References

168

32. Langhahr, P.A, Oberacker, R., and Hoffmann, M.J., Long-Term Behaviour and Application

Limits of Plasma-Sprayed Zirconia Thermal Barrier Coatings, Journal of the American

Ceramic Society, 2001, 84(6), p. 1301-1208.

33. Porter, D.L. and Heuer, A.H., Microstructural Development in MgO-Partially Stabilized

Zirconia (Mg-PSZ), Journal of the American Ceramic Society, 1979, 62, p. 298-305.

34. Bennett, A., Properties of Thermal Barrier Coatings, Materials Science and Technology,

1986, 2, p. 257-261.

35. Maier, R.D., Scheuermann, C.M., and Andrews, C.W., Degradation of Two Layer Thermal

Barrier Coating under Thermal Cycling, American Ceramic Society Bulletin, 1981, 60(5), p.

555-560.

36. Wilden J., Wank A., Application study on Ceria Based Thermal Barrier Coatings, Mat.-

wiss. u. Werkstofftech, 2001, 32, p. 654-659.

37. Lee, E.Y., Sohn, Y., Jha, S.K., Holmes, J.W., and Sisson, R.D. Jr., Phase transformations of

plasma sprayed zirconia-ceria thermal barrier coatings, Journal of the American Ceramic

Society, 2002, 85(8), p. 2065-2071.

38. Brandon, J. R. and Taylor, R., Phase Stability of Zirconia-Based Thermal Barrier Coatings

Part II. Zirconia-Ceria Alloys, Surface & Coatings Technology, 1991, 46, p. 91-101.

39. Khor, K. A. and Yang, J., Transformability of t-ZrO2 and lattice parameters in plasma

sprayed rare-earth oxides stabilized zirconia coatings, Scripta Materialia, 1997, 37(9), p.

1279-1286.

40. Hamacha, R, Fauchais, P., and Nardou, F., Influence of Dopant on the Thermal Properties

of Two Plasma-sprayed Zirconia Coatings 1. Relationship Between Powder Characteristics

and Coating Properties, Journal of Thermal Spray Technology, 1996, 5(4), p. 431-438.

41. Nicholls, J.R., Lawson K.L, Rickerby D.S., and Morrell, P., Advanced Processing of TBC's

for Reduced Thermal Conductivity, in AGARD-R-823, 1997, NASA: Aalborg, Denmark, p.

6.1-6.9.

42. Grunling, H. W. and Mannsmann, W., Plasma-Sprayed Thermal Barrier Coatings For

Industrial Gas- Turbines - Morphology, Processing and Properties, Journal de Physique IV,

1993, 3(C7), p. 903-912.

43. Morrell, P. and Taylor, R., eds. Thermal Diffusivity and Phase Structure of Zirconia-Based

Thermal Barrier Coatings. Advances in Ceramics. Vol. Volume 24: Science and

Technology of Zirconia III. 1988, The American Ceramics Society.

44. Klemens, P.G. and Gell, M., Thermal Conductivity of Thermal Barrier Coatings, Materials

Science & Engineering, 1998, A245, p. 143-149.

45. Scardi, P., Leoni, M., Cernuschi, F., and Figari, A., Microstructure and heat transfer

phenomena in ceramic thermal barrier coatings, Journal of the American Ceramic Society,

2001, 84(4), p. 827-835.

46. Nicholls, J.R., Jaslier, Y., and Rickerby, D.S., Erosion and Foreign Object Damage of

Thermal Barrier Coatings, Materials Science Forum, 1997, 251, p. 935-948.

47. Nicholls, J.R., Deakin, M.J., and Rickerby, D.S., A Comparison between the Erosion

Behaviour of Thermal Spray and Electron Beam Physical Vapour Deposition Thermal

Barrier Coatings, Wear, 1999, 235, p. 352-361.

References

169

48. Strangman, T.E., Thermal Barrier Coatings for Turbine Airfoils, Thin Solid Films, 1985,

127, p. 93-105.

49. Morrell, P. and Rickerby, D.S., Advantages/Disadvantages of Various TBC Systems as

perceived by the Engine Manufacturer, in AGARD-R-823, 1998, NASA: Denmark, p. 20-1 -

20-9.

50. Hass, D.D., Slifka, A. J., and Wadley, H. N. G., Low Thermal conductivity Vapor Deposited

Zirconia Microstructures, Acta Materialia, 2001, 49(6), p. 973-983.

51. Fisher, I.A., Variables Influencing the Characteristics of Plasma-Sprayed Coatings,

International Metals Reviews, 1972, 17, p. 117-129.

52. Fauchais, P. and Vardelle, M., Plasma Spraying : Present and Future, Pure & Applied

Chemistry, 1994, 66(6), p. 1247-1258.

53. Apelian, D., Paliwal, M., Smith, R.W., and W.F., Schilling, Melting and Solidification in

Plasma Deposition - Phenomenological Review, International Metals Reviews, 1983, 28(5),

p. 271-294.

54. Heimann, R. B., Plasma-Spray Coating Principles and applications. 1996, Weinhein,

Germany: VCH Verlagsgesellschaft.

55. Gill, S. C., Residual Stresses in Plasma Sprayed Deposits, PhD Thesis, in Department of

Materials Science and Metallurgy, 1991, University of Cambridge: Cambridge.

56. Sampath, S and Herman, H. Microstructural Development of Plasma Sprayed Coatings. in

12th International Conference on Thermal Spray. 1989, edited by Berndt, C.C., The

Welding Institute, Cambridge, U.K., p. 761-766.

57. Matejka, D. and Benko, B., Plasma Spraying of Metallic and Ceramic Materials. 1989:

John Wiley & Sons. 280.

58. Schlichting, K.W., Padture, N.P., and Klemens, P.G., Thermal Conductivity of Dense and

Porous Yttria-stabilized Zirconia, Journal of Materials Science, 2001, 36, p. 3003-3010.

59. Kuroda, S. and Clyne, T.W., The Quenching Stress in Thermally Sprayed Coatings, Thin

Solid Films, 1991, 200, p. 49-66.

60. Thompson, J.A. and Clyne, T.W., The Effect of Heat Treatment on the Stiffness of Zirconia

Top Coats in Plasma-Sprayed TBCs, Acta Materialia, 2001, 49(9), p. 1565-1575.

61. Green, D.J., An Introduction to the Mechanical Properties of Ceramics. 1 ed. Cambridge

Solid State Science Series, ed. Clarke, D.R., Suresh, S. and Ward, I. M. 1998: Cambridge

University Press.

62. Malzbender, J. and Steinbrech, R. W., Determination of the Stress-Dependent Stiffness of

Plasma-Sprayed Thermal Barrier Coatings Using Depth-Sensitive Indentation, Journal of

Materials Research, 2003, 18(8), p. 1975-1984.

63. Eskner, M. and Sandstrom, R, Measurement of the Elastic Modulus of a Plasma-Sprayed

Thermal Barrier Coating using Spherical Indentation, Surface & Coatings Technology,

2004, 177, p. 165-171.

64. Kulkarni, A, Vaidya, A. , Goland, A., Sampath, S., and Herman, H., Processing Effects on

Porosity-Property Correlations in Plasma Sprayed Yttria-Stabilized Zirconia Coatings,

Materials Science and Engineering A, 2003, 359, p. 100-111.

References

170

65. Siebert, B., Funke, C., Vassen, R., and Stover, D., Changes in porosity and Young's

Modulus due to sintering of plasma sprayed thermal barrier coatings, Journal of Materials

Processing Technology, 1999, 93, p. 217-223.

66. Guo, H. B., Xu, H. B., Gong, S. K., and Liu, F. S., Effect of surface strengthening on high

temperature oxidation and bonding strength of EB-PVD thermal barrier coating, Rare

Metal Materials Engineering, 2001, 30(4), p. 314-317.

67. Pawlowski, L. and Fauchais, P., Thermal Transport Properties of Thermally Sprayed

Coatings, International Materials Reviews, 1992, 37(6), p. 271-289.

68. Schwingel, D., Taylor, R., Haubold, T., Wirgen, J., and Gaulco, C., Mechanical and

Thermophysical properties of thick PYSZ Thermal Barrier Coatings:Correlation with

Microstructure and Spraying Parameters, Surface & Coatings Technology, 1998, 108-109,

p. 99-106.

69. Ahmaniemi, S., Vuoristo, P., and Mantyla, T., Mechanical and elastic properties of

modified thick thermal barrier coatings, Materials Science and Engineering A, 2004, 366, p.

175-182.

70. Gadow, R. and Lischka, M., Lanthanum Hexa-aluminate - Novel Thermal Barrier Coatings

for Gas Turbine Applications - Materials and Process Development, Surface & Coatings

Technology, 2002, 151, p. 392-399.

71. Pfender, E., Fundamental Studies Associated with the Plasma Spray, Surface & Coatings

Technology, 1988, 34, p. 1-16.

72. Vaidya A., Bancke G., Sampath S., and H., Herman. Influence of Process Variables on the

Plasma Sprayed Coatings: An Integrated Study. in Thermal Spray 2001 :New Surfaces for a

New Millenium. 2001, edited by Bernd C.C., K.K.A., Lugscheider E.F.,, Materials Park,

Ohio, USA: ASM International.

73. Fauchais P., Vardelle M., Vardelle A., Bianchi L., Leger A.C.,, Parameters Controlling the

Generation and Properties of Plasmas Sprayed Zirconia Coatings, Plasma Chemistry and

Plasma Processing, 1996, 16(1), p. 99S-125S.

74. Smith, R. W., Wei, D., and Apelian, D., Thermal Plasma Materials Processing -

Applications and Opportunities, Plasma Chemistry and Plasma Processing, 1989, 9, p.

135S-165S.

75. Kuroda, S., Fukushima, T., and Kitahara, S., Generation Mechanisms of Residual-Stresses

in Plasma-Sprayed Coatings, Vacuum, 1990, 41(224), p. 1297-1299.

76. Tsui, Y. C., Adhesion of Plasma Sprayed Coatings, PhD thesis, in Department of Material

Science and Metallurgy, 1996, University of Cambridge: Cambridge, UK.

77. Scardi, P., Leoni, M., Bertini, L., and Bertamini, L., Residual Stress in Partially-Stabilised

Zirconia TBCs: Experimental Measurement and Modelling, Surface & Coatings Technology,

1997, 94-5, p. 82-88.

78. Scardi, P., Leoni, M., and Bertamini, L., Residual Stresses in Plasma Sprayed Partially

Stabilised Zirconia TBCs - Influence of the Deposition Temperature, Thin Solid Films, 1996,

278, p. 96-103.

79. Watson, J.W. and Levine, S.R., Deposition Stress Effects on the Life of Thermal Barrier

Coatings on Burner Rigs, Thin Solid Films, 1984, 119, p. 185-193.

References

171

80. Teixeira, V., Andritschky, M., Fischer, W., Buchkremer, H.P., and Stover, D., Analysis of

Residual Stresses in Thermal Barrier Coatings, Journal of Materials Processing Technology,

1999, 93, p. 209-216.

81. Teixeira, V., Andritschy, M., Fischer, W., Buchkremer, H.P., and Stover, D., Effects of

Deposition Temperature and Thermal Cycling on Residual Stress State in Zirconia-Based

Thermal Barrier Coatings, Surface & Coatings Technology, 1999, 121, p. 103-111.

82. Hobbs, M. K. and Reiter, H., Residual Stresses in ZrO2-8%Y2O3 Plasma-Sprayed Thermal

Barrier Coatings, Surface & Coatings Technology, 1988, 34, p. 33-42.

83. Kingswell, R., Scott, K. T., and Sorensen, B., Measurement of Residual Stress in Plasma

Sprayed Ceramic Coatings, in 2nd Plasma-Technik-Symposium, edited by Blum-Sandmeier,

S., Eschnauer, H., Huber, P., and Nicoll, A.R., 1991, Plasma Technik. p. 377-388.

84. Kuroda, S., Kukushima, T., and Kitahara, S., In situ Measurement of Coating Thickness

during Thermal Spraying using an Optical Displacement Transducer, Journal of Vacuum

Science & Technology, 1987, A5, p. 82-87.

85. Kuroda, S., Kukushima, T., and Kitahara, S., Simultaneous Measurement of Coating

Thickness and Deposition Stress During Thermal Spraying, Thin Solid Films, 1988, 164, p.

157-163.

86. Kuroda, S., Fukushima, T., and Kitahara, S. Significance of the Quenching Stress in the

Cohesion and Adhesion of Thermally Sprayed Coatings. in 13th Int. Conf. Thermal Spraying.

1992, edited by Berndt, C.C., Orlando, Florida, USA: ASM, Metals Park, Ohio, p. 903-909.

87. Kuroda, S. and Clyne, T.W. The Origin and Quantification of the Quenching Stress

Associated with Splat Cooling during Spray Deposition. in 2nd Plasma Technik Symposium.

1991, edited by Eschenauer, H., Huber, P., Nicoll, A.R., and Sandmeier, S., Lucerne,

Switzerland: Plasma Technik, p. 273-284.

88. Stoney, G.G., The Tension of Metallic Films Deposited by Electrolysis, Proceedings of the

Royal Society of London, 1909, A82, p. 172-175.

89. Brenner, A. and Senderoff, S., Calculation of Stress in Electrodeposits from the Curvature

of a Plated Strip, Journal of Research of the National Bureau of Standards, 1949, 42, p. 105-

123.

90. Gill, S. C. and Clyne, T. W., Residual Stress Modelling and Characterisation of Thermally

Sprayed Ceramic Coatings, in High Performance Ceramic Films and Coatings, edited by

Vincenzini, P., 1991, Elsevier. p. 339-352.

91. Gill, S. C. and Clyne, T. W., Investigation of Residual Stress Generation during Thermal

spraying by Continuous Curvature Measurement, Thin Solid Films, 1994, 250, p. 172-180.

92. Matejicek, J., Sampath, C., Brand, P.C., and Prask, H.J., Quenching, Thermal and Residual

Stress in Plasma Sprayed Deposits: NiCrAlY and YSZ Coatings, Acta Materialia, 1999,

47(2), p. 607-617.

93. Levit, M., Grimberg, I., and Weiss, B.-Z., Residual Stresses in Ceramic Plasma-Sprayed

Thermal Barrier Coatings: Measurement and Calculation., Materials Science and

Engineering, 1995, A206, p. 30-38.

94. Zhuang, H. and Zhang, T. A study of the Residual Stresses in a ZrO2 + 8%Y2O3 Coatings. in

2nd Plasma-Technik-Symposium. 1991, edited by Blum-Sandmeier, S., Eschnauer, H.,

Huber, P., and Nicoll, A.R., Lucerne, Switzerland: Hafliger Druck AG, Wettingen, p. 311-

318.

References

172

95. Clyne, T.W. and Gill, S.C., Residual Stresses in Thermally Sprayed Coatings and their

Effect on Interfacial Adhesion - A Review of Recent Work, Journal of Thermal Spray

Technology, 1996, 5, p. 1-18.

96. Kesler, O., Matejicek, J., Sampath, S., Suresh, S., Gnaeupel-Herold, T., Brand, P.C., and

Prask, H.J., Measurement of Residual Stress in Plasma-Sprayed Metallic, Ceramic and

Composite Coatings, Materials Science and Engineering, 1998.

97. Lifshin, E., X-ray Characterization of Materials. 1999: Wiley-VCH.

98. Reimers, W., Broda, M., Dantz, G., Liss, K.-D., Pyzala, A., Schmakers, T., and Tschenscher,

T., Evaluation of Residual Stresses in Bulk Materials by High Energy Synchrotron

Radiation, Journal of Non-destructive Evaluation, 1998, 17(3), p. 129-140.

99. Thompson, J.A., Matejicek, J., and Clyne, T.W. Modelling and Neutron Diffraction

Measurement of Stresses in Sprayed TBCs. in Superalloys 2000. 2000, edited by Pollock,

T.M., Kissinger, R.D., Bowman, R.R., Green, K.A., McLean, M., Olsen, S.L., and Schirra,

J.J., Seven Springs, USA: TMS, p. 639-647.

100. Scardi, P., Leoni, M., Bertini, L., Bertamini, L., and Cernuschi, F., Strain Gradients in

Plasma - Sprayed Zirconia Thermal Barrier Coatings, Surface and Coatings Technology,

1998, 108-9, p. 93-98.

101. Rickerby, D. S., Scott, K. T., and Eckold, G. Analysis of the Residual Stresses in Plasma

Sprayed Coatings. in First Plasma-Technik Symposium. 1988, edited by Eschenauer, H.,

Huber, P., Nicoll, A.R., and Sandmeier, S., Wöhlen, Switzerland: Plasma Technik, p. 267-

276.

102. Bengtsson, P. and Persson, C., Modelled and Measured Residual Stresses in Plasma

Sprayed Thermal Barrier Coatings, Surf. & Coatings Tech., 1997, 92, p. 78-86.

103. Gill, S. C. and Clyne, T. W., Modeling of the generation of residual stress during thermal

spraying of ceramic coatings, High Performance Ceramic Films and Coatings, 1991.

104. Rickerby, D. S., Eckold, G., Scott, K. T., and Buckley-Golder, I. M., The Interrelationship

between Internal Stress, Processing Parameters and Microstructure of Physically Vapour

Deposited and Thermally Sprayed Coatings, Thin Solid Films, 1987, 154, p. 125-141.

105. Stover, D., Jager, D. A., and Schutz, H. G. Residual Stresses in Low Pressure Plasma

Sprayed Chromia Coatings. in Proceedings of The Fourth National Thermal Spray

Conference. 1991, edited by Bernecki, T.F., Pittsburgh, PA, USA: The Materials

Information Society, p. 215-219.

106. Elsing, R., Knotek, O., and Balting, U., Calculation of Residual Thermal Stress in Plasma-

Sprayed Coatings, Surface & Coatings Technology, 1990, 43/44, p. 416-425.

107. Tsui, Y.C. and Clyne, T.W. Adhesion of Thermal Barrier Coating Systems and

Incorporation of an Oxidation Barrier Layer. in Thermal Spray: Practical Solutions for

Engineering Problems. Proceedings of the 9th Nat. Thermal Spray Conf. 1996, edited by

Berndt, C.C., Cincinatti: ASM, p. 275-284.

108. Tsui, Y.C. and Clyne, T.W., An Analytical Model for Predicting Residual Stresses in

Progressively Deposited Coatings. Part 1: Planar Geometry, Thin Solid Films, 1997, 306,

p. 23-33.

109. Thompson, J.A., Thermomechanical Behaviour of Plasma-Sprayed Thermal Barrier

Coatings, PhD Thesis, in Department of Materials Science and Metallurgy, 2000,

University of Cambridge: Cambridge, UK.

References

173

110. Lugscheider, E., Bobzin, K., Barwulf, S., and Etzkorn, A., Mechanical properties of EB-

PVD-thermal barrier coatings by nanoindentation, Surface & Coatings Technology, 2001,

138(1), p. 9-13.

111. Wallace, J.S. and Ilavsky, J., Elastic Modulus Measurements in Plasma Sprayed Deposits,


112. Zhu, D.M. and Miller, R.A., Thermal Conductivity and Elastic Modulus Evolution of

Thermal Barrier Coatings under High Heat Flux Conditions, Journal of Thermal Spray

Technology, 2000, 9, p. 175-180.

113. Siebert, S., Funke, C., Vassen, R., and Stover, D., Changes in Porosity and Young's

Modulus due to Sintering of Plasma-Sprayed Thermal Barrier Coatings, Journal of

Materials Processing Technology, 1999, 93(217-223).

114. Leigh, S.H, Lin, C.K, and Berndt, C. C., Elastic response of thermal spray deposits under

indentation tests, Journal of the American Ceramic Society, 1997, 80(8), p. 2093-2099.

115. Vasquez, D.L., Kucuk, A., Lima, R. S., Senturk, U., and Berndt, C. C. Elastic modulus

measurements of air plasma sprayed yttria partially stabilized zirconia coatings using laser

ultrasonics and indentation techniques. in Thermal Spray 2001:New Surfaces for a New

Millenium. 2001, edited by Berndt, C.C., Khor, K.A., and Lugscheider, E.F., Ohio, USA:

ASM International, p. 1045-1050.

116. Steffens, H.D. and Fischer, U. Correlation between microstructure and physical properties

of plasma sprayed zirconia properties. in Thermal Spray: Advances in Coatings Technology.

1998, edited by Houck, D.L., Ohio, USA: ASM International, p. 167-173.

117. Duan, K. and Steinbrech, R. W., Influence of sample deformation and porosity on

mechanical properties by instrumented microindentation technique, Journal of the European


118. Pawlowski, L., The Science and Engineering of Thermal Spray Coatings. 1995, Chichester,

England: J. Wiley & Sons Ltd.

119. Kulkarni, A., Wang, Z., Nakamura, T., Sampath, S., Goland, A., Herman, H., Allen, J.,

Ilavsky, J., Long, G., Frahm, J., and Steinbrech, R. W., Comprehensive microstructural

characterization and predictive property modeling of plasma-sprayed zirconia coatings,

Acta Materialia, 2003, 51(9), p. 2457-2475.

120. Kroupa, F., and Plesek, J., Nonlinear Elastic Behaviour in Compression of Thermally

Sprayed Materials, Materials Science and Engineering, 2002, 328(1-2), p. 1-7.

121. Harok, V. and Neufuss, K., Elastic and inelastic effects in compression in plasma-sprayed

ceramic coatings, Journal of Thermal Spray Technology, 2001, 10(1), p. 126-132.

122. Kroupa, F., and Dubsky, J., Pressure Dependence of Young's Moduli of Thermal Sprayed

Materials, Scripta Materialia, 1999, 40(11), p. 1249-1254.

123. Choi, S. R., Zhu, D. , and Miller, R. A., Deformation and Tensile Cyclic Fatigue of Plasma-

Sprayed ZrO2-8wt % Y2O3 Thermal barrier Coatings, Ceramic Engineering and Science

Proceedings, 2001, 22(4), p. 427-434.

124. Rejda, E. F., Socie, D. F., and Beardsley, B., Fatigue behavior of a plasma-sprayed

8%Y2O3-ZrO2 thermal barrier coating, Fatigue &Fracture of Engineering Materials and

Structures, 1997, 20(7), p. 1043-1050.

125. Gibson, L.J. and Ashby, M.F., The Mechanics of Three-Dimensional Cellular Materials,

Proceedings of the Royal Society of London, 1982, A382, p. 43-59.

References

174

126. MacKenzie, J.K., The Elastic Constants of a Solid Containing Spherical Holes, Proceedings

of the Physical Society, 1950, B63, p. 2-11.

127. Eshelby, J.D., The Determination of the Elastic Field of an Ellipsoidal Inclusion and

Related Problems, Proceedings of the Royal Society of London, 1957, 241, p. 376-396.

128. Leigh, S.H. and Berndt, C.C., Modelling of Elastic Constants of Plasma Spray Deposits with

Ellipsoid-Shaped Voids, Acta Materialia, 1999, 47(5), p. 1575-1586.

129. Sevostianov, I. and Kachanov, M., Modeling of the anisotropic elastic properties of plasma-

sprayed coatings in relation to their microsturcture, Acta Materialia, 2000, 48, p. 1361-

1370.

130. Kachanov, M., Sevostianov, I., and Shafiro, B., Explicit cross-property correlations for

porous materials with anisotropic microstructures, Journal of Mechanics and Physics of

Solids, 2001, 49(1), p. 1-25.

131. Wang, Z, Kulkarni, A, Deshpande, S, Nakamura, T, and Herman, H, Effects of pores and

interfaces on effective properties of plasma sprayed zirconia coatings, Acta Materialia, 2003,

51, p. 5319-5334.

132. McPherson, R. and Shafer, V., Interlamellar Contact within Plasma-Sprayed Coatings, Thin

Solid Films, 1982, 97, p. 201-204.

133. Dobbins, T.A., Allen, A.J., Ilavsky, J., Long, G.G., P.R.Jemian, Kulkarni, A., and Herman,

H., Recent Development in the Characterization of Anizotropic Void Populations in Thermal

Barrier Coatings Using Ultra-small Angle X-ray Scattering, Ceramic Engineering and

Science Proceedings, 2003, 24(3), p. 517-524.

134. McPherson, R., A Review of Microstructure and Properties of Plasma Sprayed Ceramics

Coatings, Surface & Coatings Technology, 1989, 39/40, p. 173-181.

135. Strunz, P., Schumacher, G., Vassen, R., and Wiedenmann, A., In Situ SANS Study of Pore

Microstructure in YSZ Thermal Barrier Coatings, Acta Materialia, 2004, 52, p. 3305 -3312.

136. Allen, A. J., Ilavsky, J., Long, G. G., Wallace, J. S., Berndt, C. C., and Herman, H.,

Microstructural Characterisation of Yttria-stabilised Zirconia Plasma-sprayed Deposits

using Multiple Small-Angle Neutron Scattering, Acta Materialia, 2001, 49, p. 1661-1675.

137. Boukari, H., Allen, A. J., Long, G. G., Ilavsky, J., Wallace, J. S., Berndt, C. C., and Herman,

H., Small-angle neutron scattering study of the role of feedstock particle size on the

microstructural behavior of plasma-sprayed yttria-stabilized zirconia deposits, Journal of


138. Ilavsky, J., Long, G.G., Allen, A.J., and Berndt, C.C., Evolution of the Void Structure in

Plasma-Sprayed YSZ Deposits During Heating, Materials Science and Engineering, 1999,

272(1), p. 215-221.

139. Lughi, V., Tolpygo, V.K., and Clarke, D. R., Microstructural aspects of the sintering of

thermal barrier coatings, Material Science and Engineering, 2004, A272, p. 215-221.

140. Lelait, L., Alperine, S., and Mevrel, R., Alumina Scale Growth at Zirconia-MCrAlY

Interface: A Microstructural Study, Journal of Materials Science, 1992, 27, p. 5-12.

141. Lipkin, D.N. and Clarke, D. R., Measurement of the stress in oxide scales formed by

oxidaton of alumina-forming alloys, Oxidation of Metals, 1996, 45(3/4), p. 267-280.

References

175

142. Tolpygo, V.K., Dryden, J.R., and Clarke, D.R., Determination of the Growth Stress and

Strain in -Alumina Scales during the Oxidation of Fe-22Cr-4.8Al-0.3Y Alloy, Acta

Materialia, 1998, 46, p. 927-937.

143. Tolpygo, V. K. and Clarke, D. R., Spalling failure of Alumina Films grown by Oxidation.

I: Dependence on Cooling Rate and Metal Thickness, Materials Science and Engineering,

2000, A278, p. 142-150.

144. Clarke, D.R., Christenson, R.J., and Tolpygo, V., The Evolution of Oxidation Stresses in

Zirconia Thermal Barrier Coated Superalloy Leading to Spalling Failure, Surface &

Coatings Technology, 1997, 94-5, p. 89-93.

145. Ali, M.Y., Nusier, S.Q., and Newaz, G.M., Creep effects on early damage initiation in a

TBC system, Journal of Materials Science, 2004, 39(10), p. 3383-3390.

146. Tolpygo, V. K. and Clarke, D. R., Spalling failure of -Alumina Films grown by Oxidation:

II. Decohesion Nucleation and Growth, Materials Science and Engineering, 2000, A278, p.

151-161.

147. German, R.N., Sintering Theory and Practice. 1 ed. 1996, New York: John Wiley & Sons.

148. Shi, J.L., Thermodynamics and densification kinetics in solid state sintering of ceramics,

Journal of Materials Research, 1999, 14(4), p. 1398-1408.

149. Lay, K.W. Grain growth during sintering. in 3rd International Conference on Sintering and

Related Phenomena. 1972, edited by Kuczynski, G.C., Indiana, USA: Plenum Press, p. 65--

80.

150. Ashby, M.F., A First Report on Sintering Diagrams, Acta Metallurgica, 1974, 22, p. 275-

289.

151. Pask, J.A. and Hoge, C.E. Thermodynamic aspects of solid state sintering. in Sintering and

Related Phenomena. 1972, edited by Kuczynski, G.C., Notre Dame, Indiana, USA: Plenum

Press, New York, p. 229-238.

152. Burke, J.E., Role of Grain Boundaries in Sintering, Journal of the American Ceramic

Society, 1957, 40(3), p. 80-85.

153. Mendelev, M.I. and D.J., Srolovitz, Impurity effects on grain boundary migration,

Modelling and Simulation in Materials Science and Engineering, 2002, 10, p. R79-R-109.

154. Cahn, J.W., The impurity-drag effect in grain boundary motion, Acta Metallurgica, 1962, 10,

p. 769-798.

155. Brook, R.J., Howlett, S.P., and Wu, S.X. The use of solid solution additives in sintering. in

Sintering- Theory and Practice. 1981, edited by Kolar, d., Pejovnik, S., and Ristic, M.M.,

Yugoslavia: Elsevier Scientific Publishing Company, Amsterdam, p. 135-144.

156. Kroger, F.A. Point defects in compounds and their role in diffusion. in Sintering and

Related Phenomena. 1967, edited by Kuczynski, G.C., Hooton, N.A., and Gibbon, C.F.,

New York: Gordon and Breach.

157. Coblenz, W.S., Dynys, J.M., Cannon, R.M., and Coble, R.L. Initial Stage Solid State

Sintering Models. A Critical Analysis and Assessment. in Sintering and Related Phenomena.

1980, edited by Kuczynski, G.C.: Plenum Press, p. 36-43.

158. Kingery, W.D. and Berg, M., Study of the Initial Stages of Sintering Solids by Viscous Flow,

Evaporation-Condensation and Self-Diffusion, Journal of Applied Physics, 1955, 26(10), p.

1205-1212.

References

176

159. Coble, R.L., Sintering crystalline solids: I. Intermediate and Final Stage Diffusion models,

Journal of Applied Physics, 1961, 32(5), p. 787-792.

160. Kingery, W.D. and Berg, M., Study of the Initial Stages of Sintering Solids by Viscous Flow,

Evaporation-Condensation, and Self Diffusion, Journal of Applied Physics, 1955, 26(10), p.

1205-1212.

161. Kuczynski, G.C., Self-diffusion in sintering of metallic particles, Trans A.I.M.E., 1949,

185(February), p. 169-177.

162. Maximenko A.L. and Olevsky, E.A., Effective diffusion coefficients in solid-state sintering,

Acta Materialia, 2004, 52, p. 2953-2963.

163. Exner, H.E. and Petzow, G. A critical evaluation of shrinkage equations. in Sintering

Processes. 1979, edited by Kuczynski, G.C., Indiana, USA: Plenum Press, p. 107-120.

164. Coblenz. Initial stage solid state sintering models. A critical analysis and assessment. in

Sintering Processes. 1979, edited by Kuczynski, G.C., Indiana, USA: Plenum Press, p. 141-

157.

165. Exner, H.E., Solid-state sintering: critical assessment of theoretical concepts and

experimental methods, Powder Metallurgy, 1980, 4, p. 203-209.

166. Harmer, M.P. Use of Solid-Solution Additives in Ceramic Processing. in Advances in

Ceramics: Structure and Properties of MgO and Al2O3 ceramics. 1983, edited by Kingery,

W.D., Cambridge Massachusetts, USA: The American Ceramic Society.

167. Zhao, J. and Harmer, M.P., Effect of pore distribution on microstructure development: II

First- and second-generation pores, Journal of the American Ceramic Society, 1988, 71(7),

p. 530-539.

168. Herring, C., Effect of scale on sintering phenomena, Journal of Applied Physics, 1950, 21, p.

301-303.

169. Rhodes, W.H., Agglomerate and particle size effects on sintering of yttria stabilzed zirconia,

Journal of the American Ceramic Society, 1981, 64(1), p. 19-22.

170. Shi, J.L., Gao, J.H., Lin, Z.X., and Yen, T.S., Sintering Behavior of Fully Agglomerated

Zirconia Compacts, Journal of the American Ceramic Society, 1991, 74(5), p. 994-997.

171. Chappell, J.S., Ring, T.A., and Birchall, J.D., Particle size distribution effects on sintering

rates, Journal of Applied Physics, 1986, 6(1), p. 383-319.

172. German, R. M., Liquid Phase Sintering. 1985, New York: Plenum Press.

173. De Souza, D.P.F. and De Souza, M.F., Liquid Phase Sintering of RE2O3: YSZ Ceramics -

Part I - Grain Growth and Expelling of the Grain Boundary Glass Phase, Journal of

Materials Science, 1999, 34, p. 4023-4030.

174. Kingery, W.D., Introduction to Ceramics. 1960: Jonh Wiley and Sons.

175. Richerson, D.W., Modern Ceramic Engineering. 2nd ed, ed. Peter, H. 1992: Marcel Dekker

Inc.

176. del Olmo, L., Duran, P., and Moure, C. Sintering behaviour of an yttria stabilized zirconia.

in Sintering-theory and Practice. 1981, edited by Kolar, D., Pejovnik, S., and Ristic, M.M.,

Portooz, Yugoslavia: Elsevier Scientific, Amsterdam, p. 401-408.

References

177

177. Nasar, R. S., Cerqueira, M., Longo, E., and Varela, J. A., Sintering mechanisms of ZrO2

center dot MgO with addition of TiO2 and CuO, Ceramics International, 2004, 30(4), p. 571-

577.

178. Jorgensen, P.J. Diffusion controlled sintering in oxides. in Sintering and Related Phenomena.

1967, edited by Kuczynski, G.C., Hooton, N.A., and Gibbon, C.F.: Gordon and Breach,

New York, p. 401-422.

179. Brook, R.J. Preparation and Electrical Behaviour of Zirconia Ceramics. in Science and

Technology of Zirconia. 1981, edited by Heuer, A.H. and Hobbs, L.W., Cleveland, Ohio,

USA: The American Ceramic Society Inc., p. 272-285.

180. Chien, F. R. and Heuer, A. H., Lattice diffusion kinetics in Y2O3-stabilized cubic ZrO2 single

crystals: a dislocation loop annealing study, Philosophical Magazine A, 1996, 73, p. 681-

697.

181. Solmon, H., Chaumont, J., Dolin, C., and Monty, C. Zr, Y and O self diffusion in Zr(1-x)YxO2-

x/2. in Ceramic Transactions- Point Defects and Related Properties of Ceramics. 1991,

edited by Mason, T.O. and Routbort, J.L., p. 175-184.

182. Zhu, D.M. and Miller, R.A., Sintering and Creep Behaviour of Plasma-Sprayed Zirconia-

and Hafnia-Based Thermal Barrier Coatings, Surface & Coatings Technology, 1998, 109(1-

3), p. 114-120.

183. Kilo, M., Taylor, B., Argirusis, Ch., Borchardt, G., Lesage, B., Weber, S., Scherrer, S.,

Scherrer, H., Schroeder, M., and Martin, C., Cation self-diffusion of Ca, Y, and Zr in single

crystalline calcia- and yttria-doped zirconia, Journal of Applied Physics, 2003, 94(12), p.

7547-7552.

184. Kilo, M., Taylor, M.A., Argirusis, Ch., Borchardt, G., Weber, S., Scherrer, H., and Jackson,

R.A., Lanthanide transport in stabilized zirconias: Interaction between ionic radius and

diffusion coefficient, Journal of Chemical Physics, 2004, 12(11), p. 5482-5487.

185. Wu, S.X. and Brook, R.J., Sintering Aids for Zirconia Ceramics, Transactions and journal of

the British Ceramic Society, 1983, 82(6), p. 200-205.

186. Alleman, J.A., Michel, B., Marki, H.B., Gauckler, L.J., and Moser, E.M., Grain growth of

differently doped Zirconia, Journal of the European Ceramic Society, 1995, 15, p. 951-958.

187. Hwang, S-L. and Chen, I-W., Grain size control of tetragonal zirconia polycrystals using

the space change concept, Journal of the American Ceramic Society, 1990, 73(11), p. 3269-

3277.

188. Theunissen, G.S.A.M., Winnubst, A.J.A., and Burggraaf, A.J., Surface and grain boundary

analysis of doped zirconia ceramics studies by AES and XPS, Journal of Materials Science,

1992, 27, p. 5057-5066.

189. Stemmer, S., Vleugels, J., and Van Der Biest, O., Grain boundary segregation in high-

purity yttria-stabilized tetragonal zirconia polygrystals (T-TZP), Journal of the European

Ceramic Society, 1998, 18, p. 1565-1570.

190. Boutz, M.M.R., Winnubst, A.J.A., and Burggraaf, A.J., Yttria-Ceria Stabilized Tetragonal

Zirconia Polycrystals - Sintering, Grain Growth and Grain Boundary Segregation, Journal

of the European Ceramic Society, 1994, 13(2), p. 89-102.

191. Lynn Johnson, D. Impurity Effects in the Initial Sintering of Oxides. in Sintering and related

phenomena. 1967, edited by Kuczynski, G.C., Hooton, N.A., and Gibbon, C.F.: Gorden and

Breach, New York.

References

178

192. Verkerk, M., Winnubst, A., and Burggraaf, A., Effect of Impurities on Sintering and

Conductivity of Yttria-Stabilised Zirconia, Journal of Materials Science, 1982, 17, p. 3113-

3122.

193. Tekeli, S., Erdogan, M., and Aktas, B., Microstructural evolution in 8 mol% Y2O3-stabilized

cubic zirconia (8YSCZ) with SiO2 addition, Materials Science and Engineering a-Structural

Materials Properties Microstructure and Processing, 2004, 386(1-2), p. 1-9.

194. Shi, J.L., Yen, T.S., and Schubert, H., Effect of Small Amounts of Additives on the Sintering

of High-Purity Y-TZP, Journal of Materials Science, 1997, 32, p. 1341-1346.

195. Tsubakino, H., Fujiwara, T., Satani, K., and Ioku, S., Composition of grain boundary phase

formed in zirconia-3 mol% yttria containing alumina, Journal of Materials Science Letters,

1997, 16(17), p. 1472-1475.

196. Srdic, V.V., Winterer, M., and Hahn, H., Sintering Behaviour of Nanocrystalline Zirconia

Doped with Alumina Prepared by Chemical Vapour Synthesis, Journal of the American


197. Eaton, H.E. and Novak, R.C., Sintering Studies of Plasma Sprayed Zirconia, Surface &

Coatings Technology, 1987, 32, p. 227-236.

198. Vassen, R., Czech, N., Mallener, W., Stamm, W., and Stoever, D., Influence of Impurity

Conetent and Porosity of Plasma-Sprayed Yttria-Stabilized Zirconia Layers on the Sintering

Behaviour, Surface & Coatings Technology, 2001, 141, p. 135-140.

199. Klocker, T. and Clyne, T.W., The Effects of Bond Coat Roughness and Top Coat Purity on

the Adhesion of Thermal Barrier Coatings, Surface & Coatings Technology, 2002, p.

submitted.

200. Gibson, I.R., Dransfield, G.P., and Irvine, J.T.S., Sinterability of Commercial 8 mol%

Yttria-Stabilized Zirconia Powders and the Effect of Sintered Density on the Ionic

Conductivity, Journal of Materials Science, 1998, 33(17), p. 4297-4305.

201. Ilavsky, J., Long, G.G., and Allen, A.J. Evolution of the Microstructure of Plasma Sprayed

Deposits During Heating. in 5th International Thermal Spray Conference. 1998, edited by

Coddet, C., Nice, France: ASM International, p. 1641-1644.

202. Fabris, S., Paxton, A.T., and Finnis, M.W., A stabilization mechanism of zirconia based on

oxygen vacancies only, Acta Materialia, 2002, 50, p. 5171-5178.

203. Ho, S-M., On the structural chemistry of zirconium oxide, Materials Science and

Engineering, 1982, 54, p. 23-29.

204. Li, P. and Chen, I-W., Effect of dopants on zirconia stabilization - an-x-ray absorption study:

I. trivalent dopants, Journal of the American Ceramic Society, 1994, 77(1), p. 118-1128.

205. Yoshimura, M., Phase Stability of Zirconia, Ceramic Bulletin, 1988, 67(12), p. 1950-1955.

206. Yoshimura, M., Yashima, M., Noma, T., and Somiya, S., Formation of diffusionlessly

transformed tetragonal phases by rapid quenching of melts in ZrO2-RO1.5 systems (R=rare

earths), Journal of Materials Science, 1990, 25, p. 2011-2016.

207. Sheu, T-S., Tien, T-Y., and Chen, I-W., Cubic-to-tetragonal (t') transformation in zirconia-

containing systems, Journal of the American Ceramic Society, 1992, 75(5), p. 1108-1116.

208. Li, P. and Chen, I-W., Effect of dopants on zirconia stabilization - an-x-ray absorption study:

II. tetravalent dopants, Journal of the American Ceramic Society, 1994, 77(1), p. 1281-1288.

References

179

209. Mastelaro, V.R., Briois, V., de Souza D.P.F., and Silva, C.L., Structural studies of a ZrO2-

CeO2 doped system, Journal of the European Ceramic Society, 2003, 23(3), p. 273-282.

210. Brandon, J.R. and Taylor, R., Thermal Properties of Ceria and Yttria Partially Stabilized

Zirconia Thermal Barrier Coatings, Surface & Coatings Technology, 1989, 39/40, p. 143-

151.

211. Lee, C.H., Kim, H.K., Choi, H.S., and Ahn, H.S., Phase Transformation and Bond coat

Oxidation Behaviour of plasma-Sprayed Zirconia Thermal Barrier Coating, Surface &


212. Harmsworth, P. D. and Stevens, R., Phase Composition and Properties of Plasma-Sprayed

Zirconia Thermal Barrier Coatings, Journal of Materials Science, 1992, 27, p. 611-615.

213. Ilasky J., Stalick J.K., Wallace J., Thermal Spray yttria-Stabilized Zirconia Phase Changes

during Annealing, Journal of thermal Spray Technology, 2001, 10(3), p. 491-501.

214. VanValzah, J.R. and Eaton, H.E., Cooling Rate Effects on the Tetragonal to Monoclinic

Phase Transformation in Aged Plasma-Sprayed Yttria Partially Stabilised Zirconia, Surface

& Coatings Technology, 1991, 46, p. 289-300.

215. Lee, J.D., Ra, Y.H., Hong, K.T., and S.K., Hur, The Effect of Residual Stress on Martensitic

Transformation in Plasma-Sprayed ZrO2 (8 wt % Y2 O3 ) Coatings, Surface & Coatings

Technology, 1992, 54/55, p. 64-69.

216. Gupta, T. K., Lange, F. F., and Bechtold, J. H., Effect of Stress-Induced Phase

Transformation on the Properties of Polycrystalline Zirconia containing Metastable

Tetragonal Phase, Journal of Materials Science, 1978, 13, p. 1464-1470.

217. Moon, J, Choi, H, Kim, H, and Lee, C, The effects of heat treatment on the phase

transformation behaviour of plasma-sprayed stabilized ZrO2 coatings, Surface & Coatings

Technology, 2002, 155(1), p. 1-10.

218. Choi, H., Kim, H., and Lee, C., Phase evolution of plasma sprayed ceria and yttria

stabilized zirconia thermal barrier coating, Journal of Materials Science Letters, 2002, 21, p.

1359-1361.

219. Dal Maschio, R., Scardi, P., and Lutteroti, L., Influence of Ce3+

/Ce4+

ration on the phase

stability and residual stress field in ceria-yttria stabilized plasma-sprayed coatings, Journal

of Materials Science, 1992, 27, p. 5591-5596.

220. Portinha, A., Teixeira, V., Carneiro, J., Costa, M.F., Barrabas, N.P., and Sequeira, A.D.,

Stabilization of ZrO2 PVD coatings with Gd2O3, Surface & Coatings Technology, 2004,

188-189, p. 107-1115.

221. Rebollo, R.N., Fabrichnaya, O., and Levi, C.G., Phase stability of Y+Gd co-doped zirconia,

Z. Metallkd., 2003, 94(3), p. 163-170.

222. Khor, K. A. and Yang, J., Lattice parameters, tetragonality (c/a) and transformability of

tetragonal zirconia phase in plasma sprayed ZrO2-Er2O3 coatings, Materials Letters, 1997,

31, p. 23-27.

223. Khor, K. A. and Yang, J., Plasma spraying of samaria-stabilized zirconia powders and

coatings, Materials Letters, 1997, 31, p. 165-171.

224. Jones, R. L. and Mess, D., Improved tetragonal phase stability at 1400oC with scandia,

yttria-stabilized zirconia, Surface & Coatings Technology, 1996, 87-8(1-3), p. 94-101.

References

180

225. Leoni, M., Jones, R.L., and Scardi, P., Phase Stability of Scandia - Yttria Stabilized Zirconia

TBCs, Surface and Coatings Technology, 1998, 108-9, p. 107-113.

226. Stubican, V.S. and Hellman, J.R. Phase Equilibria in some Zirconia Systems. in First

International Conference on the Science and Technology of Zirconia. 1980, edited by Heuer,

A.H. and Hobbs, L.W., Cleveland, Ohio: American Ceramic Society, p. 25-36.

227. Bastide, B., Odier, P., and Coutures, J.P., Phase Equilibrium and Martensitic

Transformation in Lanthana-Doped Zirconia, Journal of the American Ceramic Society,

1988, 71(6), p. 449-453.

228. Andrievskaya E.R. and L.M., Lopato, Influence of composition of the t->m transformation

in the systems ZrO2-Ln2O3 (Ln= La, Nd, Sm, Eu), Journal of Materials Science, 2001, 30(10),

p. 2592-2596.

229. Thangadurai, P., Sabarinathan, V., Chandra Bose, A., and Ramasamy, S., Conductivity

behaviour of a cubic/tetragonal phase stabilized nanocrystalline La2O3-ZrO2, Journal of

Physics and Chemistry of Solids, 2004, 65, p. 1905-1912.

230. Vassen, R., Cao, X. Q., Tietz, F., Basu, D., and Stover, D., Zirconates as new materials for

thermal barrier coatings, Journal of the American Ceramic Society, 2000, 83(8), p. 2023-

2028.

231. Cao, X. Q., Vassen, R. and Stoever, D., Ceramic Materials for Thermal Barrier Coatings,

Journal of the European Ceramic Society, 2004, 24, p. 1-10.

232. Miller, R.A., Smialek, J.L., and Garlick, R.G. Phase Stability in Plasma-Sprayed, Partially

Stabilized Zirconia-Yttria. in First International Conference on the Science and Technology

of Zirconia. 1980, edited by Heuer, A.H. and Hobbs, L.W., Cleveland, Ohio: American

Ceramic Society, p. 241-253.

233. Berman, R., Thermal Conduction in Solids. 1976, Oxford: Clarendon.

234. Lee, D.W. and Kingery, W.D., Radiation Energy Transfer and Thermal Conductivity of

Ceramic Oxides, Journal of the American Ceramic Society, 1960, 43(11), p. 594-605.

235. Klemens, P. G., Theory of the Thermal Conductivity of Solids, in Thermal Conductivity,

edited by Tye, R.P., 1969, Academic Press: Cambridge, Massachusetts, U.S.A. p. 1-65.

236. Klemens, P. G., Thermal Resitance due to Point Defects at High Temperatures, Physical

Review A, 1960, 119(21), p. 507-509.

237. Siegel, R and Howell, J R, Thermal Radiation Heat Transfer. 1st ed. 1972, New York:

McGraw-Hill. 814.

238. Peelen, J.G. and Metselaar, R., Light scattering by pores in polycrystalline materials:

Transmission properties of alumina, Journal of Applied Physics, 1974, 45(1), p. 216-220.

239. Manara, J., Caps, R., Raether, F., and Fricke, J., Characterization of the pore structure of

alumina ceramics by diffuse radiation propagation in the near infrared, Opt. Commun.,

1999, 168(1-4), p. 237-250.

240. Parrott, J. E. and Stuckes, A. D., Thermal Conductivity of Solids. 1975, London: Pion

Limited. 157.

241. Gustafsson, S.E., Karawacki, E, and Khan, M.Nazim, Transient hot-strip method for

simultaneously measuring thermal conductivity and thermal diffusivity of solids and fluids,

Journal of Physics D, 1979, 12, p. 1411-1421.

References

181

242. Pettersson, S., Theoretical considerations on the hot-wire method applied to solids under

pressure, Review of Scientific Instruments, 1990, 61(4), p. 1308-1313.

243. Log, T. and Gustafsson, S.E., Transient Plane Source (TPS) Technique for Measuring

Thermal Transport Properties of Building Materials, Fire and Materials, 1995, 19, p. 43-49.

244. Karawacki, E and Suleiman, M., Dynamic plane source technique for simultaneous

determination of specific heat, thermal conductivity and thermal diffusivty of metallic

samples, Measurement Science & Technology, 1991, 2, p. 744-750.

245. Gustafsson, S.E. and Khan, M.N., Determination of the thermal-conductivity tensor and the

heat capacity of insulating solids with the transient hot-strip method, Journal of Applied

Physics, 1981, 52(4), p. 2596-2600.

246. Gustafsson, S.E. and Karawacki, E, Transient hot-strip probe for measuring thermal

properties of insulating solids and liquids, Review of Scientific Instruments, 1983, 54(6), p.

744-747.

247. Gustafsson, S.E., Karawacki, E., and Chohan, M.A., Thermal transport studies of

electrically conducting materials using the transient hot-strip technique, Journal of Physics

D, 1986, 19, p. 727-735.

248. Gustafsson, S.E., Transient Plane Source Techniques for Thermal Conductivity and

Thermal Diffusivity Measurements of Solid Materials, Review of Scientific Instruments,

1990, 62(3), p. 797-804.

249. Gustafsson, M., Karawacki, E., and Gustafsson, S.E., Thermal conductivity, thermal

diffusivity, and specific heat of t in samples from transient measurements with hot disk

sensors, Review of Scientific Instruments, 1994, 65, p. 3856-3859.

250. Parker, W.J., Jenkins, R.J., Butler, C.P., and Abbott, G.L., Flash Method of Determining

Thermal Diffusivity, Heat Capacity and Thermal Conductivity, Journal of Applied Physics,

1961, 32, p. 1679-1684.

251. Cape, J.A. and Lehman, G.W., Tempeature and Finite Pulse-Time Effects in the Flash

Method for Measuring Thermal Diffusivity, Journal of Applied Physics, 1963, 34(7), p.

1909-1913.

252. Clark, L.M. and Taylor, R.E., Finite Pulse Time Effects in the Flash Diffusivity Method,

Journal of Applied Physics, 1975, 46, p. 714-719.

253. Mardolcar, U.V. and Nieto de Castro, C.A., The Measurement of Thermal Conductivity at

High Temperatures, High Temperature - High Pressures, 1992, 24, p. 551-580.

254. Azzopardi, A., Mevrel, R., Saint-Ramond, B., Olson, E., and Stiller, K., Influence of aging

on structure and thermal conductivity of Y- PSZ and Y-FSZ EB-PVD coatings, Surface &


255. Raghavan, S. , Wang, H. , Dinwiddie, R. B. , Porter, W. D. , and Mayo, M. J., The Effect of

Grain Size, Porosity and Yttria Content on the Thermal Conductivity of Nanocrystalline

Zirconia, Scripta Materialia, 1998, 39(8), p. 1119-1125.

256. Alperine, S., Derrien, M., Jaslier, Y., and Mevrel, R., thermal Barrier Coatings: the

Thermal Conductivity challenge, in AGARD-R-823, 1997, NASA: Aalborg, Denmark, p. 1.

257. Zhu, D. M., Miller, R. A., Nagaraj, B. A., and Bruce, R. W., Thermal conductivity of EB-

PVD thermal barrier coatings evaluated by a steady-state laser heat flow technique, Surface

& Coatings Technology, 2001, 138(1), p. 1-8.

References

182

258. Slifka, A. J. and Filla, B. J., Thermal Conductivity Measurement of an Electron-Beam

Physical-Vapor-Deposition Coating, Journal of Research of the National Institute of

Standards and Technology, 2003, 108(2), p. 147-150.

259. Trice, R W, Su, Y J, Mawdsley, J R, Faber, K T, De Arellano-Lopez, A R, Wang, H, and

Porter, W D, Effect of Heat Treatment on Phase Stability, Microstructure, and Thermal

Conductivity of Plasma-sprayed YSZ, Journal of Materials Science, 2002, 37(11), p. 2359-

2365.

260. Zhu, D. and Miller, R.A., Thermal-Barrier Coatings for Advanced Gas-Turbine Engines,

MRS Bulletin, 2000, 25(7), p. 43-47.

261. Zhu, D., Miller, R.A., Nagaraj, B.A., and Bruce, R.W., Thermal Conductivity of EB-PVD

Thermal Barrier Coatings Evaluated by a Steady-State Laser Heat Flux Technique, in NASA

TM-2000-210238, 2000, NASA Lewis Research Center: Cleveland, Ohio.

262. Dinwiddie, R.B., Beecher, S.C., Porter, W.D., and Nagaraj, B.A. The Effect of Thermal

Aging on the Thermal Conductivity of Plasma Sprayed and EB-PVD Thermal Barrier

Coatings. in International gas turbine and aeroengine congress & exhibition. 1996,

Birmingham, UK: ASME, p. 96-GT-282.

263. Slifka, A.J., Filla, B.J., Phelps, J.M., Bancke, G., and Berndt, C.C., Thermal Conductivity of

a Zirconia Thermal Barrier Coating, Journal of Thermal Spray Technology, 1998, 7(1), p.

43-46.

264. Dutton, R., Wheeler, R., Ravichandran, K.S., and An, K., Effect of Heat Treatment on the

Thermal Conductivity of Plasma-Sprayed Thermal Barrier Coatings, Journal of Thermal

Spray Technology, 2000, 9(2), p. 204-209.

265. Taylor, R.E., Thermal Conductivity Determinations of Thermal Barrier Coatings, Materials

Science and Engineering, 1998, 245, p. 160-167.

266. Lawson, K.L, Nicholls, J.R., and Rickerby, D.S., Thermal conductivity and Ceramic

Microstructure, in High Temperature Surface Engineering, edited by Nicholls, J.R. and

Rickerby, D.S., 1997, Edinburgh:The Institute of Materials: London, UK. p. 8.

267. Gu, S., Lu, T.J., Hass, D.D., and Wadley, H.N.G., Thermal Conductivity of Zirconia

Coatings with Zig-Zag Pore Microstructures, Acta Materialia, 2001, 49, p. 2539-2547.

268. Nicholls, J. R., Lawson, K. J., Johnstone, A., and Rickerby, D. S., Methods to Reduce the

Thermal Conductivity of EB-PVD TBCs, Surface & Coatings Technology, 2002, 151, p.

383-391.

269. Singh, J., Wolfe, D.E., and Singh, J., Architecture of TBCs Produced by Electron Beam-

Physical Vapor Deposition (EB-PVD), Journal of Materials Science, 2002, 37(15), p. 3261-

3268.

270. Nicholls, J. R., Lawson, K. J., Johnstone, A., and Rickerby, D. S., Low Thermal

Conductivity EB-PVD Thermal Barrier Coatings, Materials Science Forum, 2001, 369-372,

p. 595-606.

271. Clarke, D.R., Materials Selection Guidelines for Low Thermal Conductivity Barrier

Coatings, Surface & Coatings Technology, 2003, 163-164, p. 67-74.

272. Raghavan, S., Wang, H., Dinwiddie, R. B., Porter, W. D., Vassen, R., Stover, D., and Mayo,

M. J., Ta2O5/Nb2O5 and Y2O3 co-doped zirconias for thermal barrier coatings, Journal of

the American Ceramic Society, 2004, 87(3), p. 431-437.

References

183

273. Schulz, U., Saruhan, B., Fritscher, K., and Leyens, C., Review on Advanced EB-PVD

Ceramic Topcoats for TBC Applications, International Journal of Applied Ceramic

Technology, 2004, 1(4), p. 302-315.

274. Matsumoto, M., Aoyama, K., Matsubara, H., Takayama, K, Banno, T., Kagiya Y., and

Sugita, Y., Thermal conductivity and oahse stability of plasma sprayed ZrO2-Y2O3-La2O3

coatings, Surface & Coatings Technology, 2005, 194, p. 31-35.

275. Raghavan, S., Wang, H., Porter, W. D., Dinwiddie, R. B., and Mayo, M. J., Thermal

properties of zirconia co-doped with trivalent and pentavalent oxides, Acta Materialia, 2001,

49(1), p. 169-179.

276. Zhu, D. and Miller, D A, Development of Advanced Low Thermal Conductivity Thermal

Barrier Coatings, in NASA/TM-2004-212961, 2004, Glenn Research Center: Cleveland,

Ohio.

277. Zhu, D. and Miller, R.A., Thermal Conductivity of Advanced Ceramic Thermal Barrier

Coatings Determined by a Steady-State Laser Heat-Flux Approach, in NASA/TM-2004-

213040, 2004, Glenn Research Center: Cleveland Ohio.

278. Singh, J., Wolfe, D.E., Miller, D.A., Eldridge, J.I., and Zhu, D., Tailored microstructure of

zirconia and hafnia-based thermal barrier coatings with low thermal conductivity and high

hemispherical reflectance by EB-PVD, Journal of Material Science, 2004, 36(6), p. 1975-

1985.

279. Lehmann, H., Pitzer, D., Pracht, G., Vassen, R., and Stover, D., Thermal conductivity and

thermal expansion coefficients of the lanthanum rare-earth-element zirconate system,


280. Vassen, R., Traeger, F., and Stover, D., New Thermal Barrier Coatings Based on

Pyrochlore/ YSZ Double-Layer Systems, International Journal of Applied Ceramic

Technology, 2004, 1(4).

281. Clyne, T. W. and Withers, P. J., An Introduction to Metal Matrix Composites. Cambridge

Solid State Science Series, ed. Davis, E. and Ward, I. 1993, Cambridge: Cambridge

University Press.

282. Sevostianov, I. and Kachanov, M., Anisotropic thermal conductivities of plasma-sprayed

thermal barrier coatings in relation to the microstructure, Journal of Thermal Spray

Technology, 2000, 9(4), p. 478-482.

283. Shafiro, B. and Kachanov, M., Anisotropic effective conductivity of materials with

nonrandomly oriented inclusions of diverse ellipsoidal shapes, Journal of Applied Physics,

2000, 87(12), p. 8561-8569.

284. Tzou, D. Y., The Effect of Internal Heat-Transfer in Cavities on the Overall Thermal-

Conductivity, International Journal of Heat and Mass Transfer, 1991, 34(7), p. 1839-1846.

285. Hashin, Z. and Shtrikman, S., A Variational Approach to the Theory of the Elastic

Behaviour of Multiphase Materials, Journal of Mechanics and Physics of Solids, 1963, 11, p.

127-140.

286. Bauer, T. H., A General Analytical Approach Toward the Thermal Conductivity of Porous

Media, International Journal of Heat and Mass Transfer, 1993, 36(17), p. 4181-4191.

287. Bristow, J.R., Microcracks, and the static and dynamic elastic constants of annealed and

heavily cold-worked metals, British Journal of Applied Physics, 1960, 11(2), p. 81-85.

References

184

288. McPherson, R., A Model for the Thermal Conductivity of Plasma-Sprayed Ceramic

Coatings, Thin Solid Films, 1984, 112, p. 89-95.

289. Li, C. J. and Ohmori, A., Relationships between the microstructure and properties of

thermally sprayed deposits, Journal of Thermal Spray Technology, 2002, 11(3), p. 365-374.

290. Boirelavigne, S., Moreau, C., and Saintjacques, R. G., The Relationship between the

Microstructure and Thermal- Diffusivity of Plasma-Sprayed Tungsten Coatings, Journal of

Thermal Spray Technology, 1995, 4(3), p. 261-267.

291. Lu, T. J. and Hutchinson, J. W., Thermal-Conductivity and Expansion of Cross-Ply

Composites with Matrix Cracks, Journal of the Mechanics and Physics of Solids, 1995,

43(8), p. 1175-1198.

292. Lu, T. J., Levi, C. G., Wadley, H. N. G., and Evans, A. G., Distributed Porosity as a Control

Parameter for Oxide Thermal Barriers made by Physical Vapor Deposition, Journal of the

American Ceramic Society, 2001, 84, p. 2937-2946.

293. Golosnoy, I.O., Tsipas, S.A., and Clyne, T.W., An Analytical Model for Simulation of Heat

Flow in Plasma Sprayed Thermal Barrier Coatings, Journal of Thermal Spray Technology,

2005, 14(2), p. in press.

294. Ravichandran, K.S., An, K., Dutton, R.E., and Semiatin, S.L., Thermal Conductivity of

Plasma-Sprayed Monolithic and Multilayer Coatings of Alumina and Yttria-Stabilized

Zirconia, Journal of the American Ceramic Society, 1999, 83(3), p. 673-682.

295. Lu, T. J., Levi, C. G., Wadley, H. N. G., and Evans, A. G., Distributed porosity as a control

parameter for oxide thermal barriers made by physical vapor deposition, Journal of the

American Ceramic Society, 2001, 84(12), p. 2937-2946.

296. Incropera, F.P. and Dewitt, D.P., Introduction to Heat Transfer. 3rd ed. 1996, NY, USA:

John Wiley & Sons, Inc.

297. Makino, T., Kunitomo, T., Sakai, I., and Kinoshita, H., Thermal Radiation Properties of

Ceramic Materials, Heat Transfer. Japanese Research., 1984, 13(4), p. 33-50.

298. Ferriere, A., Lestrade, L., and Robert, J. F., Optical properties of plasma-sprayed ZrO2-Y2O3

at high temperature for solar applications, Journal of Solar Energy Engineering-

Transactions of the ASME, 2000, 122(1), p. 9-13.

299. Eldridge, J.I., Spuckler, C.M., Street, K.W., and Markham, J.R. Infrared Radiative

Properties of Yttria-Stabilized Zirconia Thermal Barrier Coatings. in Ceramic and

Engineering Science Procedings. 2002, Cocoa Beach, Florida, USA: The American Ceramic

Society, p. 417-430.

300. Samarskii, A.A. and Vabishevich, P.N., Computational heat transfer. V.1. Mathematical

Modelling. 1995, Chichester: Wiley. 406.

301. Tannehill, J.C., Anderson, D.A., and Pletcher, R.H., Computational Fluid Mechanics and

Heat Transfer. 2nd ed. 1997, Washington, London: Taylor & Francis. 792.

302. Ilavsky, J., Allen, A.J., Long, G.G., Krueger, S., Berndt, C.C., and Herman, H., Influence of

Spray Angle on the Pore and Crack Microstructure of Plasma-Sprayed Deposits, Journal of

the American Ceramic Society, 1997, 80(3), p. 733-742.

303. Trapaga, G., Matthys, E.F., Valencia, J.J., and Szekely, J., Fluid Flow, Heat Transfer and

Solidification of Molten Metal Droplets Impinging on Substrates; Comparison of Numerical

and Experimental Results, Metallurgical Transactions, 1992, 23B, p. 701-718.

References

185

304. Zhang, H., Theoretical analysis of Spreading and Solidification of Molten Droplet during

Thermal Spray Deposition, int. J. Heat Mass Transfer, 1999, 42, p. 2499-2508.

305. Bennett, T. and Poulikakos, D., Splat-quench Solidification: Estimating the Maximum

Spreading of a Droplet impacting a Solid Surface, Journal of Materials Science, 1993, 28, p.

963-970.

306. Scott, H.G., Phase Relationships in the ZrO2-Y2O3 System, Journal of Materials Science,

1975, 10, p. 1527-1535.

307. Tani, E., Yoshimure, M., and Somiya, S., Revised phase-diagram of the system ZrO2-CeO2

below 1400oC, Journal of the American Ceramic Society, 1983, 66(7), p. 506-510.

308. Din, S. U. and Kaleem, A., Vickers hardness study of zirconia partially stabilized with

lanthanide group oxides, Materials Chemistry and Physics, 1998, 53(1), p. 48-54.

309. Thompson, J.A., Thermomechanical Behaviour of Plasma-Sprayed Thermal Barrier

Coatings, in Department of Materials Science and Metallurgy, 2000, University of

Cambridge: Cambridge, UK.

310. Toraya, H., Yoshimura, M., and Somiya, S., Calibration Curve for Quantitative Analysis of

the Monoclinic-Tetragonal ZrO2 System by X-ray Diffraction, Journal of the American

Ceramic Society, 1984, p. C119-121.

311. Toraya, H., Yoshimura, M., and Somiya, S., Quantitative Analysis of Monoclinic -

Stabilized Cubic ZrO2 Systems by X-ray Diffraction, Journal of the American Ceramic

Society, 1984, 67, p. C183-184.

312. Wiles, D.B. and Young, R.A., A new computer program for Rietveld analysis of X-ray

powder diffraction Patterns, Journal of Applied Crystallography, 1981, 14, p. 149-151.

313. Murphy, A.M., Clustering in Particulate Metal Matrix Composites, PhD thesis, in Materials

Science and Metallurgy, 1997, University of Cambridge: Cambridge, UK.

314. Chudoba, T. and Richter, F., Investigation of Creep Behaviour Under Load During

Indentation Experiments and its Influence on Hardness and Modulus Results, Surface &


315. Oliver, W C and Pharr, G M, An Improved Technique for Determining Hardness and Elastic

Modulus Using Load and Displacement Sensing Indentation Experiments, Journal of


316. Nimonic Alloy 80A Material Datasheet: Physical and Mechanical Properties, Special

Metals Corporation: Hereford, UK.

317. Heraeus Quarzglas - Quartz Glass for Optics: Data and Properties, 2004, Heraeus:

Germany.

318. Incropera, F.P. and Dewitz, D.P., Fundamentals of Heat and Mass Transfer. Fifth ed. 2002:

John Wiley & Sons.

319. Miller, R., Garlick, R., and Smialek, J., Phase Distributions in Plasma-Sprayed Zirconia-

Yttria, Ceramic Bulletin, 1983, 62, p. 1355-1358.

320. Heuer, A H. and Ruhle, M. Phase transformations in ZrO2-containing ceramics: I. the

instability of c-ZrO2 and the resulting diffusion-controlled reactions. in Advances in

ceramics: Science and technology of zirconia II. 1983, edited by Claussen N., Ruhle, M.,

and Heuer, A.H., Stuttgard, Germany: American Ceramic Society, Inc., p. 1-13.

References

186

321. Heuer, A.H., Chaim, R., and Lanteri, V., The Displacive Cubic Tetragonal Transformation

in ZrO2 Alloys, Acta Metallurgica, 1987, 35, p. 661-666.

322. Lanteri, V., Heuer, A.H., and Mitchell, T.E. Tetragonal Phase in the System ZrO2-Y2O3. in

Second International Conference on the Science and Technology of Zirconia. 1983, edited

by Claussen, N., Ruhle, M., and Heuer, A.H., Stuttgart: American Ceramic Society, p. 118-

131.

323. Lakiza, S.M. and Lopato, L.M., Stable and metastable phase relations the system alumina-

zirconia-yttria, Journal of the American Ceramic Society, 1997, 80(4), p. 893-902.

324. Stough, M.A., Solid solubility and precipitation in a single-crystal alumina-zirconia system,


325. Li, M.J., Feng, Z.C., Ying, P.L., Xin, Q., and Li, C., Phase transformation in the surface

region of zirconia and doped zirconia detected by UV Raman spectroscopy, Physical

Chemistry Chemical Physics, 2003, 5(23), p. 5326-5332.

326. Shannon, R.D., Revised effective ionic radii and systematic studies of interatomic distances

in halides and chalcogenides, Acta Crystallographica, 1976, A32, p. 751-767.

327. Langjahr, P.A., Oberacker, R., and Hoffmann, M.J., Long-term behaviour and application

limits of plasma sprayed zirconia thermal barrier coatings, Journal of the American


328. Brandon, J. and R.Taylor, Phase Stability of Zirconia-Based Thermal Barrier Coatings Part

I. Zirconia-Yttria alloys, Surface & Coatings Technology, 1991, 46, p. 75-90.

329. Michel, D., Perezyjo.M, and Collongu.R, Study of Order-Disorder Transformation of

Fluorite Structure with Pyrochlore Type Phases of Formula (1-X) ZrO2-XLn2O3, Materials

Research Bulletin, 1974, 9(11), p. 1457-1468.

330. Jang J.W., Kim, D.J., and Lee, D.Y., Size effect of trivalent oxides on low temperature phase

stability of 2Y-TZP, Journal of Materials Science, 2001, 36(22), p. 5391-5395.

331. Vassen, R., Dietrich, M., Lehmann, H., Cao, X., Pracht, G., Tietz, F., Pitzer, D., and Stover,

D., Development of oxide ceramics for an application as TBC, Materialwissenschaft Und

Werkstofftechnik, 2001, 32(8), p. 673-677.

332. Cao, X. Q., Vassen, R., Jungen, W., Schwartz, S., Tietz, F., and Stover, D., Thermal

stability of lanthanum zirconate plasma-sprayed coating, Journal of the American Ceramic

Society, 2001, 84(9), p. 2086-2090.

333. Anthony, A.M. Structure of point defects in ionic materials. in Sintering and Related

Phenomena. 1972, edited by Kuczynski, G.C., Indiana: Plenum press, New York, London.

334. Kroger F.A., The Chemistry of Imperfect Crystals. 1964, Amsterdam, New York: North

Holland Publ. Co. and Wiley.

335. Tilley, R.J.D., Defect Crystal Chemistry and its Applications. 1987: Blackie & Son Ltd.

336. Borg, R.J. and Dienes, G.J., An Introduction to Solid State Diffusion. 1988: Academic Press,

Inc.

337. Weber, S., Scherrer, S., Scherrer, H., Kilo, M., Taylor, M.A., and Borchardt, G., SIMS

analysis of multi-diffusion profiles of lanthanides in stabilzed zirconias, Applied Surface

Science, 2003, 203-204, p. 656-659.

References

187

338. Chaim, R., Basat, G., and Kats-Demyanets, A., Effect of oxide additives on grain growth

during sintering of nanocrystalline zirconia alloys, Materials Letters, 1998, 35(3-4), p. 245-

250.

339. Li, P., Chen, I-W., and Penner-Hahn, J.E., X-ray-absorption studies of zirconia polymorphs.

II. Effect of Y2O3 dopant on ZrO2 structure, Physical Review B, 1993, 48(14), p. 10074-

10081.

340. Li, P., Chen, I-W., and Penner-Hahn, J.E., X-ray-absorption studies of zirconia polymorphs.

I. Characteristic local structures, Physical Review B, 1993, 48(14), p. 10063-10073.

341. Kamigaito, O., Idea strength of brittle materials as a function of their melting or

sublimation points, Journal of Materials Science Letters, 1991, 10, p. 567-568.

342. Tekeli, S., Erdogan, M., and Aktas, B., Influence of alpha-Al2O3 addition on sintering and

grain growth behaviour Of 8 mol% Y2O3-stabilised cubic zirconia (c-ZrO2), Ceramics

International, 2004, 30(8), p. 2203-2209.

343. Matsumoto, M., Yamaguchi, N., and Matsubara, H., Low thermal conductivity and high

temperature stability of ZrO2-Y2O3-La2O3 coatings produced by electron beam PVD, Scripta

Materialia, 2004, 50(6), p. 867-871.

344. Malzbender, J. and Steinbrech, R.W., Mechanical methods to determine layer compliances

within multilayered composites, Journal of Materials Research, 2003, 18(6), p. 1374-1382.

345. Bengtsson, P., Ericsson, T., and Wigren, J., Thermal Shock Testing of Burner Cans Coated

with a Thick Thermal Barrier Coating, Journal of Thermal Spray Technology, 1998, 7(3), p.

340-348.

346. Guo, H.B., Vassen, R., and Stover, D., Thermophysical properties and thermal cycling

behaviour of plasma sprayed thick thermal barrier coatings, Surface & Coatings

Technology, 2005, 192, p. 48-56.

347. Hutchinson, J.W. and Suo, Z., Mixed Mode Cracking in Layered Materials, Advanced

Applied Mechanics., 1991, 29, p. 63-191.

348. Bengtsson, P. and Johannesson, T., Characterization of Microstructural Defects in Plasma-

Sprayed Thermal Barrier Coatings, Journal of Thermal Spray Technology, 1995, 4(3), p.

245-251.

349. Sampath, S., Jiang, X.Y., Matejicek, J., Leger, A.C., and Vardelle, A., Substrate

temparature effects on splat formation, microsture devlopment and properties of plasma

sprayed coatings Part I: Case study for partially stabilized zirconia, Materials Science and

Engineering, 1999, A272, p. 181-188.

350. Sampath, S. and Jiang, X.Y., Spalta formation and microstructure developemtn during

plasma spraying: deposition temperature effects, Materials Science and Engineering, 2001,

A304-306, p. 144-150.

351. Gill, S.C. and Clyne, T.W., Stress Distributions and Material Response in Thermal Spraying

of Metallic and Ceramic Deposits, Met. Trans., 1990, 21B, p. 377-385.

352. Clyne, T.W. and Gill, S.C., Heat Flow and Thermal Contraction during Plasma Spray

Deposition, in Heat Transfer in Manufacturing and Processing of New Materials, edited by

Tanasawa, I., 1991, Hemisphere: New York. p. 33-48.

353. Tsui, Y.C., Thompson, J.A., and Clyne, T.W. The Effect of Bond Coat Creep on Residual

Stresses and Debonding in Plasma Sprayed Thermal Barrier Systems. in Thermal Spray:

References

188

Meeting the Challenges of the 21st Century. Proceedings of the 15th Int. Thermal Spray

Conf. 1998, edited by Coddet, C., Nice, France: ASM International, p. 1565-1570.

354. Teixeira, V., Andritschky, M., Fischer, W., Buchkremer, H. P., and Stover, D., Effects of

deposition temperature and thermal cycling on residual stress state in zirconia-based

thermal barrier coatings, Surface & Coatings Technology, 1999, 121, p. 103-111.

355. Takeuchi, S., Ito, M., and Takeda, K., Modelling of Residual Stress in Plasma-Sprayed

Coatings: Effect of Substrate Temperature, Surface & Coatings Technology, 1990, 43/44, p.

426-435.

356. Stover, D., Pracht, G., Lehmann, H., Dietrich, M., Doring, J. E., and Vassen, R., New

Material Concepts for the Next Generation of Plasma Sprayed Thermal Barrier Coatings,


357. Miller, R.A., Current Status of Thermal Barrier Coatings - An Overview, Surface &


358. Wang, H. and Dinwiddie, R.B., Reliability of the Laser Flash Thermal Diffusivity

Measurements of the Thermal Barrier Coatings, Journal of Thermal Spray Technology,

2000, 9(2), p. 210-214.

359. Slifka, A. J., Thermal Conductivity Apparatus for Steady-State, Comparative Measurement

of Ceramic Coatings, Journal of Research of the National Institute of Standards and

Technology, 2000, 105(4), p. 591-605.

360. Taylor, R., Construction of Apparatus for Heat Pulse Thermal Diffusivity Measurements

from 300-3000K, Journal of Physics E: Scienctific Instruments, 1980, 13, p. 1193-1196.

361. Mirmira, S.R. and Fletcher, L. S., Review of the Thermal Conductivity of Thin Films,

Journal of Thermophysics and Heat Transfer, 1998, 12(2), p. 121-131.

362. Thompson, J.A., Ji, W., and Clyne, T.W., The Effect of Top Coat Sintering on Ceramic

Spallation in Plasma-Sprayed Thermal Barrier Coatings, in 13th Int. Conf. on Surface

Modific. Techn. (SMT XIII), edited by Sudarshan, T.S., Khor, K.A., and Jeandin, M., 1999,

ASM. p. 177-182.

363. Mogro-Campero, A., Johnson, C. A., Bednarczyk, P. J., Dinwiddie, R. B., and Wang, H.,

Effect of Gas Pressure on Thermal Conductivity of Zirconia Thermal Barrier Coatings,

Surface & Coatings Technology, 1997, 94-5(1-3), p. 102-105.

364. Zhu, D., Bansal, N.P., Lee, K.N., and Miller, R.A., Thermal Conductivity of Ceramic

Thermal Barrier and Environmental Barrier Coating Materials, in NASA TM-2001-211122,

2001, NASA Lewis Research Center: Cleveland, Ohio.

365. Pawlowski, L., Lombard, D., and Fauchais, P., Structure-Thermal Properties - Relationship

in Plasma Sprayed Zirconia Coatings, Journal of Vacuum Science & Technology a-Vacuum

Surfaces and Films, 1985, 3(6), p. 2494-2500.

366 Cardarelli, F., Materials Handbook: a concise desktop reference. 2000, London: Springer-

Verlag.

Documents

Bond Coat SAT PhD Thesis