study of bond coats and failure mechanisms of thermal

STUDY OF BOND COATS AND FAILURE

MECHANISMS OF THERMAL BARRIER

COATINGS

A thesis submitted to the University of Manchester for the degree of

Doctor of Philosophy

in the Faculty of Science and Engineering

2019

Chen LIU School of Materials

LIST OF CONTENTS

1

List of Contents

List of Contents .................................................................................................................... 1

List of Figures ....................................................................................................................... 5

List of Tables .......................................................................................................................11

List of Abbreviations.............................................................................................................12

List of Publication ................................................................................................................13

Abstract .............................................................................................................................14

Declaration .........................................................................................................................15

Copyright Statement ............................................................................................................16

Acknowledgements..............................................................................................................17

Chapter 1 Introduction .........................................................................................................18

1.1 Gas-turbine engine materials ........................................................................................18

1.2 Introduction of thermal barrier coatings .........................................................................19

1.3 Objectives and structure of the dissertation ....................................................................22

Chapter 2 Literature Review ..................................................................................................25

2.1 Thermal barrier coating system .....................................................................................25

2.2 Top coat ....................................................................................................................25

2.2.1 Material requirement and selection.......................................................................25

2.2.2 Deposition techniques and microstructure .............................................................29

2.2.3 New top coat candidates .......................................................................................31

2.3 Bond coat ..................................................................................................................32

2.3.1 Material requirements .........................................................................................32

2.3.2 Bond coat categories ............................................................................................33

2.3.2.1 β-NiAl based diffusion bond coat .......................................................................33

2.3.2.2 Pt-diffused γ-Ni/γ’-Ni3Al bond coat ....................................................................38

2.3.2.3 MCrAlY overlay bond coat ................................................................................41

2.4 Thermally grown oxide.................................................................................................43

2.4.1 Material requirements and selection .....................................................................43

2.4.2 TGO microstructure and stress ............................................................................44

2.4.3 TGO transformation during early stage of oxidation ..............................................46

2.5 Superalloy substrate ....................................................................................................50

LIST OF CONTENTS

2

2.5.1 Composition and microstructure ..........................................................................51

2.5.2 Physical and mechanical properties ......................................................................52

2.6 The degradation and failure of TBCs...............................................................................53

2.6.1 General failure modes..........................................................................................55

2.6.2 The failure mechanisms of APS TBCs ...................................................................58

2.6.3 The failure mechanisms of EB-PVD TBCs.............................................................60

2.6.4 Interface toughness measurement of TBCs ............................................................64

2.6.4.1 Definitions of interface adhesion & delamination and interface toughness ..............64

2.6.4.2 Interface toughness test methods .....................................................................67

2.7 Summary ...................................................................................................................70

Chapter 3 Pt Effect on Early Stage Oxidation Behaviour of Pt-diffused γ-Ni/γ’-Ni3Al Coatings ..........71

3.1 Introduction ...............................................................................................................71

3.2 Experimental procedures .............................................................................................73

3.2.1 Sample preparation and thermal treatment ...........................................................73

3.2.2 Luminescence measurement and data processing ...................................................75

3.2.3 ASTAR automated crystal orientation mapping on TEM .......................................76

3.2.4 Other characterization methods ...........................................................................77

3.3 Results ......................................................................................................................78

3.3.1 Microstructure of the as-fabricated coatings with different Pt contents ...................78

3.3.2 Transient alumina to stable α-Al2O3 transformation...............................................79

3.3.3 TGO composition and microstructure evolution ....................................................82

3.3.4 TGO growth rate & stress evolution .....................................................................84

3.3.6 PLPS studies on Ni-Al-Pt alloy samples .................................................................89

3.4 Discussion ..................................................................................................................94

3.4.1 Pt effect on the θ-Al2O3 to α-Al2O3 transformation .................................................94

3.4.2 Pt effect on TGO composition & stress ..................................................................98

3.4.3 Early stage oxidation effect on prolonged oxidation performance ............................99

3.5 Summary ................................................................................................................. 100

Appendix A. Coating/alumina orientation analysis............................................................... 100

Appendix B. Prolonged oxidation lifetime of coatings with different Pt additions ..................... 104

Appendix C. Early stage oxidation effect on the stable scale morphology & stress .................... 105

Chapter 4 Effect of Superalloy Substrate on the Lifetime and Interfacial Toughness of Electron Beam

Physical Vapour Deposited Thermal Barrier Coatings .............................................................. 107

4.1 Introduction ............................................................................................................. 107

LIST OF CONTENTS

3

4.2 Experimental procedures ........................................................................................... 109

4.2.1 Sample preparation ........................................................................................... 109

4.2.2 Thermal treatment ............................................................................................ 110

4.2.3 Microstructure characterization ......................................................................... 110

4.2.4 Interface toughness measurement by the strain-to-fail test.................................... 111

4.2.4.1 Theoretical background ................................................................................. 111

4.2.4.2 Strain-to-fail test coupled with 3D-DIC ............................................................. 112

4.2.4.3 Determination of the coating stress ................................................................. 113

4.2.4.4 Determination of the YSZ modulus .................................................................. 114

4.2.4.5 Measuring buckling radius by 3D-DIC ............................................................... 114

4.3 Results .................................................................................................................... 115

4.3.1 Microstructure of the as-received TBCs .............................................................. 115

4.3.2 Cyclic oxidation testing ...................................................................................... 116

4.3.2.1 YSZ lifetime .................................................................................................. 116

4.3.2.2 Microstructural evolution ............................................................................... 117

4.3.3 Strain-to-fail compression test coupled with 3D-DIC ............................................ 123

4.4 Discussion ................................................................................................................ 127

4.4.1 Estimation of the interfacial toughness for TBCs on N5 and X4 substrates............. 127

4.4.2 Interface degradation of TBCs on different substrates.......................................... 129

4.5 Summary ................................................................................................................. 132

Chapter 5 The Al-enriched γ’-Ni3Al-base bond coat for thermal barrier coating applications ......... 133

5.1 Introduction ............................................................................................................. 133

5.2 Experimental procedures ........................................................................................... 136

5.2.1 Sample preparation ........................................................................................... 136

5.2.1.1 Fabrication of Pt-diffused intermediate coatings................................................ 136

5.2.1.2 Fabrication of Al-enriched γ’-phase coatings by pack cementation ....................... 138

5.2.2 Thermal treatment ............................................................................................ 140

5.2.3 Characterization methods .................................................................................. 140

5.3 Results .................................................................................................................... 141

5.3.1 Synthesis of Al-enriched γ’-phase coatings by pack cementation ........................... 141

5.3.2 Microstructure of the as-received coatings........................................................... 144

5.3.3 Isothermal oxidation performance of three Pt-diffused coatings ............................ 147

5.3.3.1 Elemental diffusion of three coatings ............................................................... 147

5.3.3.2 Oxide microstructure and growth kinetics......................................................... 150

LIST OF CONTENTS

4

5.3.3.3 TGO spallation .............................................................................................. 155

5.3.4 Rumpling behaviour of three bond coats under cyclic oxidation ............................ 156

5.4 Discussion ................................................................................................................ 161

5.4.1 Pt and Al depletion of three coatings ................................................................... 161

5.4.2 Effect of bond coat composition on the selective oxidation of aluminium................ 164

5.4.3 Rumpling behaviour of three coatings ................................................................. 165

5.4.3.1 Balint and Hutchinson rumpling model ............................................................. 165

5.4.3.2 B&H model applied to the Al-enriched γ’-phase coating...................................... 166

5.5 Summary ................................................................................................................. 167

Chapter 6 Conclusions and Future Work................................................................................ 169

6.1 Conclusions .............................................................................................................. 169

6.2 Future work ............................................................................................................. 171

Reference ......................................................................................................................... 173

Word counts: 43169

LIST OF FIGURES

5

List of Figures

Fig. 1.1 Progress in the maximum allowable temperatures of Ni-base superalloys and thermal

barrier coating (TBC) since 1965. The red line indicate the sharp increase of the allowable gas

temperature by the employment of TBCs [2]. .....................................................................19

Fig. 1.2 Schematic view of a TBC system on an airfoil [4]. .................................................20

Fig. 2.1 Schematic illustration of multi-layered and multifunctional TBC system. The functions

and properties for each layer are indicated [2]. ....................................................................26

Fig. 2.2 Coefficients of thermal expansion (CTEs) of a range of materials are cross-plotted

against their thermal conductivities [3]. ..............................................................................28

Fig. 2.3 Cross-sectional microstructure of a) APS and b) EBPVD TBC [24]. .......................30

Fig. 2.4 Binary phase diagram of Ni-Al system [39]. ..........................................................34

Fig. 2.5 The cross-sectional SEM images of the β-NiAl bond coats fabricated by a) low-activity

and high-temperature; b) high-activity and low-temperature pack cementation [41]. ............36

Fig. 2.6 The cross-sectional SEM image of as-fabricated (by pack cementation) β-NiPtAl bond

coat on the CMSX-4 superalloy substrate. ..........................................................................37

Fig. 2.7 The typical microstructure of a Pt-diffused γ-Ni/γ’-Ni3Al bond coat on the CMSX-4

superalloy substrate. The γ’-phase: brighter contrast; γ-phase: grey contrast. .......................39

Fig. 2.8 Cross-sectional images of the as-deposited NiCoCrAlY bond coat deposited by HVOF

[71]: a) optical image and b) back scattered electron (BSE) image (high magnification) showing

the β (grey contrast) + γ (white contrast) two-phase microstructure. The black contrast areas

are interfacial pores between metal particles. ......................................................................42

Fig. 2.9 A fractured cross-sectional image of a specimen exposed for several hours at 1200℃

showing the TGO columnar and equiaxed grains [4]. ..........................................................44

Fig. 2.10 a) Schematic illustration of the PLPS technique and b) typical R1/R2 fluorescence

spectra for Cr-containing stress-free (dashed line) and stressed α-Al2O3 (solid line) [81]. .....46

Fig. 2.11 The morphology of oxide scale on the alloy after h oxidation at ℃. Note the needle

or plate shape of θ-Al2O3 [88]. ...........................................................................................50

Fig. 2.12 Alloying elements in the Ni-based superalloys (adapted from [101]). ....................51

LIST OF FIGURES

6

Fig. 2.13 Microstructure of a Ni-based single crystal superalloy revealing a high volume

fraction of γ’ phase [100]: the cuboidal γ’ precipitates (grey contrast) in the γ-matrix (white

contrast). ...........................................................................................................................52

Fig. 2.14 Five major categories of failure mechanisms documented for TBC systems [74]. ..55

Fig. 2.15 The optical images of a EBPVD TBC sample showing a) the incipient buckling of

the top coat (viewed under reflected light) and b) subsequent spallation of the top coat [3]. ..57

Fig. 2.16 A schematic illustration of four primary cracking modes in an APS TBC system [11].

.........................................................................................................................................59

Fig. 2.17 SEM micrographs showing damage evolution in an APS TBC: a) isothermal

oxidation and b) thermal cycling [60]. ................................................................................59

Fig. 2.18 The cross-sectional SEM micrograph of an EBPVD TBC on a β-NiPtAl bond coat

exhibiting the TGO rumpling after 50 1-h cycles at 1150 ℃ [117]. .....................................61

Fig. 2.19 NiPtAl specimens showing the effect of the oxide thickness on the rumpling. The

oxide layer thickness is a) ~ 5 μm and b) ~ 10 μm. The systems were subjected to the same

thermal cycling history and obviously more rumpling developed for the specimen with thicker

oxide layer [127]. ..............................................................................................................63

Fig. 2.20 Schematic of different failure modes for a thin coating system. a) A single through-

thickness crack in the coating which deflects to the interface to cause the coating failure; b)

multiple through-thickness cracks in the coating; c) edge-delamination at the interface and d)

buckling-induced delamination for a compressed film [133]. ...............................................66

Fig. 3.1 a) Luminescence spectrum showing characteristic peaks for θ-Al2O3 and α-Al2O3 of

an oxide scale; b) R peaks of an α-Al2O3 scale on a Pt-diffused γ/γ’ coating after isothermal

oxidation at 1100°C for 1 h. For comparison, the spectrum from a stress-free, polycrystalline

alumina is shown by the red line. .......................................................................................76

Fig. 3.2 Cross-sectional SEM images of as-diffused samples: a) 2 µm Pt coating; b) 5 µm Pt

coating. .............................................................................................................................79

Fig. 3.3 Typical luminescence spectrum of a) 0 Pt, b) 2 µm Pt and c) 5 µm Pt samples after 2

min oxidation at 1000 ºC....................................................................................................80

Fig. 3.4 Luminescence spectrum of a) 0 Pt, b) 2 µm Pt and c) 5 µm Pt samples after 10 min

oxidation at 1000 ºC. .........................................................................................................81

LIST OF FIGURES

7

Fig. 3.5 Fraction profiles of θ-Al2O3 as a function of oxidation time. ...................................81

Fig. 3.6 a) - c): FIB/SEM cross-sectional images after 2 min oxidation; d) - f): FIB/SEM cross-

sectional images after 10 min oxidation at 1000 ºC of no Pt sample, 2 µm Pt coating and 5 µm

Pt coating, respectively. .....................................................................................................82

Fig. 3.7 a) - c): FIB/SEM cross-sectional images; d) - f): SEM surface images of no Pt sample,

2 µm Pt coating and 5 µm Pt coating after 30 min oxidation at 1000 ºC. ..............................84

Fig. 3.8 Oxide scale thickness evolution of no Pt sample and 5 µm Pt sample as a function of

oxidation time. ..................................................................................................................85

Fig. 3.9 The TGO stress evolution of no Pt sample and 5 µm Pt sample as a function of

oxidation time. ..................................................................................................................86

Fig. 3.10 a) ADF STEM image of 2 µm Pt sample after oxidation at 1050 ºC for 10 min; b)

combined phase map and phase reliability map obtained from automated crystal orientation

mapping in TEM, taken from the red box region in a). Green: θ-Al2O3; red colour: α-Al2O3.

.........................................................................................................................................87


combined phase map and phase reliability taken from the red box region in a). Green colour:

θ-Al2O3; red: α-Al2O3. .......................................................................................................88

Fig. 3.12 Microstructure of the as-received a) Ni-20Al, b) Ni-20Al-10Pt, c) and d) Ni-20Al-

20Pt alloy. The inset in d) shows the magnified morphology of the tiny γ channels in the γ/γ’

region................................................................................................................................91

Fig. 3.13 a) and d): the optical image; b) c) e) and f): PLPS spectra taken from different phase

regions of Ni-20Al alloy after oxidation at 1050 ºC for 2 min and 10 min, respectively. .......92


regions of Ni-20Al-10Pt alloy after oxidation at 1050 ºC for 2 min and 10 min. ...................92


regions of Ni-20Al-20Pt alloy after oxidation at 1050 ºC for 2 min and 10 min. ...................93

Fig. 3.16 a) Previously reported growth model of α-Al2O3: uniformly nucleate along the

interface, resulting in a layered structure of the oxide scale; b) and c) new growth model in our

study which illustrates lower Pt content and higher Pt content coatings, respectively. ...........95

LIST OF FIGURES

8

Fig. 4.1 Schematic view of the experimental setup: both high speed cameras take images of the

TBC coating surface (x-y plane) during the compression test, and the compression load is along

the y-direction. ................................................................................................................ 113

Fig. 4.2 As-received TBCs with different substrates, a) - d): cross-sectional SEM images; e)

and f): Ni, Al and Pt concentration profiles by SEM/EDX linescan along the lines in c) and d).

....................................................................................................................................... 115

Fig. 4.3 YSZ lifetime for the cyclic oxidation testing of TBC-coated different superalloy blades.

....................................................................................................................................... 117

Fig. 4.4 Cross-sectional SEM images: a) b) secondary electron (SE) mode; c) d) backscattered

electron (BSE) mode; e) f) corresponding SEM/EDX linescan elemental concentration profile

along the red lines in c) and d), respectively after 5 1-h cycles at 1200 °C.......................... 118

Fig. 4.5 a) b): cross-sectional SEM images (SE mode) and c) d): Ni, Al and Pt concentration

profiles by SEM/EDX linescan along the red lines in a) and b), respectively after 10 1-h cycles

at 1200 °C. ...................................................................................................................... 119

Fig. 4.6 a) and b): bond coat surface BSE images exposed by spalling after 28 cycles at 1200 °C;

c) and d): BSE images of the back side of spalled YSZ coating after 28 cycles. ................. 121

Fig. 4.7 a) TGO thickness evolution during the cycling test; b) root mean square roughness

evolution of TGO/BC interface by processing of digitized profiles. ................................... 122

Fig. 4.8 a) Optical image taken by the camera for DIC showing the sample surface prior to

applying the load. The region-of-interest (ROI) is highlighted with the red rectangular and is

used as the reference image f (x); b) deformed images at several loading scales corrected by a

DIC displacement field g (x + u (x)) and c) the corresponding residual field Φ (x) of a sample

(as-received TBC with N5 substrate) during the test. The red ellipse in c) highlights the

occurrence of buckling. .................................................................................................... 124

Fig. 4.9 Evolution of a) average buckling radii, and b) corresponding strains calculated by DIC

as a function of oxidation time at 1150 °C for specimens with the X4 and N5 substrate,

respectively. .................................................................................................................... 125

Fig. 4.10 a) Residual stress of TGO and b) average TGO thickness as a function of isothermal

oxidation time at 1150 °C for TBCs with the X4 and N5 substrate, respectively. ................ 126

Fig. 4.11 Mode Ⅰ interface toughness of TBCs on N5 (black square) and X4 substrates (red

circle) as a function of oxidation time, respectively. Data of other TBC systems are also

LIST OF FIGURES

9

included for comparison: Pt diffusion bond coat (1150 °C, X. Wang et al. [132]); NiCoCrAlY

bond coat (Yu-Fu Liu et al. [150]); β-NiPtAl bond coat (Vasinonta and Beuth [146])......... 129

Fig. 4.12 High resolution STEM/EDX analysis of the TGO/bond coat interface: a) and b)

HAADF (high angle angular dark field)/STEM image, b) is the red box area shown in a); c) -

g) STEM/EDX mapping of TGO on N5 substrate after 3 cycles at 1200 °C. ...................... 131

Fig. 4.13 High resolution STEM/EDX of the TGO/bond coat interface: a) and b)

HAADF/STEM image, b) is the red box area shown in a); c) - g) STEM/EDX mapping of TGO

on X4 substrate after 3 cycles at 1200 °C.......................................................................... 131

Fig. 5.1 a) cross-sectional FIB/SEM image and b) surface image of as-etched samples; c)

etching time-thickness plot and d) cross-sectional FIB/SEM image after Pt electroplating on

the etched substrate.......................................................................................................... 137

Fig. 5.2 Two steps to fabricate the Al-enriched γ’-phase coatings: Ⅰ. Fabricate Pt-diffused

intermediate coatings; Ⅱ. Pack cementation aluminizing on the intermediate coatings. ....... 139

Fig. 5.3 Ni-Pt-Al phase diagram at 1150°C [202]. The compositions of sample 1-5 are marked

by the different dots, respectively. The inset SEM image shows the as-fabricated cross-sectional

microstructure of sample 3. .............................................................................................. 143

Fig. 5.4 Ni-Pt-Al phase diagram at 1150°C [202]. The two horizontal red lines represent the

upper and lower limit of Al concentration for a pure γ’-phase coating, respectively. ........... 144

Fig. 5.5 XRD patterns of as-fabricated Al-enriched γ’-phase coatings, Pt-diffused γ/γ’ and β-

NiPtAl coatings. .............................................................................................................. 145

Fig. 5.6 Microstructure of as-fabricated coatings: SEM (BSE) images of: a) γ’ coating, b) γ/γ’

coating and c) β-NiPtAl coating; EBSD phase contrast map of the red box area in a - c: d) γ’

coating, e) γ/γ’ coating and f) β-NiPtAl coating; and corresponding color-coded inverse pole

figure (IPF) mapping g) - i) showing the different grain sizes of three coatings. ................. 147

Fig. 5.7 Ni, Pt and Al concentration evolution of Al-enriched pure γ’ coating by EDX linescan

after a) 0 h, b) 20 h and c) 50 h oxidation.......................................................................... 148

Fig. 5.8 Ni, Pt and Al concentration evolution of Pt-diffused γ/γ’ coating by EDX linescan: a)

0 h, b) 20 h and c) 50 h oxidation. .................................................................................... 149

Fig. 5.9 Ni, Pt and Al concentration evolution of β-NiPtAl coating by EDX linescan: a) 0 h, b)

20 h and c) 50 h oxidation. ............................................................................................... 150

LIST OF FIGURES

10

Fig. 5.10 Glancing angle (3°) XRD patterns of the oxides on a) γ’ coating, b) γ/γ’ coating and

c) β-NiPtAl coating after different oxidation time at 1150 °C. ........................................... 151

Fig. 5.11 a) - c) cross-sectional SEM images of three coatings after 20 h oxidation at 1150 °C;

d) the magnified image of the red box area in c)................................................................ 152

Fig. 5.12 Cross-sectional SEM backscattered electron (BSE) images of three coatings after 50

h oxidation at 1150 °C: a) - c) low magnification; d) - e) high magnification. ..................... 154

Fig. 5.13 Oxide thickness vs. isothermal oxidation time (at 1150 °C) for the three bond coat

systems. .......................................................................................................................... 154

Fig. 5.14 Optical surface images of three coatings after different isothermal oxidation time at

1150 °C. .......................................................................................................................... 156

Fig. 5.15 Profilometer images of (a) as-fabricated Al-enriched γ’-phase coating surface and

after (b) 5 10-minute cycles, (c) 10 cycles, (d) 25 cycles and (e) 50 cycles recorded at the

identical area. (f) The surface profiles and the corresponding Rq of the line shown in (a). ... 158

Fig. 5.16 Profilometer images of (a) as-fabricated Pt-diffused γ/γ’ coating surface and after (b)

5 10-minute cycles, (c) 10 cycles, (d) 25 cycles and (e) 50 cycles recorded at the identical area.

(f) The surface profiles of the line shown in (a). ................................................................ 159

Fig. 5.17 Profilometer images of (a) as-fabricated β-NiPtAl coating surface and after (b) 5 10-

minute cycles, (c) 10 cycles, (d) 25 cycles and (e) 50 cycles recorded at the identical area. (f)

The surface profiles of the line shown in (a). .................................................................... 160

Fig. 5.18 The local interdiffusion flux of Al (𝐽𝐴𝑙) for the Al-enriched γ’-phase coating/CMSX-

4 diffusion couple and the Pt-diffused γ/γ’ coating/CMSX-4 diffusion couple at specific

positions (𝑥𝑖= 1, 5, 10 and 20 μm) after 20 h oxidation. The arrows represent the diffusion

direction. ←: uphill diffusion from inner part to the coating surface; →: from coating to the

inner part of superalloy. ................................................................................................... 164

LIST OF TABLES

11

List of Tables

Table 2-1 Comparison between APS and EB-PVD .............................................................29

Table 2-2 The structural properties of transient alumina phases ...........................................47

Table 2-3 Physical properties of Ni-based superalloys.........................................................53

Table 3-1 Composition of CMSX-4 substrates ....................................................................74

Table 3-2 Electroplating platinum bath ...............................................................................74

Table 3-3 Electroplating time and corresponding Pt thickness, average Pt concentration and

surface roughness ..............................................................................................................78

Table 3-4 θ-Al2O3 fraction of γ/γ’ areas and γ’ areas, respectively for the three Ni-Pt-Al alloys

after 2 min oxidation at 1050 ºC .........................................................................................93

Table 3-5 θ-Al2O3 fraction of γ/γ’ areas and γ’ areas, respectively for the three Ni-Pt-Al alloys

after 10 min oxidation at 1050 ºC .......................................................................................93

Table 3-6 Low-index (hkl) planes and the corresponding d-spacing values of two Ni phases (γ’

and γ) and θ-Al2O3: the d-spacing mismatch between planes of θ-Al2O3 and the corresponding

γ’ (or γ) plane with closest d-spacing matching is calculated and listed as the strain .............97

Table 4-1 Superalloy compositions (atomic %) by EDXRF ............................................... 110

Table 4-2 Elastic moduli of YSZ top coat measured by the cross-sectional nano-indentation

test .................................................................................................................................. 126

Table 5-1 Composition of CMSX-4 superalloy ................................................................. 137

Table 5-2 Electrolyte γ-etching bath parameters................................................................ 138

Table 5-3 Different pack cementation parameters and the resulting coating composition by

EDS after vacuum anneal................................................................................................. 142

LIST OF ABBREVIATIONS

12

List of Abbreviations

APS atmosphere plasma spray

BC bond coat

BSE backscattered electrons

CMAS calcium magnesium alumina silicate

CTE coefficient of thermal expansion

DIC digital image correlation

EDS energy dispersive X-ray spectroscopy

HAADF high angle annular dark field

PLPS photoluminescence piezospectroscopy

SE secondary electron

SEM scanning electron microscope

TBC thermal barrier coating

TEM transmission electron microscope

TGO thermally grown oxide

XRD X-ray diffraction

YSZ yttria-stabilised zirconia

LIST OF PUBLICATIONS

13

List of Publication

C. Liu, X. Zhang, Y. Chen, et al., Effect of superalloy substrate on the lifetime and interfacial

toughness of electron beam physical vapour deposited thermal barrier coatings, Surface &

Coatings Technology (2019), https://doi.org/10.1016/ j.surfcoat.2019.124937.

ABSTRACT

14

Abstract

Study of Bond Coats and Failure Mechanisms of Thermal Barrier Coatings

Chen Liu A thesis submitted to the University of Manchester for the degree of

Doctor of Philosophy, 2019

Bond coats for thermal barrier coating (TBC) applications and the failure mechanisms of TBCs

are addressed in this thesis, with a focus on i) the early stage oxidation of Pt diffusion bond

coats, ii) substrate effects on TBC failure and iii) a new bond coat design.

The early stage oxidation behaviours of γ/γ’-based NiAl bond coats with different Pt additions

are investigated. Pt can slow down the θ-Al2O3 to α-Al2O3 transformation. High resolution

phase mapping by scanning diffraction analysis shows that α-Al2O3 nucleation in the θ-Al2O3

scale is inhomogeneous along the coating/scale interface. Spatially resolved PLPS

(photoluminescence piezospectroscopy) results show a clear correlation between the θ-Al2O3

to α-Al2O3 transition and the γ or γ’ microstructure in the underlying alloy: where Pt stabilises

the γ’ structure, the suppression of θ-Al2O3 to α-Al2O3 transition is observed. The slower θ-

Al2O3 to α-Al2O3 transition rate due to Pt addition leads to a lower compressive stress of the

stable oxide scale, which contributes to the long term stability of the oxide scale.

The effects of substrate composition on the lifetime of TBCs were studied by comparing two

TBCs applied to a CMSX-4 and a René N5 single crystal superalloy substrate, respectively.

Both TBCs were applied by EB-PVD on top of the Pt-diffused γ/γ’ bond coats. Cyclic oxidation

test showed that TBCs deposited on the CMSX-4 substrates exhibited an average lifetime 20%

higher than that deposited on the René N5 substrate. The TGO thickness evolution and the

roughness of the TGO/bond coat interface were comparable for the two TBCs during cyclic

oxidation. To find out the mechanism for the substrate composition effect, a strain-to-fail test

combined with 3D-DIC was employed to measure the bond coat/TGO interface toughness and

its evolution for the two TBCs. The mode I interfacial toughness (Γic) values were almost

identical for the two TBCs (~ 30 J/m2) in the as-deposited state. However, it decreased much

faster for the TBC with a René N5 substrate after oxidation. The fast decrease of interface

toughness was attributed to the sulphur segregation at the TGO/bond coat interface.

A new Al-enriched γ’-Ni3Al bond coat has been developed and its high temperature oxidation

behaviour has been examined and compared with that of the conventional Pt-diffused γ/γ’

coating and the β-NiPtAl coating. This new γ’-phase coating exhibited significantly reduced

Al and Pt depletion during oxidation compared to the two conventional diffusion coatings.

Moreover, although the Al-enriched γ’-phase coating presented faster thermal grown oxide

(TGO) growth than that of the β-NiPtAl coating during isothermal oxidation, it outperformed

the β-NiPtAl coating regarding the rumpling resistance during cyclic oxidation. The Al-

enriched γ’-phase coating also exhibited superior TGO spallation resistance compared to the

Pt-diffused γ/γ’ coating. The mechanisms for the combination of good rumpling resistance and

oxidation performance of this γ’-Ni3Al coating will be addressed.

DECLARATION

15

Declaration

No portion of the work referred to in this thesis has been submitted in support of an application

for another degree or qualification of this or any other university or other institute of learning.

COPYRIGHT STATEMENT

16

Copyright Statement

i) The author of this thesis (including any appendices and/or schedules to this thesis) owns

certain copyright or related rights in it (the “Copyright”) and s/he has given The University of

Manchester certain rights to use such Copyright, including for administrative purposes.

ii) Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may

be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended)

and regulations issued under it or, where appropriate, in accordance with licensing agreements

which the University has from time to time. This page must form part of any such copies made.

iii) The ownership of certain Copyright, patents, designs, trademarks and other intellectual

property (the “Intellectual Property”) and any reproductions of copyright works in the thesis,

for example graphs and tables (“Reproductions”), which may be described in this thesis, may

not be owned by the author and may be owned by third parties. Such Intellectual Property and

Reproductions cannot and must not be made available for use without the prior written

permission of the owner(s) of relevant Intellectual Property and/or Reproductions.

iv) Further information on the conditions under which disclosure, publication and

commercialization of this thesis, the Copyright and any Intellectual Property and/ or

Reproduction described in it may take place is available in the University IP Policy (See

http://www.campus.manchester.ac.uk/medialibrary/poilcies/intellectual-property.pdf), in any

relevant Thesis restriction declarations deposited in the University Library, The University

Library’s regulations (see http://www.manchester.ac.uk/library/aboutus/regulations) and in

The University’s policy on presentation of Theses.

ACKNOWLEDGEMENTS

17

Acknowledgements

I would like to express my gratitude to Professor Ping Xiao for offering me this chance to

conduct this research and acting as my mentor over the past years. His enthusiasm for the work

has inspired me and I will always be grateful for all the guidance, suggestions and

encouragement from him. I also thank Prof. Philip Withers, my co-supervisor, for all the

suggestions on my research.

I would like to thank my senior colleague, Dr. Ying Chen, who has guided me and helped me

since the very beginning of this project. Thanks to his expertise in the research of thermal

barrier coatings, I have learned a lot from him through all the discussions we had.

A special appreciation is given to Dr. Alexander S. Eggeman for conducting all the scanning

diffraction analysis (ASTAR) work shown here. This appreciation is extended to Dr. Xun

Zhang and Dr. Lin Qiu for the helpful suggestions on analysing the 3D-DIC data and

conducting the pack cementation experiments, respectively. I also would like to thank the

experimental officers in the School of Materials including Mr. Matthew Smith, Ms. Xiangli

Zhong, Dr. John Warren, Mr. Kenneth Gyves, Mr. Gary Harrison, Mr. Andy Wallwork, Mr.

Stuart Morse, Mr. Andrij Zadoroshnyj, Mr. Ben Spencer and Mr. Andy Forrest for the help on

my experiments.

I want to thank all the members in the ceramic coating group and all my friends in Manchester

for the help on my study and life.

I would like to thank my beloved families, my parents, Mr. Anwen Liu and Mrs. Yunchuan

Chen, and my grandparents for their love and support.

CHAPTER 1 INTRODUCTION

18

Chapter 1 Introduction

1.1 Gas-turbine engine materials

The gas-turbine engines used for the electricity generation and aircraft propulsion are Carnot

engines where their energy efficiency is proportional to the operating temperature of the turbine

[1]. However, the operating temperature of gas-turbine engines is restricted by the melting

point and high temperature properties of the turbine blade materials. The Ni-base superalloy

has been used as an exclusive structural material for the gas-turbine engines for the sake of its

high melting point (~ 1300 ℃) and superior mechanical properties at high temperatures. The

past decades have witnessed the development of the alloy composition design (for a

combination of oxidation resistance and creep resistance), directionally solidified and single

crystal Ni-base superalloys and internal cooling technologies [2, 3]. All of these have allowed

a steady increase in operating temperatures of gas-turbine engines over decades, which is

shown by the brown line in Fig. 1.1 [2].

A pursuit of higher engine efficiency in industrial applications has required an even higher

operating temperature of the engine. While a big hurdle to achieve this is that the current

operating temperature of the engines has exceed the melting point of the Ni-base superalloys

by ~ 200 - 300 ℃. This suggests that further improvement of the operating temperature is

unlikely to be achieved by optimising cooling technologies or alloy composition design alone.

Thermal barrier coatings (TBCs) were first proposed in 1980s to overcome this hurdle and

achieve higher operating temperatures of the engines. As shown by the red line in Fig. 1.1 [2],

the employment of TBCs has significantly increased the engine’s operating temperature, thus

increasing the engine efficiency.


19

Fig. 1.1 Progress in the maximum allowable temperatures of Ni-base superalloys and thermal

barrier coating (TBC) since 1965. The red line indicates the sharp increase of the allowable gas

temperature by the employment of TBCs [2].

1.2 Introduction of thermal barrier coatings

TBC is a complex, multi-layered and multifunctional system. A typical TBC system on an

airfoil is shown in Fig.1.2 [4], which consists of four layers: a ceramic top coat, a thermally

grown oxide (TGO) layer, a metallic bond coat and a superalloy substrate. The ceramic top

coat which is typically composed of 7-8 wt. % yttria stabilized zirconia (YSZ) is deposited on

the metallic bond coat. It can act as the thermal insulator due to its significantly low thermal

conductivity. The primary function of the metallic bond coat (which is firstly applied to the

substrate) is to provide a compatible bonding between the metallic substrate and the YSZ top

coat. More importantly, it also offers an oxidation resistance for the superalloy substrate by

forming a dense and protective TGO layer during high temperature oxidation. The wide

application of TBCs together with the state-of-art cooling technologies have effectively


20

lowered the surface temperature of the superalloy components, thus significantly increasing

the engine efficiency and prolonging the lifetimes of gas-turbine engines.

Fig. 1.2 Schematic view of a TBC system on an airfoil [4].

Although the use of TBCs have allowed improvements in the operating temperature, there are

still some problems and challenges with the current TBCs raised by further demands in better

durability and reliability of TBC systems [5]. Firstly, a prominent concern with TBCs is the

loss of adhesion and spallation from the underlying metal during service [6]. The underlying

bare metal would be directly exposed to the hot gas environment if the TBC spalls, which is

catastrophic. Hence, it is vital to build a reliable lifetime prediction model of TBCs based on a

comprehensive understanding of the failure mechanism of TBCs. Many phenomena including

phase transformation, oxidation, diffusion, sintering and thermal & mechanical deformation

occur concurrently in a TBC system during high temperature exposures. Therefore, it is of great

difficulty to describe the evolution of microstructures, stresses and thermal & mechanical


21

properties of TBCs during service and to further elucidate the failure mechanisms based on

these descriptions. For instance, despite years of research efforts, the effect of bond coat and

substrate chemistry on the degradation mechanism leading to early spallation of TBCs has not

been well understood [7]. To clarify this chemical composition effect, in-depth investigations

and analysis on the TBC system regarding various aspects (e.g. interfacial microstructure

evolution, oxide growth kinetics, etc.) are required.

Second, since bond coat is the key component in a TBC system and plays a crucial role in

controlling durability of TBCs, new bond coat design has attracted increasing attention to meet

the demand in higher temperature and higher efficiency. Future bond coat design should

consider three aspects for the sake of an improved performance of TBCs [8]:

i) The layer compatibility; the bond coat serves as an intermediate compliant layer to maintain

the top coat adhesion to the metal substrate.

ii) The bond coat must have adequate oxidation properties to form and maintain a protective

TGO scale (usually α-Al2O3).

iii) The mechanical properties; an ideal bond coat should exhibit superior high temperature

strength to resist deformation and is also ductile at room temperatures to be strain tolerant.

The combination of these three aspects makes ideal bond coat designs but it is not

straightforward due to the complexity and interactions of the TBC system.

Third, with increasing operating temperature in advanced engines, TBCs are increasingly

suspected to degrade by molten salts such as calcium-magnesium-alumina-silicate (CMAS) [9],

which have posed challenges for new TBC materials and engineering processing techniques.


22

1.3 Objectives and structure of the dissertation

The primary objective is to investigate the bond coat and substrate chemical composition effect

on the oxidation behaviour of TBCs in terms of TGO phase transformation, stress evolution,

surface rumpling and how these phenomena are correlated with the compositional changes of

the bond coat/substrate. In addition, a new bond coat design which combines the three

principles described in Section 1.2 has been proposed and successfully implemented. Detailed

descriptions for each chapter are listed as follows:

In Chapter 1, a brief introduction of gas-turbine materials and TBC system is given, followed

by the challenges in current TBC studies.

In Chapter 2, different components in a TBC system are reviewed in respect of material

selection, material properties and the processing methods. The degradation and failure

mechanisms for different TBC systems are also reviewed.

In Chapter 3, the effect of platinum addition on the early stage oxidation behaviour of Pt-

diffused γ-Ni/γ’-Ni3Al bond coats is investigated. Many previous studies have reported that Pt

can greatly improve the prolonged oxidation performance of the γ-Ni/γ’-Ni3Al bond coats.

However, only limited work has focused on the Pt effect on the early stage oxidation behaviour,

which can also affect the bond coat oxidation performance. Moreover, some controversial

views regarding Pt effect on the early stage oxidation of Pt-diffused bond coats have been

reported, so further study is required to clarify this issue. In this study, the early stage oxidation

behaviours of three Pt-diffused γ-Ni/γ’-Ni3Al bond coats with different Pt additions were


23

studied and compared in terms of the TGO phase transformation, residual stresses of TGO and

the TGO growth rate. Pt effect on the TGO phase transformation during early stage oxidation

is summarized and discussed. A new phase transformation mechanism based on high resolution

phase mappings by scanning diffraction analysis is proposed to explain the Pt effect on the

TGO phase transition.

In Chapter 4, the substrate composition effect on the lifetime of TBCs is studied by comparison

of two TBC systems with different superalloy substrates. Although a number of studies have

reported that the superalloy substrate composition can affect the cyclic lifetime of TBCs, the

mechanism of this substrate composition effect has not been fully understood. In this study,

two TBCs with Pt-diffused γ-Ni/γ’-Ni3Al bond coats are applied to a CMSX-4 and a René N5

single crystal superalloy substrate, respectively. Cyclic oxidation tests were carried out on these

two TBCs to compare their lifetimes as well as microstructural evolution and TGO growth

kinetics. A strain-to-fail test combined with 3D-DIC (digital image correlation) was employed

to measure the bond coat/TGO interface toughness and its evolution for the two TBC systems.

Finally, the mechanism of the substrate composition effect on the interface toughness evolution

and TBC lifetime was explored.

In Chapter 5, a new Al-enriched γ’-Ni3Al bond coat was designed according to the three

principles for the future bond coat design (as described in Section 1.2). This new bond coat

was deposited on the CMSX-4 superalloy substrate by selective etching of the CMSX-4

substrate combined with the low-temperature pack cementation. The isothermal oxidation

behaviours of this new Al-enriched γ’-Ni3Al bond coat have been studied and compared to the


24

conventional industry-standard Pt-diffused γ/γ’ coating and β-NiPtAl coating in terms of TGO

microstructure & growth rate, Pt & Al depletion and TGO spallation resistance. Furthermore,

thermal cycling to a maximum temperature of 1150 °C was conducted on all coatings to

compare their rumpling behaviours. The mechanisms for the good combination of rumpling

resistance and oxidation performance of this new γ’-Ni3Al-base coating are discussed.

In Chapter 6, the main conclusions and future work are summarised.

CHAPTER 2 LITERATURE REVIEW

25

Chapter 2 Literature Review

2.1 Thermal barrier coating system

The application of TBCs in conjunction with external cooling technologies can achieve a

temperature gradient up to ~ 170 ℃ across the coating [10]. As a result, the inlet gas

temperature and the efficiency of gas turbine engines have been significantly improved. A

state-of-the-art TBC system consists of four layers. Each layer and its functions and properties

are shown in Fig. 2.1 [2]. The nickel-base superalloy substrate is the structural component

which can sustain creep and cyclic fatigue. A bond coat layer is coated on the superalloy

substrate, which can provide adhesion between the superalloy substrate and ceramic top layer.

The TGO layer (usually α-Al2O3) forms between the metallic bond coat and the ceramic top

coat due to bond coat oxidation during high temperature exposures. The topmost layer is a

ceramic layer which acts as the actual heat shield. It usually has a combination of low thermal

conductivity and high strain tolerance. Due to the coupled diffusional and mechanical

interactions between each layer, the TBC system is dynamic during service and all layers

interact with each other to control the performance and durability of the system [11].

2.2 Top coat

2.2.1 Material requirement and selection

The first and foremost requirement for the top coat material is the low thermal conductivity

because its primary function is to provide the thermal insulation for the underlying metal

components [3]. In addition to the low thermal conductivity, the extreme thermomechanical

working environment has led to other requirements for the top coat material:


26

(1) The top coat should have phase stability during prolonged thermal exposure because the

disruptive volume change accompanied with the phase transformation can cause damage of the

material [5].

(2) The top coat material must be stable to resist sintering and erosive pollutant attack (e.g.

CMAS) during high temperature exposures in an oxidising environment [12].

(3) The top coat should be strain-tolerant and fracture-resistant during thermal cycling. The

strain tolerance requires the material to withstand the strains induced by the thermal misfit

between the top coat and the metallic substrate during thermal cycling, and the fracture

resistance can mitigate the deformation during thermal cycling and impact damage from

airborne particles.

(4) The material should have thermodynamic compatibility with the TGO material (usually

alumina) to ensure a good interfacial adhesion of the TBC system.

Fig. 2.1 Schematic illustration of the multi-layered and multifunctional TBC system. The

functions and properties for each layer are indicated [2].


27

All of these requirements have provided the guidelines for the top coat material selection and

development in the last decades and have laid the foundation for state-of-the-art TBC materials.

Currently, the top coat is typically composed of ~ 7 wt. % yttria stabilised zirconia (7YSZ)

because of its desirable properties which coincides with the above requirements for top coat

materials. First and foremost, as can be seen from Fig. 2.2 [3], YSZ has a low thermal

conductivity (~ 1-3 W/mK). This can be ascribed to its high concentration of point defects

including solute cations and oxygen vacancies, which can reduce the lattice thermal

conductivity by scattering lattice waves [13]. Moreover, despite being a ceramic material, the

coefficient of thermal expansion (CTE) of YSZ is well matched to that of the Ni-based

superalloys (Fig. 2.2), leading to reduced thermal misfit strains during thermal cycling

compared to other ceramics.

As described above, one of the requirements for the top coat material is the phase stability upon

thermal cycling to avoid volume changes associated with phase transformations that can lead

to the degradation. The dopants, such as Y2O3, MgO, CeO2, CaO and Sc2O3 can significant ly

enhance the phase transformation resistance compared to pure zirconia. Despite the fact that

all of these dopants can retard the phase transformation of ZrO2 to some extent, a large number

of studies have found that ZrO2 doped with 7 wt.% Y2O3 (7YSZ) exhibited the longest thermal

cycling lifetime and therefore is the most suitable candidate for the top coat [11]. Under

equilibrium condition, 7YSZ is stabilised as the tetragonal phase above 1050 ℃. This

tetragonal YSZ is transformed into a mixture of monoclinic and cubic YSZ upon cooling unless

mechanically constrained [14, 15]. However, due to the non-equilibrium fabrication process of

the top coats (e.g. electron beam physical vapour deposition), as-deposited 7YSZ coatings

typically have a metastable tetragonal prime (t’) phase instead of the equilibrium tetragonal (t)

phase [16]. Although these two phases are similar in structure, the t’ phase has been considered


28

to be a non-transformable phase because it does not undergo any phase transformation after

prolonged thermal exposure at 1200 ℃ [17]. This makes the 7YSZ a good choice for TBC

applications.

As a refractory ceramic, YSZ has long-term sintering resistance during the high temperature

oxidation. In addition, the porous microstructure of TBC top coat resulting from the

manufacture process can contribute to the strain tolerance of the coating. Another merit of YSZ

is the high fracture toughness due to the ferroelastic toughening [18] combined with the phase

transformation toughening (martensitic transformation from tetragonal to monoclinic phase)

[19]. Thus the YSZ can fulfil the third requirement as listed above. Finally, YSZ exhibits

chemical compatibility and strong bonding with the alumina, which ensures the long-term

stability of the TBC system [3].

Fig. 2.2 Coefficients of thermal expansion (CTEs) of a range of materials are cross-plotted

against their thermal conductivities [3].


29

2.2.2 Deposition techniques and microstructure

Currently, TBC top coats are generally deposited by plasma spraying (PS) or by electron beam

physical vapour deposition (EBPVD). PS uses a high-temperature plasma jet, to melt and

accelerate the powder feedstock (the material to be deposited) toward the substrate. The plasma

jet can be generated by passing a gas (usually Ar, He or N2) through an electric arc, during

which the gas will be ionised and form the plasma jet. Then the powder feedstock is injected

into the plasma jet either internally or through an external feed-port [20]. After the

instantaneously melting of the powder feedstock, the semi-molten powders are rapidly

accelerated towards the cold substrate. A mechanically bonded coating is deposited

immediately on the substrate surface by spreading and solidification. Depending on the process,

PS can be done at ambient conditions (atmospheric plasma spray, APS) or at controlled

conditions (e.g. vacuum). EBPVD was first introduced to fabricate TBCs in 1980s [21].

Electron beam is designed to transform target atoms into the gaseous form. These atoms then

precipitate into solid form and coat onto the preheated substrate [22]. Coatings deposited by

EBPVD are mainly used for extreme thermomechanical working conditions (e.g. blades of

aeroengines), while APS are more commonly used nowadays because of its operation

robustness and economic viability compared to EBPVD [23]. Table 2.1 gives a comparison of

these two coating techniques.

Table 2-1 Comparison between APS and EB-PVD

Methods Occasions [12] Interfacial bonding Equipment cost [12]

APS

Stationary parts on aeroengines;

stationary and rotating parts of

land-based power generation

engines

Mechanical bonding £0.6-1.1 million

EBPVD Vanes or blades of aircraft

engines Chemical bonding £10-20 million


30

As mentioned above, due to the non-equilibrium fabrication process of the top coats (e.g.

EBPVD and APS), as-deposited 7YSZ coatings typically have a metastable tetragonal prime

(t’) phase instead of the equilibrium tetragonal (t) phase. In addition, APS and EBPVD TBCs

have totally different microstructures. As a result, they exhibit different advantages in terms of

properties. The thickness of APS TBCs is about hundreds of micros to several millimetres. Fig.

2.3 a shows that APS TBCs are featured by splat grains and inter-splat plate-like pores which

are parallel to the substrate/coating interface. These inter-splat pores resulting from the rapid

solidification can effectively reduce the thermal conductivity. On the other hand, because the

pores are parallel to the interfaces, APS TBCs generally have less strain compliance and

therefore shorter thermal cycling lifetimes than EB-PVD TBCs.

EBPVD TBCs are usually ~ 120-150 μm thick and exhibit columnar grain morphology with

inter-column gaps (Fig. 2.3 b). The gaps between the disconnected columns can accommodate

the thermal misfit strain between ceramic and metallic components during cycling exposures

and thus providing better strain compliance. However, the lack of large splat pores normal to

the heat flow direction can lead to higher thermal conductivity compared to APS TBCs [12].

Fig. 2.3 Cross-sectional microstructure of a) APS and b) EBPVD TBC [24].


31

2.2.3 New top coat candidates

The upper limit of use temperature for 7YSZ is 1200 ℃. This can be explained by two reasons.

First, the t’ phase of YSZ will decompose into a high yttria cubic phase and a low yttria

tetragonal phase after the long term exposure at elevated temperatures [14]. The latter will

transform to a monoclinic phase upon cooling accompanying with a large volume expansion,

which accelerates the TBC failure. Another reason for the temperature limit is the sintering of

YSZ at temperatures above 1200 °C leading to loss of strain tolerance and early failure [23].

Therefore, considerable efforts are being invested in identifying new top coat materials with

better performance than the current industry-standard 7YSZ [23, 25-27].

Several studies found that some rare earth element doped zirconia, such as pyrochlores

(A2B2O7) and perovskites (ABO3) have lower thermal conductivity than 7YSZ [23].

Furthermore, their thermal stability and sintering resistance are better compared to 7YSZ,

which makes them promising candidates for the TBC top coat. However, relatively lower CTE

(e.g. YSZ CTE: 11×10-6 °C, La2Zr2O7 CTE: 9×10-6 °C [27]) of these materials impedes their

development because the thermal mismatch between the substrate and coating can induce

higher thermal stresses. Recent studies [26, 28] have proposed a double ceramic top coat

combining YSZ with another ceramic of lower thermal conductivity, which exhibited longer

thermal cycling life than the single YSZ layer. Currently, since no single material can meet all

the requirements for the extremely complicated TBC system, the double layer top coat seems

promising in future TBC applications.


32

2.3 Bond coat

2.3.1 Material requirements

The YSZ top coat is not deposited directly onto the substrate. Since YSZ is transparent to

oxygen diffusion, during service, oxygen would diffuse through the interconnected pores of

YSZ to the interface and oxidize the superalloy by forming fast growing Ni-rich oxides [29].

These Ni-oxides (e.g. NiO and Ni(Cr, Al, Co, Ti, Ta)2O4) are thermodynamically incompatible

to the YSZ layer and can cause early spallation of the top coat [30]. To avoid this problem,

bond coats with sufficient high temperature oxidation resistance are applied to the superalloy

substrates prior to the TBC deposition.

The bond coat functions as an intermediate adhesion layer between the YSZ top coat and the

superalloy substrate. In addition to this, the most significant function of bond coats is to provide

sufficient oxidation resistance for the superalloy substrates by forming a slow-growing and

protective TGO scale on the surface of the bond coat when it oxidizes at high temperatures.

Currently, the industry-standard bond coats are commonly made of alloys with specific

aluminium-rich compositions, which can result in the formation of a TGO scale that mainly

consists of α-Al2O3. The dense α-Al2O3 scale can prevent the oxidation of the underlying

superalloy and is also thermodynamically compatible with the YSZ top coat. Apart from the

oxidation resistance, the bond coat is also expected to minimize the interdiffusion with the Ni-

based superalloy substrate [31]. Furthermore, the bond coat material should be morphologically

stable to resist plastic deformation of the surface induced by thermal stresses during thermal

cycling. Because the surface deformation can cause interfacial debonding and spallation failure

of the coating.


33

2.3.2 Bond coat categories

There are two categories of bond coats based on the fabrication techniques: diffusion coatings

and overlay coatings [32]. The former can be further classified into two categories by their

phase constituents. One is comprised primarily of the β-NiAl phase (referred to as nickel

aluminide coatings), and the other is composed of γ-Ni/γ’-Ni3Al phase. The overlay coatings

are typically made of MCrAlY (M=Ni, Co or a combination of both) alloys. The following

sections will elaborate the deposition technique, composition, microstructure and properties of

these coatings, respectively.

2.3.2.1 β-NiAl based diffusion bond coat

The β-NiAl based diffusion bond coats are fabricated by a high temperature interdiffusion

annealing process between an external aluminium source and the nickel superalloy substrate.

The external aluminium source can be applied by vapour deposition methods such as vapour-

phase aluminizing (VPA) [33], chemical vapour deposition (CVD) [34, 35] and pack

cementation [36, 37]. These methods all involve in a reaction between an aluminium donor and

a halide activator firstly to generate the gaseous aluminium halide. Then this gaseous

aluminium halide reacts with the nickel superalloy substrate and forms the β-NiAl based nickel

aluminide coatings at elevated temperatures. Among these deposition methods, the pack

cementation has been widely applied due to its low cost, processing simplicity and flexibility

for different specimen dimensions and geometries [38]. For the pack cementation process, the

pack powder, which consists of the aluminium source (usually pure Al), an inert filler (e.g. α-

Al2O3) and a halide activator such as NH4Cl or AlCl3, is ground and mechanically mixed firstly.

Then the specimen is buried in the pack powders charged into an air-tight alumina crucible.

The crucible is then heat treated in a protective atmosphere (e.g. argon gas). During the heat

treatment, two procedures take place simultaneously in the pack [38]. One is the

activation/migration process, which includes the chemical reaction to create the aluminium


34

halide (AlXn, X=F, Cl or Br and 1≤n≤3) vapour and the migration of the vapour to the specimen

surface. Another process (termed as deposition/diffusion process) is the deposition of

aluminium by the oxidation-reduction reaction between the aluminium halide and the metal,

followed by the interdiffusion between the deposited aluminium and alloying elements (e.g.

Ni) in the metal substrate. The key step is the generation of the aluminium halide vapour. This

gaseous phase will react with the metal to deposit aluminium by the following reaction:

𝐴𝑙𝑋𝑛 +𝑛

𝑛 − 1𝑁𝑖 → 𝐴𝑙 +

𝑛

𝑛 − 1𝑁𝑖𝑋𝑛−1

The β-NiAl phase has a B2 crystal structure, which is an ordered body centred cubic (bcc)

structure consisting of two simple cubic interpenetrating sublattices. From the Ni-Al binary

phase diagram (Fig. 2.4 [39]), it can be seen that the β-NiAl phase composition range is wide,

indicating that the Al concentration of β-phase can vary significantly from its NiAl

stoichiometric composition.

Fig. 2.4 Binary phase diagram of the Ni-Al system [39].


35

The microstructure of the β-NiAl coating is relied on the pack cementation process.

Specifically, there are two categories of pack cementation according to the heat treatment

temperature: the low-activity and high-temperature (above ~ 1000 ℃) pack cementation and

the high-activity and low-temperature (below ~ 950 ℃) pack cementation [40]. The first type

(low-activity) coating has two zones: the outer zone (Zone 1 in Fig. 2.5 a) and inner zone (Zone

2 in Fig. 2.5 a). Both zones have a nickel-rich β-NiAl-phase matrix. Zone 1 contains various

amounts of substrate alloying elements (e.g. Cr, Mo, Co, and Ti) in solution, and Zone 2 has a

variety of dispersed phases including MC and M23C6 carbides and σ(Cr, Mo, Co) phases [41].

While the high-activity coating has three zones: the outer Zone 1, the middle Zone 2 and the

inner Zone 3 (Fig. 2.5 b), and all of them have an Al-rich β-NiAl-phase matrix. Specifically,

Zone 1 (Fig. 2.5 b) contains dispersed α(Cr, Mo) phases and carbides, and the latter are identical

to those found in the underlying Ni-based superalloy [41]. The middle Zone 2 (Fig. 2.5 b) is

comprised of σ-phase and carbides in the β-NiAl-phase matrix, which is similar to the outer

Zone 1 of the low-activity coating. And Zone 3 (Fig. 2.5 b) also shows similar microstructure

to that of Zone 2 for the low-activity coating (Fig. 2.5 a).


36

Fig. 2.5 The cross-sectional SEM images of the β-NiAl bond coats fabricated by a) low-activity

and high-temperature; b) high-activity and low-temperature pack cementation [41].

The distinct microstructures of these two types of β-NiAl coatings can be explained by the

coating formation mechanisms. The formation of low-activity coatings are based on

predominant outward diffusion of Ni. On the other hand, the high-activity coating forms as a

result of the predominant inward diffusion of Al, and a higher amount of Al can diffuse into

the coating during the aluminizing process due to the significantly higher Al activity. This can

explain the Al-rich β-phase matrix for the high-activity coating. Another characteristic

difference is that the low-activity coating has much less amount of complex precipitates

throughout the coating, especially in the outer zone. However, the high-activity coating has

numerous precipitates in the entire coating. This is due to the fast inward diffusion of Al during

coating formation. The slow diffusing alloying elements (e. g. Ta, Mo etc.), which are

originally from the substrate, may become trapped in the fast-forming β-NiAl matrix, and

precipitate out due to their limited solubility in the β-phase [42].


37

The β-NiAl phase, due to its high Al concentration, can serve as an Al reservoir for the

formation and continuous growth of the protective Al2O3 scale during thermal exposures. This

dense alumina scale can protect the underlying superalloy from being oxidized. However,

during long-term oxidation, especially under cyclic conditions, the alumina scale can spall off,

which results in the loss of YSZ top coat and failure of the TBC system. Thus, the TGO scale

adhesion is significant for evaluating the oxidation performance of the bond coat. Pt addition

is the most widely applied method to improve the adhesion of alumina scales [42-44]. The

state-of-art industrial process to fabricate the platinum modified nickel aluminide (β-NiPtAl)

coatings includes a Pt electroplating process (5-7 μm Pt layer), followed by an aluminizing

process by pack cementation or CVD [5]. The microstructure of the as-fabricated β-NiPtAl

bond coat is similar to that described for the β-NiAl coating, as shown in Fig. 2.6.

Fig. 2.6 The cross-sectional SEM image of the as-fabricated β-NiPtAl bond coat on the CMSX-

4 superalloy substrate.


38

The β-NiPtAl coatings exhibit significantly improved TGO scale adhesion and spallation

resistance, compared to the unmodified β-NiAl coatings. For instance, Hou and Tolpygo [44]

have conducted cyclic oxidation tests at 1150°C in air for both NiAl and NiPtAl coatings on

the same single crystal superalloy substrate. The TGO scale on the β-NiAl coating spalled after

~ 300 10-min cycles, while the oxide scale of the β-NiPtAl coating remained adherent even

after 2000 10-min oxidation cycles. Numerous studies [42, 45-47] have reported that Pt

addition can improve the TGO adherence, and some possible mechanisms have been proposed

to explain this Pt effect. One possibility is that Pt addition can directly strengthen the interface

bonding between the alumina and the NiAl alloy. However, Svensson et al. [48] have calculated

the work of separation of the α-Al2O3(0001)/β-NiAl (111) interface, in pure and Pt-rich NiAl

materials respectively using the density functional theory (DFT). They found that the interfacial

bonding is decreased with Pt addition, which excludes the interfacial strengthening effect of Pt.

Some studies [44, 46] confirmed that platinum can reduce the interfacial pores. However, it is

uncertain whether this effect alone can fully explain the improved scale adhesion. It has also

been proposed that Pt can prevent the sulphur segregation at the TGO scale/coating interface

[45, 49], or limit the outward diffusion of minor alloying elements (e.g. Ta, Ti, Re etc.) from

the substrate [45], thus improving scale adhesion. In a word, currently there is no well-accepted

mechanism for this Pt effect on the TGO scale adhesion.

2.3.2.2 Pt-diffused γ-Ni/γ’-Ni3Al bond coat

The success of Pt addition to the β-NiAl bond coats has inspired the invention of another

diffusion coating, the Pt-diffused γ-Ni/γ’-Ni3Al bond coat. Since a problem associated with the

β-NiPtAl bond coats is that the conventional aluminizing process (e.g. pack cementation) can

introduce undesirable elements or impurities that can lead to poor adhesion between the bond

coat and the top coat. The Pt-diffused γ-Ni/γ’-Ni3Al bond coats have eliminated the


39

aluminizing process by electroplating a thin layer of Pt (7 - 12 μm) on the single crystal

superalloy substrate followed by a diffusion heat treatment at 1100 -1200 ℃ up to several hours

[50]. This not only reduces the fabrication cost, but also increases the coating stability because

the γ-Ni/γ’-Ni3Al phases have good chemical compatibility with the superalloy substrate [51].

The Pt-diffused γ-Ni/γ’-Ni3Al bond coats are formed by the interdiffusion between the Pt layer

and the nickel-based superalloy. During the process, the inward diffusion of Pt into the

superalloy will destroy the original γ/γ’ lattices of the superalloy and expand the unit cells due

to the large atom radius of Pt. The new γ’ and γ phases precipitate out with Pt in solid solution

and exhibit a strip-like microstructure. After extended heat treatment, the strip-like

microstructure is elongated, as shown in Fig. 2.7. The Pt-enriched γ-Ni phase has a fcc (face

centred cubic) structure and the Pt-enriched γ’-Ni3Al phase has a L12 structure. The γ’-phase

shows a brighter contrast with more Pt enriched in the Z contrast image.

Fig. 2.7 The typical microstructure of a Pt-diffused γ-Ni/γ’-Ni3Al bond coat on the CMSX-4

superalloy substrate. The γ’-phase: brighter contrast; γ-phase: grey contrast.


40

Platinum plays a crucial role in the oxidation performance of the γ/γ’-based bond coat. First,

Pt in the γ/γ’-based bond coat has sustained the benefit of improving oxide scale adhesion as

described in Section 2.3.2.1 for the β-NiAl bond coat. Second, Pt can promote the selective

oxidation of aluminium. In other words, Pt can promote the growth of the protective Al2O3

scale and inhibit the growth of detrimental Ni-oxides such as NiO or spinel. According to Izumi

and Gleeson [52], the extents of NiO and spinel formation decreased significantly with

increasing Pt content during cyclic oxidation of γ/γ’-based nickel alloys at 1150 ℃, which in

turn improved the oxidation resistance of the γ/γ’ alloys. Briefly, there are three factors

contributing to the selective oxidation of aluminium due to Pt addition [53, 54]:

(1) The inert nature of Pt can ensure that no Pt involved oxidation reaction takes place even at

extremely high temperatures (~ 1500 ℃).

(2) The subsurface enrichment of Pt at the TGO/metal interface can reduce the chemical

activity of aluminium (aAl) at the interface, causing an uphill diffusion of aluminium from the

inner part of the metal to the surface, which promotes the exclusive growth of the Al2O3 scale.

(3) Pt replaces almost solely to the Ni sites in the γ/γ’ structure, which increases the Al:Ni atom

ratio on a given crystallographic plane containing both Ni and Al. As a result, the aluminium

oxidation is favoured rather than nickel.

Apart from the Pt contribution, a recent study [55] has pointed out that numerous grain

boundaries near the surface of the Pt-diffused γ-Ni/γ’-Ni3Al bond coat can provide fast

diffusion paths of aluminium at initial stage of oxidation, which also contributes to the

exclusive growth of Al2O3 scale.


41

As mentioned above, the Pt-diffused γ-Ni/γ’-Ni3Al bond coat has some advantages over the β-

NiPtAl bond coat including low fabrication cost and better phase stability. Furthermore, this

bond coat also exhibits negligible morphological instability during cyclic oxidation due to its

higher creep resistance compared to the β-based coating [56]. However, a concern with this

bond coat is its relatively low aluminium concentration, leading to the formation of detrimental

Ni-oxides after long-term high temperature oxidation. These Ni-oxides such as spinel can

significantly degrade the oxide/coating interface adhesion due to their brittleness, which leads

to premature failure of the coating system [57]. A new bond coat design by improving the

aluminium concentration while maintaining the structure of γ’-phase seems to be attractive for

optimizing the TBC system. Actually, some efforts have been made to improve the aluminium

concentration of Pt-diffused γ-Ni/γ’-Ni3Al bond coats by a secondary aluminizing process

(pack cementation) [58, 59]. However, the as-fabricated Al-enriched γ-Ni/γ’-Ni3Al bond coats

only exhibited limited oxidation performance improvement due to the Al depletion during

oxidation. Therefore, further studies on this issue are demanding for optimizing the current

TBC system.

2.3.2.3 MCrAlY overlay bond coat

The MCrAlY overlay bond coats are directly sprayed onto the surface of Ni-based superalloy

substrates using physical deposition techniques such as APS [60, 61], high velocity oxygen

fuel spraying (HVOF) [55, 62, 63], low-pressure plasma spraying (LPPS) [12, 64] and EBPVD

[65]. Unlike the diffusion coatings, the overlay coating composition and thickness can be

tailored by the coating source and deposition time with great flexibility and accuracy, which

makes this coating a desirable candidate in some applications.

The compositions of MCrAlY (M=Ni, Co or a combination of both) overlay bond coats are

typically: (in wt. %) ~ 15 - 25 % Cr, ~ 10 - 15 % Al, ~ 0.2 - 1% Y and Ni (Co) in balance [12].


42

Typical cross-sectional images of a NiCoCrAlY bond coat deposited by HVOF are shown in

Fig. 2.8. From Fig. 2.8 a), it can be seen that this deposition technique can produce a dense

coating with the uniform thickness and low oxide content. The as-deposited NiCoCrAlY bond

coat exhibits a β (grey contrast) + γ (white contrast) two-phase microstructure (Fig. 2.8 b). The

β-phase can serve as the main aluminium source to improve its oxidation resistance. The TGO

scale on this bond coat is mainly composed of α-Al2O3. However, after prolonged high

temperature oxidation, some Y/Al rich oxides are also observed in the TGO scale apart from

α-Al2O3 [66]. On the other hand, the γ-phase is designated to improve its mechanical ductility.

Chromium is added to enhance the hot corrosion resistance of the coating [67]. In addition,

some studies also reported that Cr can promote the selective growth of α-Al2O3 by a third-

element effect mechanism [68], thus also contributing to the oxidation performance. The minor

additions of reactive element (RE) yttrium in this bond coat can significantly improve the TGO

scale adhesion thus extending the lifetime of the coating. Although several mechanisms have

been proposed to explain this RE effect, it is commonly accepted that yttrium can improve the

scale adhesion by segregating to the metal/scale interface thereby preventing the detrimental

sulphur segregation [69, 70].

Fig. 2.8 Cross-sectional images of the as-deposited NiCoCrAlY bond coat deposited by HVOF

[71]: a) optical image and b) back scattered electron (BSE) image (high magnification) showing


43

the β (grey contrast) + γ (white contrast) two-phase microstructure. The black contrast areas

are interfacial pores between metal particles.

2.4 Thermally grown oxide

2.4.1 Material requirements and selection

The oxygen from the engine environment can diffuse through the YSZ top coat during the

service of gas turbine engines. As a result, the bond coat will oxidise and form a thermally

grown oxide (TGO; 1 - 10 μm in thickness) layer between the bond coat and the YSZ top coat

[72]. Two factors have contributed to the oxygen diffusion from the engine environment to the

bond coat surface during oxidation. First, the interconnected pores in the YSZ top coat have

provided diffusion paths for oxygen to the bond coat surface. Second, the high oxygen

diffusivity of the YSZ top coats makes it ‘oxygen transparent’ [11, 73].

The extreme operating conditions have raised several requirements for the TGO material:

1) The TGO material should be phase compatible with YSZ to ensure thermodynamic stability

of the top coat for long-term high temperature exposure [3].

2) It should be a slow-growing and stable oxide at high temperatures [3]. The fast-growing

oxides can result in very thick TGO layer during high temperature exposures, which increases

the trend for oxide spallation due to an increasing strain energy in the TGO scale.

3) It should have low oxygen diffusivity [74]. The TGO layer must perform as an oxygen

diffusion barrier to protect the underlying metallic part from being oxidised.

4) It should be mechanically robust to resist fracture, especially in highly cyclic scenarios.

Considering all of these requirements, the major classes of bond coats have been developed to

form a TGO scale which is predominantly composed of α-Al2O3 during high temperature

exposures in air.


44

2.4.2 TGO microstructure and stress

The pure Al2O3 TGO layer usually exhibits two microstructural zones after prolonged oxidation:

a columnar zone (CZ, columnar grains) next to the bond coat and an equiaxed zone (EZ,

equiaxed grains) next to the YSZ (Fig. 2.9 [4]). The inner columnar oxide grains are formed

by the inward diffusion of oxygen anions and the outer equiaxed grains are formed by outward

diffusion of aluminium cations.

Fig. 2.9 A fractured cross-sectional image of a TGO scale showing the columnar grains formed

by the inward diffusion of oxygen and equiaxed grains formed by the outward diffusion of

aluminium [4].

Generally, the residual stress of TGO at room temperature consists of two components: the

thermal mismatch stress and the growth stress. The former is generated in the TGO scale by

cooling from the elevated temperature to the ambient temperature due to the thermal expansion

mismatch between the TGO and the Ni-based superalloy [75]. Since α-Al2O3 has a much lower

coefficient of thermal expansion (CTE; 8 - 9 × 10-6 m·C-1 [74]) than that of the Ni-based

superalloy (12 - 16 × 10-6 m·C-1), the thermal mismatch stress is usually compressive (~ -3 - 5

GPa). The growth stress is generated mainly due to the lateral growth of TGO scale [76]. The


45

lateral elongation of TGO is constrained by the underlying metal substrate thus generating a

compressive growth stress in the scale.

While the thermal mismatch stresses can be calculated quite well, the magnitude of the growth

stress is determined by the dynamic competition between two opposite processes at high

temperatures: the stress generation and concurrent creep relaxation of the TGO layer.

Consequently, the magnitude of the growth stress is evolving during the oxidation process and

is dependent on the lateral growth strain, creep strain of the TGO layer and plastic deformation

of the underlying metal [77]. Several studies have reported the magnitude of growth stress of

alumina scales on various alloys. For instance, Schumann et al. [78] have measured the growth

stress of α-Al2O3 scales on NiAl alloys during oxidation at 1100 ℃ by X-ray diffraction, and

concluded that the compressive growth stress was very low (less than ~ 50 MPa). On the other

hand, the growth stress of alumina scales on FeCrAl alloys measured by the same technique

was ~ -1 GPa [79].

Knowledge of the residual stress of TGO is crucial because the strain energy associated with

the TGO stress and thickness results in the delamination of the TGO scale from the underlying

metal [4]. In practice, the premature damage of the TGO scale is usually accompanied with

variations of local TGO stress (e.g. decrease in magnitude or change of stress state) in the

damaged area. Since the TGO is buried underneath a thick YSZ top coat, there is a need for

non-destructive tools to probe through the entire YSZ top coat and detect the TGO stress before

the failure occurs [80]. One promising tool is photoluminescence piezospectroscopy (PLPS).

PLPS technique can measure the TGO stress underneath the YSZ layer based on the R

luminescence (Fig. 2.10 [81]), which is generated by the phonon emission of chromium (Cr3+)

impurity in the Al2O3 scale when appropriately excited by an argon laser [82, 83]. Two


46

fluorescence transitions of the dopant ion Cr3+ in the Al2O3 crystal correspond to the R1 and

R2 fluorescence doublet as shown in Fig. 2.10. Since the fluorescence transition is extremely

sensitive to the local ionic environment in the host crystal, stresses (or deformations) which

alter the interionic distances will shift the position of the doublet. The linear relationship

between the frequency shift of R luminescence and the in-plane equi-biaxial TGO stress

magnitude has been calibrated with a precision of ~ 10 MPa by J. He and co-workers [84]. In

addition, other luminescence parameters are also good indicators for the damage quantification

in TBCs, such as the peak shape, intensity ratio of R1 and R2, peak broadening and peak

separation [80, 85, 86].

Fig. 2.10 a) Schematic illustration of the PLPS technique and b) typical R1/R2 fluorescence

spectra for Cr-containing stress-free (dashed line) and stressed α-Al2O3 (solid line) [81].

2.4.3 TGO transformation during early stage of oxidation

As mentioned above, α-Al2O3 is the predominant oxide in the TGO during the service of TBCs.

However, before the stable growth of the α-Al2O3 scale, one or more alumina polymorphs


47

usually grow firstly during early stage of oxidation. These phases, also called transient

aluminas, will transform to the stable α-Al2O3 at higher temperatures or after longer exposures.

The following phase transformation sequence has been reported in the literature during the

early stage oxidation of alumina-forming alloys [87-90]:

γ 750℃→ 𝛿

900℃→ 𝜃

1000℃→ 𝛼

The structural properties of these transient alumina phases are summarized in Table 2.2. The

lattice parameters, space groups and the orientation relationships with respect to γ-Al2O3 of

these three transient alumina phases are listed. δ-Al2O3 has been described as a superlattice of

the spinel structure with ordered cation vacancies and has a tetragonal symmetry. θ-Al2O3 is

the most widely studied polymorph of alumina and has a monoclinic symmetry and space group

c2/m. The aluminium cations are equally distributed over the octahedral and tetrahedral sites.

Table 2-2 The structural properties of transient alumina phases

Phase Lattice parameters

[88]

Space group Cations/unit cell

[89,90]

Orientation

relationship

with respect to

γ-Al2O3 [89,90]

γ aγ= 7.9 Å Fd3̅m 64/3 -

δ aδ= 7.9 Å

bδ = 15.8 Å

cδ = 11.9 Å

P212121 64 (100)δ║ (100)γ

[100]δ║ [001]γ

θ aθ= 11.9 Å

bθ = 2.8 Å

cθ = 5.7 Å

β=104°

C2/m 8 (100)θ║ (001)γ

[010]θ║[110]γ


48

Overall, these transient alumina phases are all based on a defective spinel structure. The

transient alumina polymorphs can be described as a fcc array of oxygen anions with aluminium

cations partially filling the tetrahedral and octahedral interstices [91]. For these alumina phases

with the defective spinel structure, the ratio of octahedral to tetrahedral sites occupied by Al3+

decreases from γ-Al2O3 (the ratio is 2) until θ-Al2O3 (the ratio is 1). In other words, the

transformations within the transient regime can be described in terms of a change in site

occupancy of cations [91]. For example, some researchers [91] have applied electron

diffraction to study the alumina transitions and concluded that both γ-Al2O3 and δ-Al2O3 were

based on the fcc packing of the oxygen anions but with a higher degree of order for the

interstitial cations in the δ phase. The stable α-Al2O3 is trigonal symmetry with rhombohedral

centring (space group R3c). The oxygen anions are hexagonal close packing (hcp) and the

aluminium cations occupy the octahedral sites in the anion sublattice [92].

The investigations considering the early stage oxidation of alumina-forming alloys are

intensively focused on the θ-Al2O3 formation, its transformation to α-Al2O3 and the effect of

additives on this phase transformation. For example, Prasanna et al. [88] have studied the effect

of θ-Al2O3 formation on the growth kinetics of FeCrAlY alloys during oxidation. They

demonstrated that θ-Al2O3 formation can significantly enhance the oxidation rate of alumina-

forming alloys. Furthermore, they have utilised a two-stage oxidation test which was performed

in 𝑂16 2 / 𝑂18 2 gas at 900 ℃ and secondary ion mass spectropy (SIMS) analysis to elucidate the

θ-Al2O3 to α-Al2O3 transformation process. The needle-shaped θ-Al2O3 (Fig. 2.11 [88]), which

grew by outward diffusion of aluminium, transformed to equiaxed small grains of α-Al2O3

which grew by inward diffusion of oxygen after early stage oxidation. Yang et al. [92] have

investigated the transient oxidation stage of single crystal (001) NiAl alloys by the electron

diffraction analysis and found that α-Al2O3 nucleated at the oxide/alloy interface with random


49

orientations. Other studies have investigated the transient oxide formation on NiAl alloys by a

variety of techniques including scanning electron microscopy (SEM) [53, 89, 91], transmission

electron microscopy (TEM) [90, 93], PLPS [94] and thermogravimetric analysis (TGA) [89,

95, 96].

The effect of different additives on the θ-Al2O3 to α-Al2O3 phase transformation is also of

interest because different additives can retard or accelerate this transformation thus affecting

the oxide growth kinetics. For instance, hafnium (Hf) can slow down the θ-Al2O3 to α-Al2O3

transformation and result in a consistent fast growth of oxide scale during the transient

oxidation stage of Pt-modified γ’-Ni3Al-based alloys [96]. Conversely, Brumm and Grabke [89]

concluded that Cr addition in the Ni-Al alloys can accelerate the θ-Al2O3 to α-Al2O3

transformation. This Cr effect is explained by the Cr2O3 nuclei formation in the initial stage of

oxidation which serves as nucleation sites for α-Al2O3, causing a faster transition rate. The

yttrium (Y) effect is more complex. According to Jedlinski [97], Y can accelerate or retard the

transient alumina to stable α-Al2O3 transformation depending on the amount and form of

yttrium in the alloy as well as its form in the oxide scale. Specifically, small amount of Y

accelerates the phase transition by provision of heterogeneous nucleation sites. While higher

amount of Y can be incorporated into the lattice of transient alumina by forming mixed Y-Al

oxides such as Y3Al5O12, and retard their transition into stable α-Al2O3. As for the Pt addition,

controversies exist regarding its effect on this phase transformation [94, 98, 99]. Unlike Cr or

Y, Pt is inert and cannot form any oxide or dope into the oxide scale in ion form. This suggests

that other mechanisms should be responsible for the observed Pt effects on this transformation.


50

Fig. 2.11 The morphology of oxide scale on the Fe-20Cr-Al alloy after 72 h oxidation at 900 ℃.

Note the needle or plate shape of θ-Al2O3 [88].

In summary, different additions of alumina-forming elements can affect the transient alumina

to stable α-Al2O3 transition, which in turn has an influence on the oxide growth kinetics.

However, reliable mechanisms to explain the effect of additives are still demanding mostly due

to lack of microscopic information of the transient oxide distribution and α-Al2O3 nucleation.

2.5 Superalloy substrate

The Ni-based superalloys have been applied as the structural materials of aircraft and power-

generation turbine blades over the last few decades. After years of alloy development, the state-

of-the-art Ni-based superalloys can tolerate average temperatures ~ 1050 ℃ with occasional

local hot spots to temperatures ~ 1200 ℃, which is ~ 90% of the melting point of the Ni alloy


51

[100]. The following sections will briefly review the composition, microstructure and

characteristic properties of the Ni-based superalloys.

2.5.1 Composition and microstructure

The nickel-aluminium binary system is the basis for the Ni-based superalloy composition for

gas-turbine applications. Apart from this, at least five to ten alloying elements are also added

to the superalloy which can account for up to 40 wt. %. Typical alloying elements of the Ni-

based superalloys are given in Fig. 2.12 [101].

The fcc γ-phase is the basic constitute of the Ni-based superalloy. With increasing aluminium

contents, a precipitate γ’-Ni3Al phase forms, which has an ordered intermetallic L12 structure.

These two phases are the major constitutes of the Ni-based superalloy. Fig. 2.13 [100] shows

the microstructure of a Ni-based single crystal superalloy with a high volume fraction of

cuboidal γ’ precipitates in the grid-shape γ matrix. Single crystal Ni-based superalloys

including CMSX-4, SRR 99 and René N5 all show this typical microstructure with different

minor element additions, which will be illustrated below.

Fig. 2.12 Alloying elements in the Ni-based superalloys (adapted from [101]).


52

Fig. 2.13 Microstructure of a Ni-based single crystal superalloy revealing a high volume

fraction of γ’ phase [100]: the cuboidal γ’ precipitates (grey contrast) in the γ-matrix (white

contrast).

Some alloying elements, such as W, Mo and Re are solid solution strengtheners of the γ-phase.

While other alloying elements including Ti, Ta and Nb can strengthen the γ’-phase by forming

the γ’- Ni3(Al, Ti, Ta, Nb) intermetallic compound [102]. For example, both the CMSX-4 and

SRR 99 single crystal superalloy contain Ti as strengtheners, while the René N5 superalloy

excludes this strengthener. Furthermore, Cr can improve the corrosion resistant of the

superalloy and Y can contribute to the oxidation resistance of the superalloy [103]. The trace

elements, e.g. C, B, Zr and Hf can form carbides or borides which are often located at the grain

boundaries. These elements are added to control the grain structure thus affecting the

mechanical properties that are strongly influenced by the grain boundaries [100].

2.5.2 Physical and mechanical properties

A large fraction of the structural components in gas-turbine engines are made of the Ni-based

superalloys for the sake of their exceptional combination of physical and mechanical properties.

Table 2.2 lists typical physical properties of Ni-based superalloys. In practical applications, it


53

is usually worth considering the density normalised properties, especially for the rotating

components in the gas-turbine engines.

Table 2-3 Physical properties of Ni-based superalloys

Properties Typical ranges

Density 7.7 - 9.1 g/cm3 [100]

Melting point 1320 - 1450 ℃

Thermal conductivity 9 - 11 W/(m·K) (RT)

CTE 12 - 18 ×10-6 /℃ [104]

Ni-based superalloys exhibit relatively high yield tensile strength (~ 900 - 1300 MPa at room

temperatures). However, the tensile properties show a significant decay at temperatures above

850 ℃. The single crystal CMSX-3 superalloy exhibits a yield strength ~ 200 MPa at

temperatures of 1000 ℃ [105]. Modern Ni-based superalloys are optimised to improve the high

temperature creep resistance, which is crucial because during service the superalloys are under

stress for extended periods at high temperatures. Creep properties are influenced by the

alloying elements and microstructure of the superalloys. For example, the additions of

refractory elements Re, W and Mo have been successful in improving the creep resistance of

Ni-based superalloys. However, further incorporation of these refractory elements is limited

due to the formation of topologically close packed (TCP) phases at elevated temperatures,

which is associated with the initiation of creep damage. New generation superalloys have

incorporated Ru to suppress TCP formation resulting in enhanced creep properties [106].

2.6 The degradation and failure of TBCs

The core value of applying the YSZ top coat is to protect the underlying superalloy substrates

from the high temperature environment. Any degradation (erosion) or failure (spalling away

from the superalloy) of the top coat will lead to direct exposure of the superalloy to the hot


54

gases with temperatures over its melting point, which is catastrophic. Therefore, establishing

reliable lifetime models and appropriately evaluating the TBC lifetime are very important for

the development of prime-reliant TBCs. The prerequisite for establishing reliable lifetime

models is to gain a comprehensive understanding of the failure mechanisms for various TBC

systems under different exposure conditions.

In general, TBC systems exhibit various failure mechanisms, and some well-accepted ones are

listed in Fig. 2.14 [74]. Based on the nature of these mechanisms, they can be summarised into

intrinsic and extrinsic mechanisms. The former is associated with thermal processes such as

elemental interdiffusion (i and ii in Fig. 2.14), TGO growth (iii in Fig. 2.14) and bond coat

deformation (iv). The extrinsic mechanism refers to the environmental degradation including

foreign object damage (FOD, v) and CMAS (calcium-magnesium-alumino-silicate) attack.

Specifically, in some cases, the alternative oxides such as spinels form either between the TGO

(α-Al2O3) and the bond coat due to Al depletion (i) or between the top coat and the TGO

because of the Ni diffusion through the TGO (ii) [10]. The brittle spinel can result in the

interfacial delamination. Alternatively, some failures are dominated by the strain energy

density of the oxide scale, and its interplay with the imperfections in the vicinity (non-planar

interface, iv) or within the TGO (planar interface with imperfections, iii) [6]. This failure

mechanism will be emphasized in this study and will be elaborated in Section. 2.6.1. In other

cases, the particle impact (FOD) may locally compress the porous top coat, leading to local hot

spots which can accelerate oxidation and contribute to the failure process [107]. Since the

extrinsic failure mechanism will not be involved in the next chapters, the following discussions

will only focus on the intrinsic mechanisms which will be organised as follows: firstly, a brief

introduction of the general failure modes of TBCs, followed by the failure mechanisms for

specific TBC systems under different exposure conditions.


55

Fig. 2.14 Five major categories of failure mechanisms documented for TBC systems [74].

2.6.1 General failure modes

As mentioned above, the TGO develops a large compressive stress when cooling from the

service temperature. This stress in the thin TGO layer can cause the undulation (morphological

instability) of the scale, which can result in the failure of the system. This failure process can

be envisioned as a sequence of the crack nucleation, propagation and coalescence events [74,

108, 109]. First of all, tensile stresses which are normal to the bond coat/TGO interface are


56

induced as a result of the TGO imperfections. This process is associated with an strain energy

release rate, thus initiating small cracks in the vicinity [6]. Then the stresses around the

imperfections and the associated energy release rates govern the propagation of these small

cracks or separations. The TBC remains attached at the remnant ligaments during the crack

propagation process. Finally, when cracks from the nearby TGO imperfections coalesce and

the remnant ligaments are detached, the TBC spalls either by edge delamination or large-scale

buckling (LSB) [74].

The above-mentioned two competing failure mechanisms: edge and buckle driven

delamination, both have been documented in previous studies [6, 110-112]. In this work the

emphasis will be on the buckling driven delamination, which is prevalent for TBC failures. Fig.

2.15 [3] shows a typical buckling driven failure of an EBPVD TBC system, which is common

for the compressive thin films. Consider a thin film subject to an equi-biaxial compressive

stress state, σ0, which is given by [110]:

𝜎0 = 𝐸∆𝛼∆𝑇/(1− 𝜈) (2.1)

where ∆𝑇 is the temperature drop from which the film is stress-free. E is the young’s modulus

of the film and ν is Poisson’s ratio. ∆𝛼 is the difference between the CTE of the substrate

(having a higher CTE, thus ∆𝛼 > 0) and the thin film. A buckling index is then defined as

[110]:

𝛱 = (1 − 𝜈2)(𝜎0/𝐸)(𝐿/ℎ)2 (2.2)

where h is the film thickness and L is the size of the separation which is present at the

film/substrate interface. When Π exceeds a critical value Πc (Πc =4.89 for a circular buckle

[110]), the buckling can be initiated. Equating Π=Πc =4.89 into Eq. (2.2), a critical separation

size Lb at the onset of buckling can be expressed as [110]:

𝐿𝑏/ℎ = 2.21√𝐸/𝜎0 (2.3)


57

where 𝐸 is the plane strain modulus of the film, 𝐸 = 𝐸/(1 − 𝜈2). Here, 𝐿𝑏 represents the

smallest separation size to initiate a buckle in absence of any TGO imperfection. By

substituting typical 𝐸 and 𝜎0 values into Eq. (2.3), Lb is ~ 20 ℎ in magnitude, which is about

several mm when considering a film thickness of ~ 100 μm. This relatively large size of flaw

can hardly be seen in the as-deposited TBCs. Hence, the initiation and growth of small

separations must occur until these separations reach the critical size needed for LSB. In addition,

the calculation of energy release rates of separations that are induced by the imperfections of

thin films have validated that TGO imperfections (e.g. undulations or heterogeneities) can

decrease the critical flaw size for the buckle initiation [110].

Fig. 2.15 The optical images of a EBPVD TBC sample showing a) the incipient buckling of

the top coat (viewed under reflected light) and b) subsequent spallation of the top coat [3].

In brief, the general failure mode for a TBC system includes a sequence of crack initiation,

growth & coalescence, and buckle driven delamination events. The TGO heterogeneities and

undulations play a key role in the failure process by re-distributing the stress in the vicinity of

the TGO imperfections, thus facilitating the above sequence of events. APS TBCs and EBPVD

TBCs are totally different in their microstructure and thermophysical properties, so different


58

failure processes apply respectively [113]. The following sections will discuss the failure

process for the two TBC systems in detail.

2.6.2 The failure mechanisms of APS TBCs

As described in Section 2.2.2, YSZ top coats deposited by APS have inter-splat pores and

cracks which are parallel to the substrate/coating interface. In addition, since the bond coat

surface is roughened (the roughness average Ra can be tens of micro) before the deposition of

the YSZ layer to enhance the mechanical bonding, the metal/ceramic coating interface is highly

undulated. Because of the complex microstructure, the mechanics of events which proceeds

the edge- or buckle-driven delamination is still not fully understood [113]. The undulations at

the bond coat/TGO/top coat interfaces can give rise to out-of-plane tensile stresses which are

normal to the metal/ceramic interface. These stresses, combined with the interface

imperfections lead to the failure of the APS TBC system. Fig. 2.16 [11] gives a schematic

illustration of four primary cracking modes documented for APS TBCs. Thermal mismatch

stresses are developed at the TGO/bond coat interface upon cooling, and the interface

undulations can redistribute the thermal misfit stresses: the tensile stresses at the undulation

crests and the compressive stresses at the troughs [114]. The out-of-plane tensile stresses

increase as the TGO thickens, causing the crack initiation at the bond coat/TGO interface (type

Ⅰ crack in Fig. 2.16) of the crest. The type Ⅱ crack is initiated at the top coat/TGO interface of

the crest, and type Ⅲ corresponds to the cracking within the top coat in the vicinity of the crest.

These types of cracks are also resulted from the out-of-plane tensile stresses in the vicinity of

the TGO/top coat interface. The mechanism for type IV crack in Fig. 2.16 is more complicated.

When the TGO scale is very thin, the thermal mismatch stress between the bond coat (BC) and

the top coat (tbc) (determined by the CTE difference, αBC - αtbc >0) results in out-of-plane

tension in the top coat above the undulation crests and compression in the troughs. As the TGO

thickens, it constitutes a good fraction of the bond coat asperity, therefore the thermal mismatch


59

stress is locally dominated by the CTE mismatch between the bond coat/TGO ‘composite’ and

the top coat, rather than just the bond coat [115]. Beyond a certain critical TGO thickness, the

CTE of the bond coat/TGO ‘composite’ becomes lower than that of the top coat (αtbc), thereby

reversing the stress in the YSZ undulation troughs from compression to tension [11]. This

tensile stress drives the cracking within the troughs of the YSZ layer as shown by type IV.

Fig. 2.16 A schematic illustration of four primary cracking modes in an APS TBC system [11].

Lastly, the thermal treatment can affect the cracking mode for the APS TBCs [116]. According

to Trunova et al. [60], the isothermal heat treatment tends to promote crack propagation within

the TGO, while cyclic exposure can shift the crack path towards the top coat. Fig. 2.17 [60]

shows examples of different crack paths caused by the isothermal and cyclic oxidation.

Fig. 2.17 SEM micrographs shows the damage evolution in an APS TBC: a) isothermal

oxidation and b) thermal cycling [60].


60

2.6.3 The failure mechanisms of EB-PVD TBCs

Unlike the APS TBCs, the crack initiation of EBPVD TBC systems is mainly at the TGO/top

coat interfaces or the TGO/bond coat interfaces, rather than inside the TGO. Here, the failure

mechanisms of EB-PVD TBCs will be addressed in terms of different bond coat types.

The top coat deposited by EBPVD usually applies to the MCrAlY or β-NiPtAl bond coat on a

relatively flat surface (Ra is about a few micros). However, during prolonged thermal exposures,

especially thermal cycling, the bond coat surface is gradually roughened, due to the progressive

displacement of TGO into the bond coat, as shown in Fig. 2.18 [117]. This undulation that

develops in the TGO layer, also termed as rumpling, ratcheting or undulation instability, has

attracted considerable interest as a specific form of TBC failure with the MCrAlY or β-NiPtAl

bond coats [118-121]. Because the rumpling growth can generate tensile stresses across the top

coat/TGO/bond coat interfaces. These tensile stresses can initiate interfacial cracking between

the top coat and the superalloy substrate, leading to LSB and spallation failure of the TBC

[121].

Previous studies have extensively investigated the rumpling of TGO on the MCrAlY or β-

NiPtAl bond coats by experiments or simulations in the last decades [118, 121-124]. Among

these, Tolpygo and Clarke [121-123] have carried out systematic studies to find out the

mechanisms behind this rumpling behaviour. By conducting a series of variable-controlling

experiments on the surface roughness evolution of the nickel-aluminide bond coats, they have

ruled out some previously suggested factors that contribute to the rumpling and revealed that

the rumpling is mainly driven by a combination of TGO lateral growth and the thermal

mismatch between the coating and the underlying superalloy. For instance, they have

conducted thermal cycling above the martensitic transformation temperature Ms of the β-


61

NiPtAl bond coat and observed the same roughness evolution compared to the thermal cycles

with martensitic transformation, which excluded the influence of martensitic transformation on

the rumpling [122]. They also pointed out that volume changes in the bond coat (due to Al

depletion induced phase transformation, e. g. β-NiAl to γ’-Ni3Al) are not necessary for the

rumpling initiation. Because the rumpling initiation can take place at very beginning of

oxidation before the phase transformation (due to Al depletion) occurs [121].

Fig. 2.18 The cross-sectional SEM micrograph of an EBPVD TBC on a β-NiPtAl bond coat

exhibiting the TGO rumpling after 50 1-h cycles at 1150 ℃ [117].

A mechanistic model (Balint & Hutchinson, B&H model [125]) has been developed to simulate

the rumpling behaviour of the TGO layer under various exposure conditions. This model can

provide an analytical approximation of the rumpling growth by considering factors including

TGO thickening and high temperature yielding of the TGO. The fundamental idea of this model

is that the rumpling deformation is driven by the lateral growth strain of TGO and assisted by

concurrent creep of the bond coat in compliance with the TGO deformation. The predictions

of rumpling growth on systems with the top coat and without the top coat have been presented

using this model [125]. For the TBC system with the top coat, more than a ten-fold reduction

of the total rumpling growth is predicted compared to that without the top coat. This prediction


62

agrees well with the observations in the literature. For example, it is observed that the rumpling

is exclusively in the form of downward displacement of the TGO layer into the bond coat,

rather than both upward and downward displacements of TGO when no top coat is present

[126]. Because the top coat can significantly restrain the upward displacements of the TGO

and reduce the total rumpling growth. Furthermore, Balint and Hutchinson [125] have applied

this model to explore several important effects on the rumpling growth such as TGO thickness

effect and thermal history effect. The B&H model predicts more rumpling for the thicker oxide

layers, which is consistent with the experimental results by Tolpygo and Clarke (Fig. 2.19

[127]).

In addition to the TGO rumpling induced by thermal cycling, the MCrAlY bond coat has

another form of TGO imperfection which can also initiate cracking - the TGO thickness

heterogeneity due to the growth of yttrium-rich oxides (such as Y2O3 or yttrium aluminium

garnet, YAG) [128]. The TGO heterogeneities, also termed as ‘pegs’, can be developed and

enlarged in regions where the yttrium-rich oxides form which have a much higher oxygen

diffusivity than that of α-Al2O3. When this TGO heterogeneity develops to a critical size,

tensile stresses are generated around it followed by interfacial separations [74]. However, the

influence of this TGO heterogeneity on TBC failure is in debate. Other studies suggested that

these pegs can mechanically anchor the TGO layer in the bond coat, thus increasing the

interfacial adhesion and contributing to TBC lifetimes [128, 129].


63

Fig. 2.19 NiPtAl specimens show the effect of the oxide thickness on the rumpling. The oxide

layer thickness is a) ~ 5 μm and b) ~ 10 μm. The systems were subjected to the same thermal

cycling history and obviously more rumpling developed for the specimen with thicker oxide

layer [127].

Unlike the MCrAlY or β-NiPtAl bond coat, the EBPVD TBCs based on the Pt-diffused γ/γ’

bond coats are not suspected to develop significant rumpling during cyclic exposure [86]. In

addition, the TGO on this bond coat did not exhibit thickness heterogeneity due to the

formation of fast-growing oxides [130]. It has been widely observed that for this TBC system,

the failure occurs mainly at the TGO/bond coat interface. Zhao and Xiao [131] have proposed

that during oxidation, impurities such as sulphur and refractory elements segregate to the bond

coat/TGO interface and degrade the TGO adhesion, which causes failure of the TBC. Other

researchers have found that the undulations of the initial non-planar bond coat surface can lead

to local interface separations due to the normal tensile stress across the interface [130]. These

separations gradually coalescence, causing spallation failure of the system. In summary, the

failure mechanism of EBPVD TBCs with the Pt-diffused γ/γ’ bond coat is still not fully

understood and remains to be further investigated.


64

2.6.4 Interface toughness measurement of TBCs

The above sections have summarised different factors that can lead to crack initiation for the

APS TBC and the EBPVD TBC systems, respectively. In general, the lifetime of TBCs is

determined by the interaction between the driving force for delamination (the energy release

by crack initiation) and the resistance to crack propagation (toughness) [132]. Therefore, in

addition to the driving force of failure, the resistance force, i.e. the interface toughness is also

critical for understanding the failure mechanisms and establishing reliable lifetime models of

TBCs. This section will firstly introduce the definition of interface adhesion & delamination

and interfacial toughness, followed by some general failure modes for a thin film system.

Finally, some widely-used approaches to investigate the interface toughness for a thin ceramic

film on the metal substrate will be reviewed, which might be applied to the TBC systems.

2.6.4.1 Definitions of interface adhesion & delamination and interface toughness

The interface adhesion can be defined as: ‘the state in which two surfaces are held together by

interfacial forces including electrostatic forces, Van der Waals forces or chemical bonding’

[133]. Delamination, on the other hand, is the phenomenon that a coating separates from the

underlying substrate, which is driven by external or internal stresses, corrosion, etc.

The true work of adhesion for the interface is the amount of energy required to create free

surfaces from the bonded layers [134]:

𝑊𝐴 = 𝛾𝑓 + 𝛾𝑠+ 𝛾𝑓𝑠 (2.4)

where 𝛾𝑓 and 𝛾𝑠 are the specific surface free energies of the film and the substrate, respectively.

𝛾𝑓𝑠 is the specific free energy of the interface. The true work of adhesion is an intrinsic property

of the interface and can be determined by contact-angle experiments [135]. However, in

practice, the delamination is usually associated with energy dissipation due to different


65

mechanisms (e. g. contamination, plastic deformation, etc.) during most of the adhesion tests.

Thus, the true work of adhesion is difficult to extract from the measurements. Instead, a

practical work of adhesion or the so-called critical strain energy release rate is defined as [133]:

𝐺𝑖𝑛𝑡 = − 𝜕𝑈

𝜕𝐴 (2.5)

where U is stored elastic energy released and A is the area of the interfacial crack. Then the

interface toughness, Kint can be defined by 𝐾𝑖𝑛𝑡 = √𝐸𝑖𝑛𝑡𝐺𝑖𝑛𝑡, where 𝐸𝑖𝑛𝑡 is a representative

Young’s modulus for the coating/substrate system.

The above definitions of the interface delamination and the critical strain energy release rate

provide the basic theories for interface toughness measurements. Then some general modes of

failure are introduced prior to the introduction of interface toughness tests. Fig. 2.20 outlines

four common failure types of a coating system [133]:

(a) A single through-thickness crack in the coating, and it propagates to the interface and

induces the coating delamination [136];

(b) A periodic array of through-thickness cracks which divert to the interface and induce the

failure [137, 138];

(c) Crack is generated at the interface and propagates along the interface [110];

(d) For a compressed coating, an initial crack at the interface can grow and lead to the buckling-

driven failure of the coating [132].


66

Fig. 2.20 Schematic of different failure modes for a thin coating system. a) A single through-

thickness crack in the coating which deflects to the interface to cause the coating failure; b)

multiple through-thickness cracks in the coating; c) edge-delamination at the interface and d)

buckling-induced delamination for a compressed film [133].

Consider a well-developed interfacial defect of radius c, from a simple fracture mechanics

approach, the interface toughness 𝐾𝑖𝑛𝑡 is given by:

𝐾𝑖𝑛𝑡 = 𝑎𝜎√𝜋𝑐 (2.6)

where 𝑎 is the geometry constant which is ~ 1 for a circular defect. 𝜎 is the stress to cause the

coating delamination.

The above-mentioned energy dissipation during the delamination process depends on the mode

mixty angle (phase angle 𝛹), a measure of the relative amount of shear and normal stress

components at the crack tip [134], given by:

𝛹 = tan−1(𝐾Ⅱ/𝐾Ⅰ) (2.7)

where 𝐾Ⅱ and 𝐾Ⅰ are the interface toughness of mode Ⅱ failure (pure shear, Ψ = 90°) and mode

Ⅰ failure (pure opening fracture Ψ = 0°), respectively. The amount of energy dissipation

increases as the phase angle increases from 0° to 90°. The interface fracture toughness as a


67

function of the mode mixty angle 𝛹 has been proposed by several studies, and the widely-used

one is defined by Hutchinson and Suo [139]:

𝛤(𝛹) = 𝛤0{1+ 𝑡𝑎𝑛2[(1 − 𝜆) ψ]} (2.8)

where 𝛤0 is the mode Ⅰ interface toughness for Ψ = 0°. λ is a parameter ranging from 0 to 1,

which is dependent on the interfacial friction.

2.6.4.2 Interface toughness test methods

In this section, several common tests to determine the interface toughness will be briefly

reviewed, including the indentation tests, pull-off tests, pushout methods, micro-cantilever tests

and four-point bending tests. The focuses for each method will be the experimental setup, the

strengths & limitations and its applications to the TBC systems.

Indentation tests. Indentation is the most common technique to measure the mechanical

properties of hard coatings such as Young’s modulus and hardness. It can also be used to model

the fracture behaviour, thus measuring the interface fracture toughness. Especially for thin

films, indentation is an effective method for sampling a small area of the interface with

common laboratory equipment for toughness measurement [140, 141]. Different indenters can

be selected according to the coating systems and the testing conditions, such as spherical

indenter (blunt), Vickers indenter (sharp), etc. The fracture modes are determined by the

indenter geometry and material properties. There are five common fracture modes for brittle

films including the cone cracks, Palmqvist radial cracks, median cracks, lateral cracks and the

half-penny radial cracks [142]. The corresponding fracture toughness analysis for each fracture

mode has been well established by the stress-based approach [143, 144] or the energy-based

approach [138, 142, 145]. Although the indentation test is featured by its experimental

simplicity and no requirement for the sample geometry, only limited work has been done to

test the toughness of TBCs by this method [132, 143, 146]. This can be attributed to the facts


68

that indentation is highly dependent on the plastic deformation behaviour of the substrate and

the pores/cracks in the YSZ layer, both of which are hard to determine.

Barb pushout methods. The barb pushout method has been successful in evaluating the

interface toughness of EBPVD and APS TBCs [147]. The experimental setup is built on a ‘barb’

geometry, which originates from the pushout test for the fibre reinforced ceramic/metal matrix

composites [148]. The TBC top coat layer is notched at a distance of several mm from the end

of the specimen, and the remaining segment of the top coat is carefully removed by polishing.

Then two specimen pieces like this are affixed back-to-back, to form a ‘T-bone’ shape

specimen for the barb pushout test. This specimen is supported by two blocks during the

pushout test and the force-displacement curve is recorded. Then the steady-state energy release

rate can be simply estimated by the elastic properties of the YSZ top coat and the substrate,

combined with the maximum load and specimen dimensions. This methodology was initially

developed to analyse the crack developing process, where the crack propagation occurs

predominantly under the pure shear (mode Ⅱ) loading [149]. Liu et al. [150] have further

determined the phase angle of this loading condition to be ~ 65 - 70°. The benefit of this test is

that it is not sensitive to the anisotropic structure of YSZ top coats. While the main

disadvantage of this technique is the relatively complex sample preparation process.

Micro-cantilever tests. This method has been regarded as a particularly useful tool to measure

the mechanical properties and the interface toughness of coating systems with anisotropic

materials [151]. The micro-cantilever can be made by the focused ion beam (FIB) which is

usually coupled with SEM. Then the nanoindentation can be applied on the micro-cantilever

to deflect it until it snaps. The interface toughness can be expressed in terms of the maximum

applied load, the size of the notched crack and the dimensions of the cantilever beam based on


69

the simple beam theory [152]. This method has been successfully applied to measure the

interface toughness of various ceramic coatings on the metallic substrates including the

SiO2/copper system [152], TaN/copper system [153] and the APS TBC system [154]. The main

disadvantage of this technique is the tedious sample preparation process because it requires a

symmetric cross-section of the cantilever beam with accurate dimensions to meet the simple

beam theory conditions.

Pull-off methods. There are two types of pull-off tests, (i) the tape peel test [141] and (ii) the

tangential lap shear test [155]. The tape peel test uses a piece of pressure-sensitive tape to peel

off the film. By measuring the force at which the film is removed, the peeling energy can be

obtained. This is a simple method but it can only be applied to the weak bonding coatings such

as polymers. In the tangential lap shear test, the coating is pulled off by an adhesively mounted

rod with a tensile force. This test can measure harder coatings. However, a perfect alignment

to ensure uniform loading across the coating/substrate interface is challenging for some

systems, especially those with special specimen shapes [133]. Thus, so far this method has

limited applications for evaluating the TBC systems.

Four-point bending. In this technique, a notch must be machined through the coating and

symmetric interfacial precracks are generated. Then the notched bending beam specimen is

monitored during the onset of the delamination with increasing load, and the interface

toughness can be simply calculated provided that there is no plastic deformation. For some

coating systems, the plastic deformation of the substrate can occur. So a stiffening layer is

attached to the coating system in order to avoid the plastic deformation, which is so-called

modified four-point bending test [156, 157]. The four-point bending test has been widely

utilised to test the interface toughness of ceramic coatings on softer substrates. However, the


70

application of this technique is limited due to the following reasons. First, it requires special

specimen geometry and the pre-cracking can be difficult for some systems which involves

complex specimen preparation [141]. Second, in practice, the discontinuous crack development

of a non-planar crack front can occur, which diverts the failure from the computable mode.

Thus, only limited work has been done by this technique for the TBC systems. For instance,

Zhao et al. [157] have reported the energy release rate to be ~ 50 J/m-2 for the as-sprayed APS

TBCs by the modified four-point bending test.

To sum up, these tests all involve inducing coating delamination in a controlled process to

allow for a quantitative analysis for the interface toughness. The measured toughness values

can be affected by test-specific factors and the residual stress of the film [158]. The interface

toughness is a critical parameter for the TBC applications, which requires the analysis of failure

mechanisms at specific test conditions and appropriate models to assess it. Currently, direct

measurements of the interface toughness in complete TBC systems are still sparse because of

the complex shape and the anisotropy of the TBC systems. Further studies are needed to

compensate the current data base on this important issue.

2.7 Summary

In this chapter, the components of the multi-layered and multifunctional TBC system have been

reviewed regarding material requirements, microstructure and properties. The failure

mechanisms and the interface toughness measurement of different TBC systems were also

discussed. The establishment of reliable lifetime models of TBCs requires a more

comprehensive understanding of the failure mechanisms of TBCs that involve material

structures and properties. In addition, further research and development (R&D) including new

coating design to improve the durability of TBCs is still crucial to meet the growing demands

in the gas turbine engine industry.

CHAPTER 3 PT EFFECT ON EARLY STAGE OXIDATION BEHAVIOUR OF γ/γ’ COATINGS

71

Chapter 3 Pt Effect on Early Stage Oxidation Behaviour of Pt-

diffused γ-Ni/γ’-Ni3Al Coatings

3.1 Introduction

The Ni-Al-based coating, which is commonly used for oxidation protection of the Ni-base

superalloys, is designed to develop a protective Al2O3 (thermally grown oxide, TGO) layer

during high temperature exposure. This coating fails when the TGO spalls off, which is driven

by the increasing TGO strain energy as the oxide thickens. Thus, the lifetime of the coating

usually depends on the spallation resistance of the TGO layer. Numerous studies have reported

that the addition of platinum (Pt) can significantly improve TGO spallation resistance of Ni-

Al-based alloys or coatings for long-term oxidation at elevated temperatures. For instance, Y.

Chen et al. [159] concluded that Pt addition can improve the oxide spallation resistance of the

γ/γ’ nickel aluminide alloys mainly due to the selective oxidation of aluminium promoted by

Pt. P. Y. Hou and V. K. Tolpygo [44] found that 5 - 8 at.% Pt can significantly enhance the

oxide spallation resistance of nickel aluminide coatings during cyclic oxidation. They pointed

out that Pt can prevent the segregation of impurities (e.g. sulfur) at the oxide/coating interface,

resulting in enhanced interfacial adhesion and improved TGO lifetime. Other researchers [45,

48, 68, 160] have reported this benefit of Pt caused by other mechanisms such as inhibit ing

void formation at the coating/oxide interface.

It is widely accepted that for Al2O3-scale forming Ni-Al-based coatings, one (usually θ) or

more transient forms of Al2O3 initially form during the early stage oxidation, followed by the

eventual transformation to stable α-Al2O3 [89, 91, 161]. Previous studies have suggested that

the early stage oxidation behaviour can affect the prolonged oxidation behaviour of the Ni-Al


72

coating. For example, V. K. Tolpygo and co-workers [162] found that different transient

alumina to stable alumina transition rate during early stage oxidation leads to local TGO

thickness heterogeneity, which can induce local tensile stresses and is detrimental to TGO

lifetime for prolonged oxidation. Therefore, it is important to investigate the early stage

oxidation behaviour for a reliable estimation of TGO performance. However, despite the large

number of studies about Pt effect on long-term oxidation as mentioned above, only limited

attempts [53, 94, 98, 163] have focused on the impact of Pt on early stage of oxidation

behaviour of Ni-Al alloys or coatings. In addition, some controversial views regarding the Pt

effect on the early stage θ-Al2O3 to α-Al2O3 transformation have been reported. For instance,

J. Jedlinski et al. [94] found that Pt resulted in an earlier development of α-Al2O3 (faster θ- to

α-alumina transition) for a β-NiPtAl alloy at 1100°C oxidation, whereas Y. Cadoret, et al. [98]

concluded that Pt can slow down the θ-Al2O3 to α-Al2O3 transition by studying the early stage

oxidation of Ni50Al50 and Ni40Pt10Al50 alloys at 900 and 1100°C, respectively. H. Svensson and

co-workers [163] concluded that Pt has no effect on this phase transformation by studying the

initial oxidation of β-NiPtAl alloys. However, for the Pt-diffused γ-Ni/γ’-Ni3Al coating system,

little work has been done on the Pt effect on the θ-Al2O3 to α-Al2O3 transformation. The only

related work to the best of our knowledge, suggested that a higher Pt content led to a larger

amount of θ-Al2O3 in the oxide scale by photo-luminescence piezo-spectroscopy (PLPS) [131].

This indicates a slower θ-Al2O3 to α-Al2O3 transformation because of Pt addition for this γ/γ’

coating system. However, no mechanism has been proposed in their work about this Pt effect

on the phase transformation rate, mainly due to the lack of microscopic information about the

distribution of the two alumina polytypes and their relation to the coating composition (Pt

contents).


73

In this chapter, the early stage oxidation behaviours of γ/γ’ Ni-Al coatings with different Pt

additions are studied and compared in terms of TGO microstructure, transient oxide to stable

α-Al2O3 transformation rate and scale stress evolution, to provide a more comprehensive

understanding of the Pt effect on the early stage oxidation of this coating. Moreover, in order

to explore the mechanism of the Pt effect on alumina phase transition, the phase mapping of

the alumina polytypes (transient θ-Al2O3 and stable α-Al2O3) in the scale during the early stage

was conducted by the ASTAR (NanoMEGAS) automated crystal orientation mapping on

transmission electron microscopy (TEM) [164]. This technique offers significantly improved

lateral spatial resolution over the recently developed transmission electron backscatter

diffraction (t-EBSD) [165], which enables the diffraction pattern indexed phase mapping of

both θ-Al2O3 and α-Al2O3 during early stage oxidation for the first time, to the best of our

knowledge. A new transformation mechanism has been proposed based on these results to

explain the different transformation rate observed on coatings with different Pt contents.

3.2 Experimental procedures

3.2.1 Sample preparation and thermal treatment

CMSX-4 single crystal Ni-based superalloy (Table 3.1, Rolls-Royce plc) was used as substrates

throughout this study. The as-received superalloy bars were cut into buttons (20 mm diameter

and 3 mm height) using a SiC cutting blade in a precision cut-off machine (Accutom 5, Struers).

All substrates were ground and polished to 1 µm finish and then washed in acetone before Pt

electroplating. Platinum was electroplated on the substrate surface by using Q salt

(Tetraammineplatinum(II) hydrogen phosphate, Johnson Matthey) (See Table 3.2). The anode

was the platinum foil and the cathode was the CMSX-4 substrate. The electroplating

temperature was maintained in the required range (91~93 °C) for sufficient cathode efficiency.

The pH value was maintained in the recommended pH range (10.0-10.6) by regular addition


74

of ammonia solution during the electroplating process. In addition, the regular addition of

distilled water was necessary in order to maintain the volume of the bath. A magnetic stirrer

was applied in the bath to maintain a uniform Pt concentration of the bath. Three groups of

samples (each group contains three button samples) were fabricated, with electroplating time

0 min, 20 min and 50 min, respectively. All electroplated samples were washed in hot distilled

water (~80 ℃) for 0.5 h to remove the remaining salts from the electroplating bath. The

electroplated Pt layer thickness of each sample was examined by the focus ion beam system

(FIB; FEI Quanta 3D). Finally, the substrates with different electroplating time were annealed

in vacuum at 1150 °C for 2 h to obtain γ/γ’coatings with different Pt contents.

Table 3-1 Composition of CMSX-4 substrates

Element Ni Al Cr Co Ta Ti W Re

Wt. % 61.4 5.6 6.4 9.6 6.6 1.0 6.4 2.9

Table 3-2 Electroplating platinum bath

Chemical formula [Pt(NH3)4](HPO4)

Platinum content

pH

Temperature

Cathode current density

4~6 g/L

10.0~10.6 (optimum 10.5)

91~93 °C

2~7 mA/cm2

Isothermal oxidation of the γ/γ’ coatings with different Pt contents was performed at 1000ºC

in an automated rig (CMTM) in laboratory air. The samples were oxidized for different periods

of time up to 30 min, followed by air quenching.


75

3.2.2 Luminescence measurement and data processing

Phase identifications of alumina polytypes as well as measuring residual stresses in oxide scales

was carried out at room temperature using PLPS on a Renishaw Invia Raman system

(RenishawTM, Gloucestershire, UK) with an argon laser source (λ=633 nm). The laser spot size

was ~5 µm. Although according to the subsequent results, some of the oxide scales studied

here have a Ni-oxide over-layer above the alumina layer, it was shown by Lipkin and Clarke

[83] that this over-layer does not prevent either the excitation or collection of the fluorescence

signal from the Cr3+ in alumina scale.

Before each experiment, the spectrometer was calibrated by taking a spectrum from a standard

pure silicon sample. Cr3+ fluorescence spectra were collected for each measurement with one

second acquisition time. Measurements for each sample were taken on a square grid of 200×

200 µm with a pitch of 20 µm, thus one map contains 121 measurement points for each sample.

To determine the peak positions in each spectrum, all spectra were deconvoluted in Wire 4.2

software (RenishawTM) with an automatic fitting program by two mixed Gaussian-Lorentzian

functions [166]. It is well accepted that the R1 and R2 doublet at ~14400 cm-1 is produced by

Cr3+ fluorescence in α-Al2O3, and θ-Al2O3 produces luminescence peaks at ~14546 cm-1 (T1),

~14626 cm-1 (T2) and ~14330 cm-1 (T3) [86], as shown in Fig. 3.1 a. The relative intensities

of the α-Al2O3 and θ-Al2O3 luminescence lines provide a semi-quantitative indicator of the θ-

Al2O3 content in the spot area, which is given by [86]:

Cθ=A𝑇1+𝐴𝑇2+𝐴𝑇3

A𝑇1+𝐴𝑇2+𝐴𝑇3+𝐴𝑅1+𝐴𝑅2 (3.1)

where A𝑇1, 𝐴𝑇2 and 𝐴𝑇3 are peak areas for three characteristic peaks of θ-Al2O3, respectively,

and 𝐴𝑅1 and 𝐴𝑅2 are peak areas of the characteristic R1 and R2 peaks.


76

Since the R2 line has a nearly linear dependence on stress, the stress calculation was based on

the peak shift of the R2 line with respect to that of the unstrained single crystal sapphire (Fig.

3.1 b) by [84]:

∆𝜈 = 5.07(𝑐𝑚−1𝐺𝑃𝑎−1)σ (3.2)

where ∆𝜈 is the frequency shift of the R2 line and σ is the residual stress by assuming an

equibiaxial plane stress state in the scale.

Fig. 3.1 a) Luminescence spectrum showing characteristic peaks for θ-Al2O3 and α-Al2O3; b)

R peaks of an α-Al2O3 scale on a Pt-diffused γ/γ’ coating after isothermal oxidation at 1100°C

for 1 h. For comparison, the red line shows the spectrum of a stress-free polycrystalline alumina.

3.2.3 ASTAR automated crystal orientation mapping on TEM

The high resolution phase mapping of the alumina polytypes was performed using the ASTAR

automated crystal orientation mapping on TEM (transmission electron microscope, FEI Tecnai,

F30) [167], operating at 300 keV. FIB (FEI Quanta 3D) in-situ lift-out technique [168] was

used to prepare site-specific specimens for this analysis. Firstly, the electron beam scanned

over the area of interest on the FIB sample, and the digital image of diffraction patterns were

saved by a high-rate digital camera at each point with a spot size of 5 nm. The crystal

information for each phase present in the scanned area was extracted from crystal information

files (CIFs) produced from powder diffraction data [169-171]. Theoretical generated patterns

b) a)


77

(the templates) for each phase were then produced using the pattern generation method by

Zaefferer [172]. The acquired experimental patterns were automatically matched with the

generated templates by the so-called template matching process [173]. The degree of matching

is quantitatively given by the correlation index Q, given by:

𝑄 =∑ 𝑃(𝑥𝑖,𝑦𝑖)𝑇(𝑥𝑖,𝑦𝑖 )𝑚𝑖=1

√∑ 𝑃2(𝑥𝑖,𝑦𝑖)𝑚𝑖=1 √∑ 𝑇2(𝑥𝑖,𝑦𝑖 )

𝑚𝑖=1

(3.3)

where P(x, y) is the intensity function of a pattern; T(x, y) is the intensity function of the

corresponding template; x and y are bounded by the picture width and height respectively.

Furthermore the reliability index assesses the likelihood of the match being unique and is

calculated by [173]:

R=100(1-Q2/Q1) (3.4)

where Q1 and Q2 stand for the two highest values of the correlation indexes for one pattern.

For a practical point of view, a reliability >15 is considered to be a safe solution.

3.2.4 Other characterization methods

An optical profilometer (ContourGT, Bruker) was used to examine the surface roughness of

the as-fabricated coatings. Scanning electron microscopy (SEM, FEI Quanta 650) and an

optical microscope (Olympus BH2-UMA) were used to examine the coating surface

microstructure after oxidation. Investigations of cross-sectional microstructure of oxide scales

were conducted by FIB (FEI Quanta 3D). Conventional angular dark field (ADF) images were

obtained by TEM (FEI Tecnai, F30) with STEM (scanning transmission electron microscope)

detector on the same samples for the orientation mapping in Section 3.2.3.


78

3.3 Results

3.3.1 Microstructure of the as-fabricated coatings with different Pt contents

Table 3.3 shows the electroplating time, corresponding Pt layer thickness, average Pt

concentration of as-diffused coatings and root mean square roughness (Rq) of the coating

surface. 20 min and 50 min electroplating time result in 2 µm and 5 µm Pt, and the

corresponding average Pt concentration of the coating is 12.6 at.% and 21.3 at.%, respectively.

The root mean square roughness values are almost identical for the two as-diffused coatings,

and also very close to that of the uncoated substrate (~ 0.45 µm). The nearly identical roughness

values of all samples are prerequisite for the subsequent phase transformation rate analysis,

since the surface roughness has a significant effect on the alumina phase transformation in

oxide scales [162].

Fig. 3.2 shows the cross-sectional SEM images of the as-fabricated γ/γ’ coatings with 2 µm

and 5 µm Pt respectively. The bright phases of the coating are γ’ and the dark phases are γ. The

coating thickness and the fraction of γ’ phase in the coating both increase with increasing Pt

contents.

Table 3-3 Electroplating time and corresponding Pt thickness, average Pt concentration and

surface roughness

Electroplating time, min 0 20 50

Pt layer thickness, μm 0 2 5

Average Pt content, at. % 0 12.6 21.3

Rq, μm 0.45 0.40 0.48


79

Fig. 3.2 Cross-sectional SEM images of as-diffused samples: a) 2 µm Pt coating; b) 5 µm Pt

coating.

3.3.2 Transient alumina to stable α-Al2O3 transformation

The alumina phase identification of scales on samples with different Pt additions after short-

term oxidation was carried out by PLPS. The spectra taken from different places on each

sample were highly reproducible. Therefore, only representative spectra are presented for each

sample. Fig. 3.3 and 3.4 show the PLPS spectra for all samples after 2 min and 10 min oxidation

at 1000 ºC respectively. The characteristic peaks for θ-alumina (if present) and α-alumina were

marked in each spectrum, and the semi-quantitative fractions of θ-alumina in the oxide scale

were calculated according to equation (1) based on the average of 10 randomly chosen spectra

for each sample.

It is clear that for no Pt addition sample, only stable α-Al2O3 can be identified without any

signals from θ-Al2O3 after just 2 min oxidation at 1000 ºC (Fig. 3.3 a). In other words, θ to α-

Al2O3 transition finishes within 2 min for samples without Pt. Conversely, a large amount

(~55.8 %) of θ-alumina is detected in the scale on 5 µm Pt coating after 2 min oxidation (Fig.

3.3 c), and for the coating with 2 µm Pt, nearly half of the scale consists of the θ polymorph

(Fig. 3.3 b). After 10 min oxidation, α-Al2O3 gradually prevails in the scale of 2 µm Pt coating

5 µm 10 µm

γ

γ’

~ 10 µm coating thickness ~ 20 µm coating thickness

γ’ γ

a) b)

2 µm Pt 5 µm Pt


80

but a weak photoluminescence signal from θ-Al2O3 is still detected (Fig. 3.4 b). And the scale

of 5 µm Pt coating still contains a high content (~ 43.9 %, Fig. 3.4 c) of θ-Al2O3 after 10 min.

The θ-Al2O3 fractions of all samples during early stage oxidation are summarized in Fig. 3.5.

As can be seen, after 30 min oxidation, the θ-Al2O3 to α-Al2O3 transformation almost finish on

2 µm Pt sample (only ~ 5.0 % θ-Al2O3), while on 5 µm Pt coating, ~ 17 % θ-Al2O3 was still

detected. Clearly, the fraction of θ-Al2O3 decreases as the oxidation time increasing during

early stage oxidation of the two coatings with Pt addition. Moreover, the more Pt addition

results in more θ-Al2O3 in the scale, which supports the results in [98, 99, 131]. Hence, it can

be concluded that the transient θ-Al2O3 to stable α-Al2O3 transformation is retarded due to Pt

addition for the Pt-diffused γ/γ’ coating.

Fig. 3.3 Typical luminescence spectrum of a) 0 Pt, b) 2 µm Pt and c) 5 µm Pt samples after 2

min oxidation at 1000 ºC.

b)

c)

a)


81

Fig. 3.4 Luminescence spectrum of a) 0 Pt, b) 2 µm Pt and c) 5 µm Pt samples after 10 min

oxidation at 1000 ºC.

Fig. 3.5 Fraction profiles of θ-Al2O3 as a function of oxidation time.

a)

b)

c)


82

3.3.3 TGO composition and microstructure evolution

Fig. 3.6 shows the FIB cross sections of three samples after short-term oxidation at 1000ºC.

After 2 min, it can be seen from Fig. 3.6 a - c that the oxide scale thickness increases as the Pt

content increases. This is due to the fact that Pt addition can promote the growth of transient θ-

Al2O3, as has been confirmed in section 3.3.2, which grows much faster than the stable α-Al2O3.

It is worth noting that at this stage, all scales are composed of pure alumina. After 10 min

oxidation, however, the oxide scales of no Pt and 2 µm Pt sample become duplex in structure:

with outer transient oxide (NiO and/or spinel) layer and inner alumina layer (Fig. 3.6 d and e).

As for 5 µm Pt coating with the highest Pt content, the scale still composes of pure alumina

after 10 min oxidation, without any Ni-oxides formation (Fig. 3.6 f).

Fig. 3.6 a) - c): FIB/SEM cross-sectional images after 2 min oxidation; d) - f): FIB/SEM cross-

sectional images after 10 min oxidation at 1000 ºC of no Pt sample, 2 µm Pt coating and 5 µm

Pt coating, respectively.

Fig. 3.7 shows surface images and FIB cross-sectional images of three samples after 30 min

oxidation. The surface of no Pt sample exhibits fine faceted morphology consisting of NiO

grains (Fig. 7 a), and larger faceted spinel grains are observed on the 2 µm Pt coating (Fig. 3.7

b). While the surface of 5 µm Pt coating shows a combination of nodules and short whiskers


83

(Fig. 3.7 c). The short whisker morphology is typical for an oxide growth as a result of outward

cation diffusion, as expected for the growth of the transient θ-Al2O3 at the gas/oxide interface

in this temperature regime [91, 98]. The short whiskers are only observed on the surface of 5

µm Pt coating. This is in agreement with the results in Fig. 3.5, which shows that after 30 min

oxidation, θ-Al2O3 to α-Al2O3 transition has almost completed on the no Pt sample and 2 µm

Pt coating. As for the cross-sectional microstructure, the oxide scale formed on no Pt sample

is multi-layered, consisting of an outer NiO layer, an intermediate spinel layer and an inner

alumina layer (Fig. 3.7 d). The oxide scale on 2 µm Pt coating is similar to that on no Pt sample,

but without the outermost NiO layer (Fig. 3.7 e). The scales on these two samples are not fully

dense, and a large number of pores can be observed at the interface of spinel/alumina. The

formation of the pores is attributable to the solid-state reaction between NiO and Al2O3, which

induces volume contractions and leads to formation of pores [30]. On the other hand, a dense

and uniform alumina layer (except for some Cr2O3 particles as shown by the arrow in Fig. 3.7

f) is observed on 5 µm Pt coating. After 30 min oxidation, the oxide scale of no Pt and 2 µm

Pt coating is much thicker than that of 5 µm Pt coating because the growth of Ni-oxides is

much faster than alumina.


84

Fig. 3.7 a) - c): SEM surface images; d) - f): FIB/SEM cross-sectional images of no Pt sample,

2 µm Pt coating and 5 µm Pt coating after 30 min oxidation at 1000 ºC.

As shown in Fig. 3.6 and Fig. 3.7, sufficient Pt addition results in an exclusive alumina TGO

layer, compared to samples without Pt (multi-layer TGO structure) or with less Pt addition

(duplex TGO structure). This confirms that Pt can promote alumina growth by inhibiting Ni-

oxides growth at very early stage of oxidation.

3.3.4 TGO growth rate & stress evolution

Fig. 3.8 is the oxidation time at 1000 ºC vs. TGO scale thickness plot of the sample without Pt

and with 5 µm Pt, respectively. As can be seen, initially, 5 µm coatings exhibit a faster TGO

growth compared to the no Pt sample. This is because within 2 min oxidation, θ-Al2O3 to α-

Al2O3 transition finishes in the scale on no Pt sample (Fig. 3.3 a) and the TGO only composes


85

of the slow-growing α-Al2O3. While fast-growing θ-Al2O3 accounts for a large proportion of

the scale on 5 µm Pt-diffused γ/γ’ coating (Fig. 3.3 c), which grows much faster.

Fig. 3.8 Oxide scale thickness evolution of no Pt sample and 5 µm Pt sample as a function of

oxidation time.

However, after 30 min oxidation, the average scale thickness of 5 µm Pt sample is only half of

that on no Pt sample (Fig. 3.8). The TGO growth rate of 5 µm Pt sample is significantly reduced

due to the phase transformation θ- to α-Al2O3, and α-Al2O3 gradually becomes dominant in the

scale of this sample without any Ni-oxide layer formation (Fig. 3.5 and Fig. 3.7 f). But for no

Pt sample, the oxide growth rate increases significantly as a result of Ni-oxide layers formation

(Fig. 3.7 d). Therefore, during the early stage of oxidation, Pt firstly increases oxide growth by

promoting growth of θ-Al2O3, and then it significantly slows down the oxide growth by

promoting the selective oxidation of Al and inhibiting the formation of Ni-oxides/spinel. In

other words, our study confirmed that Pt has two effects on the oxide scale composition during

the early stage oxidation: 1) it can slow down θ- to α-Al2O3 phase transformation; 2) Pt can

promote the selective oxidation of Al at very early stage of oxidation, in agreement with [53].


86

The TGO stress evolution of samples without Pt and with the highest Pt content (5 µm) as a

function of oxidation time is shown in Fig. 3.9. After 2 min oxidation, the TGO stresses of the

two samples are similar. However, afterwards, the compressive TGO stress of no Pt sample

increases significantly from ~ - 2.4 GPa (2 min) to ~ - 3.8 GPa (30 min), whereas for the 5 µm

Pt sample the compressive stress gradually decreases from ~ - 2.2 GPa (2 min) to ~ - 1.4 GPa

(30 min). This simply implies that Pt addition can reduce the compressive TGO stress during

the early stage oxidation of γ/γ’ Ni-Al coatings.

Fig. 3.9 The TGO stress evolution of no Pt sample and 5 µm Pt sample as a function of

oxidation time.

3.3.5 Automated crystal orientation mapping with TEM

Fig. 3.10 and Fig. 3.11 show the results of conventional ADF-STEM and crystal orientation

mapping on TEM of samples with 2 μm Pt coating and 5 μm Pt coating respectively after 10

min oxidation at 1050 ºC. ADF-STEM allows the overall structure (coating-alumina-plat inum

capping layer) to be determined. However, the similarity in composition of the different

alumina polytypes in the scale means that these cannot be determined by the ADF-STEM


87

contrast. Instead scanning diffraction analysis was used to identify the polytypes and the crystal

orientation of the alumina at a high spatial resolution (~ 5 nm). TEM diffraction patterns were

recorded in a raster scan across the sample and then compared to libraries of templates

generated for the different crystal structures present in the sample.


combined phase map and phase reliability map obtained from automated crystal orientation

mapping in TEM, taken from the red box region in a). Green: θ-Al2O3; red colour: α-Al2O3.

From the analysis, maps of the θ-Al2O3 (indicated by green pixels in Fig. 3.10 and 3.11) and α-

Al2O3 (indicated by red pixels) could be obtained. These phase maps were overlaid on phase

reliability maps (see Eq. 4) to allow grain details of the underlying coating to be shown. The

general effect of platinum slowing down θ to α transformation appears to be borne out by this


88

analysis, by inspecting the proportion of α-Al2O3, which was higher in Figure 3.10 b compared

to Figure 3.11 b. This is in agreement with the results in Section 3.3.2.


combined phase map and phase reliability taken from the red box region in a). Green colour:

θ-Al2O3; red: α-Al2O3.

The spatial distribution of θ and α-Al2O3 in the scale can be seen in the maps. Generally, both

maps (Fig. 3.10 b and Fig. 3.11 b) reveal a fine level of detail with grains as small as ~10 nm

in the inner region of the scale, while large columnar grains (100 - 200 nm) typically exists in

the outer region of the scale. This is consistent with a previous study on NiAl alloy by bright

field TEM imaging [163]. Importantly there is variation in the distribution of the different

polytypes along the surface of the coating. By inspecting individual grains which can be

distinguished by the greyscale contrast in the phase reliability map (e. g. the regions ‘A’ and

‘B’ in Figure 3.10 b are distinct grains), there are regions where α-Al2O3 has formed at the

coating surface, while at others θ-Al2O3 is present. Looking further across the scale, it is clear

that some regions have not transformed from θ to α as there are columnar grains of θ-Al2O3


89

extending across the entire scale thickness. These observations suggest that the θ-Al2O3 to α-

Al2O3 transformation rate is inhomogeneous along the coating/oxide interface and related to

the underlying coating grains.

3.3.6 PLPS studies on Ni-Al-Pt alloy samples

From PLPS macroscopic results (Section 3.3.2), we conclude that higher Pt content results in

slower θ-Al2O3 to α-Al2O3 transition. However, the mechanism of this Pt effect is still unclear.

Given the absence of any strong crystallographic coherency between the alumina grains (near

the coating surface) and the coating grains (Appendix. A), it is therefore important to

investigate the effect of the local platinum distribution in the coating since the inhomogeneous

distribution of platinum near the surface of the coating could result in the inhomogeneous

nucleation of α-Al2O3, as confirmed in Section 3.3.5.

To validate the above assumption, a direct method is to examine the coating grains individually

to determine their phase structure and correlate this to the oxide phase grown on each grain

during the early stage oxidation of Pt-diffused γ/γ’ coatings. Nevertheless, there are problems

with this method. Primarily, the alumina growth results in the depletion of aluminium in the

coating and hence the γ’ to γ phase transition can take place. Thus the TEM samples did not

retain any of the original γ/γ’ phase distribution even for very short oxidation exposure.

Furthermore, the optical microscopy was also unable to distinguish the underlying γ’ or γ phase

from the oxide surface of these coatings.

In order to circumvent these problems, the early stage oxidation behaviour of Ni-Al-Pt alloy

with γ/γ’ microstructure was studied. Y. Chen et al. [159] have shown that this alloy maintains

phase contrast of γ’ and γ phase under optical microscope on the oxide surface after up to 4 h

oxidation at 1150 ºC. In addition, γ’ to γ phase transformation is prevented during the early


90

stage oxidation due to a sufficient aluminium source in the underlying alloy allowing the

ordered γ’ phase to exist even after aluminium depletion. Hence, it is possible to study the early

stage oxidation of this alloy by PLPS with relation to the underlying alloy phase structure to

verify the above assumption.

Ni-20Al-xPt (x= 0, 10, 20 at.%) alloys with γ/γ’ microstructure were investigated. The optical

surface images of as-received alloys are shown in Fig. 3.12. The bright areas in Fig. 3.12 a - c

represent γ’ and the dark dendritic areas consist of tiny γ’ blocks embedded in the γ network-

structure channels (Fig. 3.12 d). With increasing Pt content, the dendrites become narrower and

the phase fraction ratio of γ’ to γ increases.

The oxides grown on γ’ and γ/γ’ region, respectively, were studied by PLPS to analyse the

phase after short time exposure at 1050 ºC. The results of three alloys are shown in Fig. 3.13-

3.15, respectively. The γ’ and γ/γ’ region can be easily distinguished after oxidation for all

alloys. As the Pt content increases, the θ-Al2O3 fraction also rises, no matter which region (γ’

or γ/γ’) is detected. This also confirms Pt effect on retarding θ-Al2O3 to α-Al2O3 transformation.

For Ni-20Al alloys, the average θ-Al2O3 fraction of γ’ regions (5 measurements on random

locations) is ~ 21.8 % after 2 min oxidation (Fig. 3.13 b), while θ-Al2O3 fraction of γ/γ’ regions

at this stage is 14.9 % (Fig. 3.13 c), which is 31.6 % lower than that on γ’ areas. After 10 min,

the average θ-Al2O3 fraction of γ’ areas for Ni-20Al is ~ 9.2 % (Fig. 3.13 e), whereas very

weak θ-Al2O3 signal can be detected on γ/γ’ areas (Fig. 3.13 f). The results of Ni-20Al-10Pt

and Ni-20Al-20Pt alloys also show the same trend. For instance, after 2 min, θ-Al2O3 fraction

of γ/γ’ areas of Ni-20Al-10Pt (53.4 %, Fig. 3.14 c) is 30.9 % lower than that on γ’ areas (77.7 %,

Fig. 3.14 b). And for Ni-20Al-20Pt alloy after 10 min oxidation, θ-Al2O3 fraction of γ/γ’ areas

is 35.8 % lower than that on γ’ areas (Fig. 3.15 e and f). The results in Figs. 3.13 - 3.15 have


91

been summarised in Table 3.4 and 3.5. From Table 3.4 and 3.5, it can be concluded that θ-

Al2O3 fraction of γ/γ’ areas is significantly lower than that on γ’ areas for three alloys

throughout the early stage oxidation, which coincides with our assumption that γ phase can

promote θ-Al2O3 to α-Al2O3 transformation, resulting in a lower θ-Al2O3 fraction of γ/γ’ regions.

In other words, these results support that γ grains near the coating/oxide interface promote the

transformation from θ-Al2O3 to α-Al2O3 while γ’ grains retard it.

Fig. 3.12 Microstructure of the as-received a) Ni-20Al, b) Ni-20Al-10Pt, c) and d) Ni-20Al-

20Pt alloy. The inset in d) shows the magnified morphology of the tiny γ channels in γ/γ’ region.


92


regions of Ni-20Al alloy after oxidation at 1050 ºC for 2 min and 10 min, respectively.


regions of Ni-20Al-10Pt alloy after oxidation at 1050 ºC for 2 min and 10 min.


93


regions of Ni-20Al-20Pt alloy after oxidation at 1050 ºC for 2 min and 10 min.

Table 3-4 θ-Al2O3 fraction of γ/γ’ areas and γ’ areas, respectively for the three Ni-Pt-Al

alloys after 2 min oxidation at 1050 ºC

Alloys Cθ (%), γ/γ’ areas Cθ (%), γ’ areas

Ni-20Al 14.9 21.8

Ni-20Al-10Pt 53.4 77.7

Ni-20Al-20Pt 73.2 100

Table 3-5 θ-Al2O3 fraction of γ/γ’ areas and γ’ areas, respectively for the three Ni-Pt-Al

alloys after 10 min oxidation at 1050 ºC

Alloys Cθ (%), γ/γ’ areas Cθ (%), γ’ areas

Ni-20Al 0 9.2

Ni-20Al-10Pt 21.6 46.2

Ni-20Al-20Pt 45.3 70.6


94

3.4 Discussion

3.4.1 Pt effect on the θ-Al2O3 to α-Al2O3 transformation

Platinum addition has been shown to alter the transformation dynamics of θ-Al2O3 to α-Al2O3

[94, 98]. However, the precise mechanism for this is still unclear. The results in this article

suggest that one effect of platinum is the local variation in the nucleation of α-Al2O3 in the θ-

Al2O3 scale. This alters the conventional model of oxide transformation on these coatings.

According to Hayashi and Gleeson [93], α-Al2O3 nucleates uniformly along the coating/oxide

interface, leading to a duplex layered structure of the oxide scale, as shown schematically in

Fig. 3.16 a. However, in this study, γ grains near the interface promote α-Al2O3 nucleation

while γ’ grains retard this transformation (Fig. 3.16 b and c). Under this circumstance, we can

explain why Pt addition slow down this transformation and the non-uniform distribution of α-

Al2O3 along the interface. The θ-Al2O3 to α-Al2O3 transformation is accelerated on coatings

with lower Pt content due to more γ grains near the coating/oxide interface (Fig. 3.16 b).

Conversely, for coatings with higher Pt content, this transformation is retarded because of the

higher γ’ phase to γ phase fraction ratio near the coating/oxide interface (Fig. 3.16 c). Then,

the early α nuclei on γ grains grow toward the gas/oxide interface and finally grow into large

grains, while α-Al2O3 nucleation happens later on γ’ grains, which is consistent with the

microstructural observations in Fig. 3.10 b and Fig. 3.11 b. A previous study suggested that β

grains of NiCrAlY coatings have a large tendency to form θ-Al2O3 than the γ grains [174],

which seems to be supportive to our mechanism considering that both γ’ and β phase contain

more aluminium than the γ phase because aluminium diffusion might affect the θ to α

transformation rate [98].


95

Fig. 3.16 a) Previously reported growth model of α-Al2O3: uniformly nucleate along the

interface, resulting in a layered structure of the oxide scale; b) and c) new growth model in our

study which illustrates lower Pt content and higher Pt content coatings, respectively.

Through the use of spatially resolved PLPS measurements on the NiPtAl alloys with γ/γ’

structure, it was possible to show that where Pt stabilises the ’ structure in nickel the

suppression of the θ-Al2O3 to α-Al2O3 transition was seen. This offers a validation for the

proposed new model in Fig. 3.16. In the absence of significant crystallographic coherence

between the alumina grains (both α and θ) and the coating (Appendix A), it is concluded that

the original distribution of ’ in the coating (associated with the local Pt content) directly

determines the phases present in the oxide scale. This study combined with other studies on

MCrAlY coatings [55, 174] can provide a thorough understanding of the relationship between

the oxide phase/composition and the underlying coating microstructure.


96

In search of the reason(s) why the alumina phase transformation is retarded on the ’ phase

region, firstly, as a noble metal, Pt is non-reactive and does not form any oxide during oxidation

of the coatings. There is also no evidence that Pt can diffuse into the lattice of Al2O3 or

segregate to the grain boundaries of Al2O3 in the literature. Therefore, it is very unlikely that

the retarded θ to α transformation on the γ’ region is caused by the direct chemical interaction

between Pt and Al2O3.

On the other hand, since Pt strongly partitions into γ’, the γ and γ’ phase have different lattice

parameters and corresponding interplanar spacing (d-spacing) [159]. This would affect the

epitaxial strain in the θ-Al2O3 at the very initial stage of oxidation. A low d-spacing mismatch

between θ-Al2O3 and the underlying metal phase is expected to favour the formation of θ-Al2O3

due to the low energy barrier, thus slowing down the θ to α phase transformation. For this

reason, it is proposed that a possible reason for the retarded θ to α transformation on the γ’

regions is the lower minimum d-spacing mismatch between θ-Al2O3 and γ’ phase.

To support this argument, the possible orientation relationships in terms of minimum

interplanar spacing (d-spacing) mismatch between θ-Al2O3 and two coating phases (γ’ and γ)

were investigated. In order to find the possible orientation relationships between θ-Al2O3 and

original γ’ (and γ) phases, the transmission diffraction data including the d-spacing values for

γ’, γ and θ-Al2O3 phase were generated by SingleCrystal ™ software with the basic lattice

parameters for each phase as input [171, 217]. Note that the lattice parameters for γ’ and γ were

measured at 1000℃ in [217], while the lattice parameters for θ-Al2O3 was measured at room

temperatures [171]. Thus the d-spacing values of θ-Al2O3 has been multiplied by a factor of

(1+αAl2O3∆𝑇), where αAl2O3 is the coefficient of thermal expansion of alumina (~8 ppm℃-1) and

∆𝑇 is the temperature drop between the oxidation temperature and room temperature


97

(~1000℃). Table 3.6 gives the data for the two Ni-based phases (γ’ and γ) and θ-Al2O3

including the low-index (hkl) planes and the corresponding d-spacing values.

Table 3-6 Low-index (hkl) planes and the corresponding d-spacing values of two Ni phases

(γ’ and γ) and θ-Al2O3: the d-spacing mismatch between planes of θ-Al2O3 and the

corresponding γ’ (or γ) plane with closest d-spacing matching is calculated and listed as the

strain

θ-Al2O3 γ’ γ

(hkl) dhkl, Å (hkl) dhkl, Å strain (hkl) dhkl, Å strain

(201̅) 4.5307

(201) 3.5542

(400) 2.8660

(1̅11) 2.5700

(111) 2.4497

(310) 2.3172

(202) 2.2652

(311) 2.0298

(1̅12) 2.0232

(11̅2) 1.9078

(003) 1.8211

(2̅03̅) 1.6273

(1̅13) 1.5728

(113) 1.4909

(020) 1.4562

(220) 1.4113

(221) 1.3475

(222̅) 1.2850

(404) 1.1326

(223) 1.0854

(024) 0.9962

(100) 3.6000 1.30%

(110) 2.5456 0.96%

(111) 2.0785 -

(200) 1.8000 1.17%

(210) 1.6100 1.00%

(211) 1.4697 0.92%

(220) 1.2728 0.96%

(221) 1.2000 -

(310) 1.1384 0.51%

(311) 1.0853 0.009%

(222) 1.0329 -

d-spacing mismatch between

planes of θ-Al2O3 and γ’:

ε=𝑑ℎ𝑘𝑙𝜃 −𝑑ℎ′𝑘′𝑙′

𝛾′

𝑑ℎ𝑘𝑙𝛾′

For example, the closest d-

spacing mismatch between θ-

Al2O3 and γ’:

ε=𝑑(223)𝜃 −𝑑

(311)

𝛾′

𝑑(311)𝛾′ =0.009%

(100) 3.5623 0.23%

(110) 2.5189 2.03%

(111) 2.0567 -

(200) 1.7725 2.74%

(210) 1.5931 2.14%

(211) 1.4543 0.13%

(220) 1.2533 2.53%

(221) 1.1874 -

(310) 1.1265 0.54%

(311) 1.0689 1.54%

(222) 1.0234 -

d-spacing mismatch between

planes of θ-Al2O3 and γ:


𝛾

𝑑ℎ𝑘𝑙𝛾

For example, the closest d-

spacing mismatch between θ-

Al2O3 and γ:

ε=𝑑(223)𝜃 −𝑑

(311)

𝛾

𝑑(311)𝛾 =1.54%

*Note: The data for θ-Al2O3 is listed as a decreasing order for d-spacing values, and only the

d-spacing values which are within the range of γ’ and γ phases are listed.

The d-spacing mismatch between planes of θ-Al2O3 and the corresponding γ’ (or γ) plane with

closest d-spacing matching is defined as:


𝛾′

𝑑ℎ𝑘𝑙𝛾′


98

As can be seen from Table 3.6, in the case of 11 low-index planes of γ’ (or γ), there is a clear

d-spacing mismatch advantage which favours the coherent growth of θ-Al2O3 on the γ’ region

by working through the γ’ (or γ) planes as follows:

3 of the 11 low-index planes do not have a close match to the planes of θ-Al2O3 (these planes

are (111), (221) and (222)). As for the remaining 8 planes, 6 of them have a d-spacing mismatch

(strain, ε) below 1% between θ-Al2O3 and γ’ (with the other two exhibiting a strain up to 1.3%).

On the contrary, for the γ planes, only three of them have a strain below 1% and the others

show a much higher strain (2 - 2.5%) than that for the γ’ phase. 1% has been chosen as a

reasonable threshold for the elastic limit of alumina beyond which the coherency is likely to

fail. Hence, there are many more possible coherent orientations between θ-Al2O3 and γ’ for a

given surface facet of the metal substrate. This accounts for the strong tendency for the growth

of θ-Al2O3 on the γ’ phase region, rather than the γ phase.

3.4.2 Pt effect on TGO composition & stress

The benefit of Pt for promoting exclusive alumina growth and inhibiting Ni-oxide growth, as

already discussed in previous publications [53, 68, 159, 175], can be ascribed to two reasons.

Firstly, Pt is inert and does not react with oxygen at temperatures up to 1200 ºC. But it has

strong preference of replacing Ni site in the ordered L12 structure of γ’, resulting in an increase

of the Al:Ni atom ratio with increasing Pt content on a given crystallographic plane containing

both Al and Ni. This increment of Al:Ni ratio would kinetically favour the formation of alumina

in preference to NiO on that plane, when considering only the reacting constituents, Al and Ni

[53]. Secondly, the enrichment of Pt in the subsurface of the Pt-containing coatings during the

early stage oxidation also contributes to this Pt benefit [49]. This enrichment not only favours

the formation of alumina as stated in the first aspect, it can also decrease the chemical activity

of Al (aAl) at the oxide/coating interface. The decrease of aAl would, in turn, facilitate the


99

diffusion flux of Al from the inner part of the coating to the oxide/coating interface [175], thus

providing a sufficient and consistent Al source for alumina growth.

Reduction of the TGO compressive stress of Pt-containing coatings during early stage

oxidation can be explained by the slower transformation from transient alumina to stable α-

Al2O3. Firstly, the slower transformation rate due to presence of Pt allows for an extended stress

relaxation process, which is due to the tensile stresses generated by the large volume shrinkage

(~10 %) associated with the transformation from the monoclinic θ phase to the hexagonal α

phase with higher density [162, 176]. This tensile stress relaxation can lead to significant

reductions of TGO compressive stress [177, 178]. Secondly, the increase ratio of θ/α in the

oxide scale in the presence of Pt may lower the growth stress since θ-Al2O3 is more plastic than

α-Al2O3, which also contributes to the lower stress of higher Pt coatings [98].

3.4.3 Early stage oxidation effect on prolonged oxidation performance

As mentioned above, Pt additions can significantly promote the growth of metastable θ-Al2O3,

which is less protective than the stable α-Al2O3. Moreover, the growth rate of θ-Al2O3 is about

an order of magnitude higher than α-Al2O3 [88]. An extended lifetime of less protective and

faster growing θ-Al2O3 due to Pt addition seems disadvantageous to the scale spallation

resistance at first sight. However, the long term oxidation investigations of coatings used in

this study (Appendix B) shows that coatings with highest Pt contents still presented the longest

TGO lifetime, despite the extended lifetime of the θ-Al2O3 scale. This is mainly attributed to

the fact that Pt can inhibit detrimental Ni-oxide formation at early stage of oxidation (Fig. 3.7).

In addition to the TGO lifetime, the early stage oxidation can also affect the scale morphology

and stress evolution during the prolonged oxidation. It is found that the slower θ to α transition


100

rate during early stage oxidation caused by Pt addition can result in a different oxide

morphology and low compressive stress of the stable scale, compared to coatings without Pt

(Appendix C). This stress reduction can also contribute to the longer TGO lifetime of Pt-

containing coatings, as reported elsewhere [98].

3.5 Summary

Pt addition has three effects on the early stage oxidation of γ/γ’-based Ni-Al coatings: 1) retard

θ-Al2O3 to α-Al2O3 transformation thus extending transient θ-Al2O3 lifetime; 2) promote the

growth of alumina and inhibit the growth of Ni-oxides; 3) significantly reduce the TGO stress

during the early stage of oxidation. Crystal orientation mapping results show that the nucleation

of α-Al2O3 is inhomogeneous along the oxide/coating interface and might be related to the

variation of coating compositions due to Pt additions. Spatially resolved PLPS study of Ni-Pt-

Al alloy (with γ/γ’ microstructure) shows that where Pt stabilises the γ’ structure in nickel, the

suppression of θ-Al2O3 to α-Al2O3 transition is observed. Based on these findings, a new

mechanism has been proposed to explain this Pt effect on θ-Al2O3 to α-Al2O3 transformation:

γ grains near the coating/oxide interface promote α-Al2O3 nucleation while γ’ grains retard this

transformation.

Appendix A. Coating/alumina orientation analysis

Three regions (red boxes in Fig. A1) were investigated to explore the possibility of coherent

relationships affecting the phase transformation in the TGO scale. Through a non-negative

matrix factorisation (NMF) decomposition [179], representative diffraction patterns for

different phases were isolated at the interface region, which also gave a general idea of their

location in the sample. Therefore, phase maps that were complementary to the ASTAR analysis

can be built and compared. Fig. A2 - A4 exhibit the results for Region 1 - 3.


101

Fig. A1 Virtual bright-field image of the FIB-TEM sample (5 µm Pt coating after 10 min

oxidation at 1050 ºC).

For Region 1, a small region of θ-Al2O3 close to the nickel-alumina interface (the upper phase

map in Fig. A2) can be seen. The diffraction patterns for these θ-Al2O3 have partial coherency

with the underlying nickel (overlapping reflections as shown by the upper diffraction pattern

image). Indexing the remainder of the region, it was found that the majority of the alumina has

transformed to α-Al2O3 (lower phase map in Fig. A2). By overlaying these diffraction patterns,

both θ-Al2O3 and α-Al2O3 in this region have no strong orientation relationship with the nickel

substrate, as shown by the lower diffraction pattern image.

There is a twinned area in the middle of the nickel in Region 2 (Fig. A3, twin boundaries are

indicated). Similar to Region 1, a small coherent θ-Al2O3 region near the surface of nickel is

observed. In addition, the coherency is more prominent in the twinned region than the outer

twin region (as the greater coincidence in the diffraction pattern from the inner-twin suggests).


102

Importantly, both α grains and large θ grains that form further out from the interface layer

exhibit little or no coherency with the substrate, as indicated by the lower diffraction patterns

in Fig. A3.

Fig. A2 Phase maps and corresponding diffraction pattern images of different phases in

Region 1.

Region 3 is predominantly α-Al2O3 (Fig. A4). Like other regions, Region 3 also shows no

orientation relationship between the alumina grains and the nickel substrate apart from a

reasonably coherent θ-Al2O3 orientation that is present in the lower part of the region (upper

diffraction pattern image).

Overall, θ-Al2O3 always nucleates partial-coherently along the nickel surface whereas with

continued growth this transforms into incoherent α-Al2O3, with no obvious relationship to the


103

underlying nickel orientation. This suggests that the nickel substrate orientation has no effect

on θ to α-Al2O3 phase transformation.

Fig. A3 Phase map and corresponding diffraction pattern images in Region 2.

Fig. A4 Phase map and corresponding diffraction pattern images in Region 3.


104

Appendix B. Prolonged oxidation lifetime of coatings with different

Pt additions

Fig. B1 shows the optical images of three coatings with different Pt additions after prolonged

oxidation. For no Pt sample, local TGO spallation occurs after 50 h (bright areas in Fig. B1 a),

and spallation is observed nearly all over the surface after 200 h (Fig. B1 c). In contrast, the 5

µm Pt coating scale remains intact up to 200 h oxidation (Fig. B1 e).

Fig. B1 Optical images of sample surface after different periods of prolonged oxidation at

1050 ºC. The bright area is the surface of the underlying metal due to spallation of the TGO.


105

Appendix C. Early stage oxidation effect on the stable scale

morphology & stress

The TGO stress evolution of no Pt sample and 5 µm Pt coating after prolonged oxidation is

shown in Fig. C1. The TGO stress of 5 µm Pt coating is lower than that of no Pt sample, until

the TGO spallation on Pt sample which causes a significant drop of stress (Fig. C1 a). Moreover,

there are two stages of stress evolution of 5 µm Pt coatings. During Stage Ⅰ (up to 10 h), the

TGO stress remains at a low level (~ 3.5 GPa) compared to no Pt coatings. After 10 h (Stage

Ⅱ), the stress shows a significant rise (~ - 4.5 GPa). Due to Pt addition, the growth of θ-Al2O3

is extended and thus numerous θ-Al2O3 whiskers form on the surface of 5 µm Pt coating during

early stage oxidation. These whiskers totally transform into stable α-Al2O3 and sustain their

whisker shape after phase transformation on the surface of the stable scale (Fig. C1 b, after 5 h

oxidation). These whiskers, being less constrained by the underlying coating, can cause the R

peaks of α-Al2O3 in PLPS spectrum shift to much higher frequencies, resulting in lower TGO

stresses. In this case, it should be assured that the scale is intact, because any crack would give

similar stress-free signals, which has been ruled out by the microstructural study of the oxide

scale. While for no Pt sample, since θ to α transformation is much faster, no whiskers remain

on the stable scale. Therefore the TGO stress of this sample is higher until TGO spalls. It is

noticed that whiskers on the stable scale of 5 µm Pt coating become smoother with oxidation

time (Fig. C1 c) as a result of surface diffusion driven by a reduction of surface energy, which

causes a slight rise of stress during the Stage Ⅰ. Eventually, these whiskers on the surface

evaporate after prolonged exposure at high temperatures (Fig. C1 d), thus leading to a

significant rise of the TGO stress.


106

Fig. C1 a) Residual stress evolution in TGO scale on no Pt sample and 5 µm Pt coating with

oxidation; b) - d) scale surface morphologies of 5 µm Pt coating after 5h, 10 h and 50 h

isothermal oxidation at 1050 ºC.

CHAPTER 4 EFFECT OF SUPERALLOY SUBSTRATE ON THE LIFETIME OF EBPVD TBCs

107

Chapter 4 Effect of Superalloy Substrate on the Lifetime and

Interfacial Toughness of Electron Beam Physical

Vapour Deposited Thermal Barrier Coatings

4.1 Introduction

Thermal barrier coatings (TBCs) are multilayered and multifunctional material assemblies that

have been widely applied for an improved performance and efficiency of gas turbine engines

[31]. State-of-the-art TBCs consist of a ceramic top coat of yttria stabilized zirconia (YSZ)

deposited by electron beam physical vapour deposition (EB-PVD) or plasma spraying, a β-

PtAl or γ/γ’ based diffusion or MCrAlY (M=Ni, Co or a mixture of two) overlay bond coat,

and a nickel-based superalloy substrate. Since the YSZ top coat is permeable to oxygen, the

bond coat oxidizes and forms a thermally grown oxide (TGO) scale (usually Al2O3) on top of

itself during service at high temperature. The TGO layer is subject to very large compressive

stress when the system is cooled due to the thermal mismatch with the metal substrate. Various

failure modes have been observed for TBCs, and the most common one for EB-PVD TBCs is

cracking at the bond coat/TGO interface leading to buckling delamination [110].

A number of studies have reported that the superalloy substrate composition can affect the

cyclic lifetime of TBCs [180-184]. For example, R. T. Wu et al. [185] found that the cyclic

lifetime varied significantly for TBCs with Pt-diffusion bond coats on different superalloy

substrates including SRR99, CMSX-4, etc., although the reason for this difference remains

unclear. B. A. Pint et al. [186] have pointed out that titanium (Ti) in the CMSX-4 superalloy

substrate was suspected to have degraded the alumina scale adhesion for Pt-diffusion bond

coats, thus was detrimental to the lifetime of TBC systems. However, they also found that the


108

poorer coating performance for CMSX-4 superalloy was not found in all cases because the Pt-

modified aluminide bond coat lifetime on CMSX-4 substrate was significantly higher

compared to other superalloys. Thus, more comprehensive analysis is needed to understand the

coating performance on CMSX-4 superalloy and the effect of Ti.

Another important factor affecting the TBC lifetime is the resistance to crack propagation

(toughness) of the relevant interfaces. The coating durability is dependent on the interplay

between the cracking driving forces and the resistance to crack propagation through the coating

or along the relevant interfaces. So measuring the relevant interface toughness of TBCs by

reliable test methods is critical to establish reliable lifetime models [149, 187]. Methods have

been developed to measure the interface fracture toughness in ceramic coating systems

including: (1) bending test of notched multilayer beams [134, 188]; (2) indentation [141, 189];

(3) blister methods [190]; (4) the push-out test [147, 149]. However, some of these methods

have limited applicability to TBC systems [132]. For instance, some TBCs on the turbine blade

samples cannot be tested by the methods which require specific geometry of samples, such as

push-out tests. Moreover, the crack path is difficult to control in a multilayer TBC system. In

some cases, cracks will divert away from the interface of interest into the vertical YSZ columns,

which makes it impossible to measure that interface. Recently, X. Wang et al. [132] have

utilized a cross-sectional indentation (CSI) method to measure the fracture energy of the EB-

PVD TBCs with the Pt-diffused bond coats. CSI does not require a designed sample geometry

and can produce controlled interface fracture. But complex finite element modelling is required

to analyse the plastic deformation of the substrate for the fracture energy calculation. X. Zhao

et al. [191] have employed a strain-to-fail test to measure the interface fracture toughness based

on buckle delamination mechanisms. Their method circumvents the complex modelling of

plastic deformation of the substrate. In addition, the in-plane compression load applied to the


109

top coat during the strain-to-fail test can inhibit the crack deflection into the YSZ layer. The

energy release rate associated with the buckling-driven delamination is estimated using the

buckle radius, coating stress and modulus and is taken as the interfacial toughness [191].

However, this requires that the buckle propagates stably without any ridge cracks. Failure to

fulfil these conditions will lead to an under-estimation to the interfacial toughness.

This chapter has two goals. Firstly, two single crystal superalloy blades (CMSX-4 and René

N5) coated with Pt-diffused γ/γ’ bond coats and YSZ top coats deposited by EB-PVD will be

investigated in terms of their cyclic oxidation behaviour in order to elucidate the effect of

superalloy substrate on TBC lifetimes. Secondly, the strain-to-fail test proposed by X. Zhao et

al. [191] is combined with the 3D-DIC technique to compare the coatings’ interfacial toughness

and its evolution with oxidation. The 3D-DIC technique can analyse the whole stable buckling

propagation process before any ridge crack initiation, thus allowing a reliable determination of

interface toughness values based on the well-established buckle-driven delamination

mechanism [139].


4.2.1 Sample preparation

The TBC blades investigated in this study were provided by Rolls-Royce plc. The composition

of the two superalloy substrates (CMSX-4 and René N5, denoted as X4 and N5, respectively)

determined by energy-dispersive X-ray fluorescence (EDXRF; PANalytical MiniPal 4) is given

in Table 4.1. These two superalloy blades were coated with Pt-diffused γ/γ’ bond coats

followed by the YSZ top coat. Firstly, the blades were grit blasted (alumina particles) and then

electroplated with Pt followed by annealing in vacuum at high temperatures

for Pt diffusion. The annealing resulted in a Pt-diffused bond coat ~ 30 µm thick. A ~ 175 µm


110

YSZ top coat (~ 7 - 8 wt.% Y2O3) was deposited on one side of the blades by EB-PVD using a

commercial process. A total of five blades for each type of superalloy substrate were fabricated.

Table 4-1 Superalloy compositions (atomic %) by EDXRF

Superalloy Ni Cr Al Ta Ti W Re Mo Co

René N5 Bal. 7.8 13.1 2.2 - 1.6 0.9 0.9 8.0

CMSX-4 Bal. 6.8 9.8 2.2 1.3 1.8 0.9 0.5 10.2

4.2.2 Thermal treatment

Cyclic oxidation was performed on some blades in laboratory air between room temperature

and 1200 °C in a cycle furnace (CM™). Each cycle consisted of 10 min ramping period, 1 h

holding time at 1200 °C and 10 min fan-assisted air quenching.

Other blades were exposed to isothermal oxidation at 1150 ºC in laboratory air in a box furnace

(CM™) for different periods of time up to 45 h. These isothermally oxidized specimens were

used for the strain-to-fail test for an evaluation of interface toughness. Isothermal oxidation at

a lower temperature (1150 ºC) can ensure that no sub-critical cracks initiation at the interface

after oxidation, thus the intrinsic interface toughness can be measured and compared for the

two TBCs with different substrates. Flat parts of the blades were cut by SiC blade into 6×6×1.5

mm samples using precision machine (Accutom 5, Struers) and used for the strain-to-fail test.

4.2.3 Microstructure characterization

For cross-sectional investigation of the samples, firstly they were ground by SiC paper up to

1200# and then polished using diamond paste to 0.25 μm finish. Scanning electron microscopy

(SEM; FEI Quanta 650) equipped with an energy dispersive X-ray spectrometer (EDS) was

used to examine the microstructure and composition of cross-sections. The bond coat surface


111

after spallation of the YSZ top coat was carbon coated and also examined by SEM.

Transmission electron microscope (TEM; FEI Talos F200A) with STEM (scanning

transmission electron microscope)/EDX (energy-dispersive X-ray) detector was used to

examine the TGO/coating interface. Focused ion beam (FIB; FEI Quanta 3D) in-situ lift-out

technique [168] was used to prepare cross-sectional site-specific specimens for TEM analysis .

4.2.4 Interface toughness measurement by the strain-to-fail test

4.2.4.1 Theoretical background

A strain-to-fail test combined with 3D-DIC technique was employed to measure the interface

toughness of EBPVD TBCs based on the buckle delamination mechanism. A sequence of

events involved in the buckling failure mode of a compressed coating includes [112]: ⅰ)

separations develop between the substrate and the coating; ⅱ) the film buckles once the

separation reaches a critical size; ⅲ) the buckle propagates along the coating/substrate interface;

iv) the crack deflects toward the coating surface leading to the coating spallation. Buckling of

a coating subject to an equi-biaxial compression, σ, occurs at a critical stress [192]:

σc=1.22[𝐸𝑐

1−𝜐𝑐2](ℎ/𝑏)

2 (4.1)

where b is the separation radius, h is the coating thickness, Ec and υc are the elastic modulus

and Poisson ratio of the coating, respectively. When the stress in the coating exceeds the critical

stress σc, the coating buckles away from the substrate and the energy release rate G is described

as [192]:

G=[1 + 0.9(1 − 𝜐𝑐)]−1𝐺0 [1 − (𝜎𝑐/𝜎)

2] (4.2)

where G0 is the elastic energy stored in the unbuckled coating per unit area, given by [192]

G0=1−𝜐𝑐

𝐸𝑐𝜎2ℎ (4.3)

The mode-dependent interface toughness Γ(ψ) is defined as [139]:


112

Γ(ψ)= Γicf(ψ) (4.4)

where Γic is the mode Ⅰ toughness and the criterion for a buckle to propagate is [112]

G= Γ(ψ) (4.5)

and

f(ψ)=1+ 𝑡𝑎𝑛2[(1 − 𝜆) ψ] (4.6)

where λ is a coefficient reflecting the interface friction (λ is about 0.3 for the bond coat/TGO

interface [112]). ψ is the loading phase angle, which represents the proportion of mode Ⅱ to

mode Ⅰ fracture, given by [139]:

tan ψ = (cos𝜔 + 𝜂sin 𝜔)/(ηcos𝜔 − sin𝜔) (4.7)

where 𝜔 is the phase angle shift generated by the elastic mismatch. For an alumina/Ni interface,

𝜔 = 52° [112].

And η is given by:

η = 0.25(1+𝜐𝑐)[(σ/𝜎𝑐 − 1)/g2]1/2 (4.8)

with

g2 = 0.25(1+𝜐𝑐) + 0.22(1−𝜐𝑐2) (4.9)

Therefore, if the stress σ of the coating, the buckling radius b and the elastic modulus Ec of the

coating are known, we can calculate the interfacial toughness Γ ic for the bond coat/TGO

interface based on the above expressions.

4.2.4.2 Strain-to-fail test coupled with 3D-DIC

Although both the TBC top coat and the alumina layer are in compression after thermal

exposure mainly due to the thermal mismatch between the ceramic layers and the metal

substrate, this stress is not sufficient to make the coating buckle from the substrate

spontaneously [191]. So an external compressive strain is applied to the pre-stressed coating to

facilitate the buckling process as schematically shown in Fig. 4.1. The compression load was

applied along the direction parallel to the TBC/substrate interface (y-axis direction in Fig. 4.1)


113

by a Static Testing Machine Instron 5569H1549 at a loading rate of 0.07 mm∙min-1. The

buckling process of the EBPVD TBCs was monitored by two high speed cameras of the 3D-

DIC system (LaVision 3D DIC VP18-0021), at a frequency of 5 Hz (~200 ms exposure time

for each frame).

Fig. 4.1 Schematic view of the experimental setup: both high speed cameras take images of the

TBC coating surface (x-y plane) during the compression test, and the compression load is along

the y-direction.

4.2.4.3 Determination of the coating stress

The stress in the coating upon buckling is originally composed of both the residual stress and

the externally-applied stress. The applied stress can be calculated by the strain determined by

DIC, which will be illustrated later. The residual stress of alumina scale was measured by

luminescence spectrum on Renishaw Invia Raman system (RenishawTM, Gloucestershire, UK)

with an argon laser source (λ=633 nm) [193]. All measurements were performed through the

top coat surfaces. Before each experiment, the spectrometer was calibrated by taking a

spectrum from a standard pure silicon sample. The fluorescence spectra from Cr3+ in alumina

scale were collected for each measurement with five seconds acquisition time. Cr3+


114

fluorescence in α-Al2O3 produces the R1 and R2 doublet at ~14400 cm-1. To determine the

peak positions in each spectrum, all spectra were deconvoluted in Wire 4.2 software

(RenishawTM) with an automatic fitting program by two mixed Gaussian-Lorentzian functions

[166]. The stress was estimated from the peak shift ∆𝜈 of the R2 line with respect to the

unstrained single crystal sapphire by assuming an equi-biaxial stress state and randomly

distributed TGO grains. The stress σ in the alumina scale was calculated by [130]:

∆𝜈 = 5.07(𝑐𝑚−1𝐺𝑃𝑎−1)σ (4.10)

4.2.4.4 Determination of the YSZ modulus

Due to the columnar structure of the EB-PVD TBCs, the stiffness of the coating can be highly

anisotropic. Since the energy release rate associated with the buckling-driven delamination

depends mainly on the coating’s in-plane elastic modulus, the nano-indentation test has been

used on the polished cross-sections of the TBCs to determine their in-plane elastic modulus. A

MTS XP nano-indenter was used. For each sample, a 2×10 (60 μm step size) array of indents

were made on the cross section using a Berkovich indenter with a penetration depth of 2 μm.

The Young’s modulus of YSZ top coat was calculated by the Oliver-Pharr method [194].

4.2.4.5 Measuring buckling radius by 3D-DIC

The recorded digital images were analysed using Davis 10.0 software. The displacement of the

coating surface was first determined by registering each frame with the reference image taken

prior to loading using a lease square matching (LSM) algorithm. In order to highlight the buckle,

the images showing the deformed coating surface were first corrected by the calculated

displacement field g (x + u (x)) and subtracted from the reference image f (x). The obtained

difference image (residual field) can clearly reveal cracks or any geometric features on the

sample during the loading process, which cannot be distinguished on as-deformed images [195].


115

4.3 Results

4.3.1 Microstructure of the as-received TBCs

Fig. 4.2 As-received TBCs with different substrates, a) - d): cross-sectional SEM images; e)

and f): Ni, Al and Pt concentration profiles by SEM/EDX linescan along the lines in c) and d).

c) N5, as-received

e) f)

20 μm

a) N5, as-received b) X4, as-received

d) X4, as-received

100 μm

YSZ ~ 180 µm

BC ~ 27 µm

γ’ γ

EDX linescan

Alumina particles (grit blasting)


116

Cross-sectional SEM images illustrating the as-received microstructure of the TBCs on two

different superalloy substrates are shown in Fig. 4.2. From Fig. 4.2 a and b, it can be seen that

both the YSZ top coat and the Pt diffusion bond coat (BC) were prepared to identical thickness

for the two superalloy substrates. The top coat had the characteristic columnar grain structure

typically produced by EB-PVD. As shown in Fig. 4.2 c and d, the Pt-diffusion bond coat

consisted of γ phase (darker contrast) and γ’ phase (brighter contrast). And the bond coat on

the X4 and N5 substrates both contained a grit-line (residual Al2O3 particles from the grit-

blasting process), which was at ~ 10 µm distance from the bond coat/top coat interface. The

EDX concentration line profiles of Ni, Pt and Al of the bond coats on two different substrates

are shown in Fig. 4.2 e and f, respectively. The Ni and Al concentration profiles on different

substrates were very similar, while the average Pt concentration of the bond coat on N5

substrate was slightly higher than that on X4 substrate.

4.3.2 Cyclic oxidation testing

4.3.2.1 YSZ lifetime

The lifetimes of TBCs with N5 and X4 substrates for 1-h cyclic test at 1200 ºC are illustrated

in Fig. 4.3. For each substrate, a total of three blades were cycled. The failure of the coating

was defined as ~ 20% spallation of the top coat (only consider the flat part of the blade). It can

be seen that the lifetime deviation for the three blades of the same substrate was small.

Moreover, three blades of the X4 substrate all exhibited longer lifetimes (30 cycles in average)

compared to that of the N5 substrate (24 cycles in average). The average lifetime of YSZ

coating on the X4 substrates reported here were similar to other studies on the same TBC

system under the same cycling condition [182, 196]. However, the higher lifetimes of TBCs

on X4 substrates than N5 substrates were in conflict with the result by B. A. Pint et al. [186],

which concluded that the Pt diffusion bond coat on superalloy X4 exhibited the shortest YSZ


117

coating lifetime among different superalloy substrates including N5. Although the reason for

this conflict result is still not clear, it might be related to the differences in the coating

preparation process.

Fig. 4.3 YSZ lifetime for the cyclic oxidation testing of TBC-coated different superalloy blades.

4.3.2.2 Microstructural evolution

Fig. 4.4 exhibits the cross-sectional SEM images and concentration profiles of Ni, Pt and Al in

the bond coat on N5 and X4 substrates, after 5 1-h cycles at 1200 °C. At this stage, cracks along

the TGO/bond coat interface were observed on the N5 substrate (Fig. 4.4 a), whereas no

interfacial crack was found for the coating on the X4 substrate (Fig. 4.4 b). This indicates that

the loss of adhesion between the bond coat and TGO was earlier for TBC on the N5 substrate

than X4 substrate, which coincides with the longer lifetime of the TBC on the X4 substrate

(Fig. 4.3). As can be seen from Fig. 4.4 c and d, a γ-phase zone arose near the bond coat surface

for both TBC systems as a result of Al depletion. This is ascribed to the interdiffusion between

the substrate and the bond coat and aluminium loss due to oxidation. It is notable that the Al

concentration near the bond coat surface on the N5 substrate was nearly identical to that on X4

substrate (~ 10 at.%), suggesting that the Al depletion with thermal cycling was similar for the


118

bond coats on the X4 and N5 substrate. While the Pt concentration near the bond coat surface

on the N5 substrate (~ 15 at.%, Fig. 4.4 e) was slightly higher than that on the X4 substrate (~

11 at.%, Fig. 4.4 f), as also found for the as-received coatings (Fig. 4.2 e and f).

Fig. 4.4 Cross-sectional SEM images: a) b) secondary electron (SE) mode; c) d) backscattered

electron (BSE) mode; e) f) corresponding SEM/EDX linescan elemental concentration profile

along the red lines in c) and d), respectively after 5 1-h cycles at 1200 °C.

N5, 5 cycles

c)

20 μm

e) f)

d)

b)

γ-phase, Al depletion zone

EDX linescan

50 μm

X4, 5 cycles

a)

Crack along the TGO/bond coat interface

No interfacial crack at this stage


119

Fig. 4.5 a) b): cross-sectional SEM images (SE mode) and c) d): Ni, Al and Pt concentration

profiles by SEM/EDX linescan along the red lines in a) and b), respectively after 10 1-h cycles

at 1200 °C.

After further thermal cycling, as shown in Fig. 4.5, the cracks still propagated predominately

along the TGO/bond coat interface for the N5 substrate (Fig. 4.5 a). For the bond coat on the

X4 substrate, on the contrary, cracks along the TGO/bond coat interface (the red arrow in Fig.

4.5 b) and within TGO (black arrows in Fig. 4.5 b) co-existed after 10 cycles. Again, the

elemental concentration profiles for bond coats on the X4 and N5 substrates at this stage were

still similar when comparing Fig. 4.5 c and d, except for the slightly higher Pt concentration of

N5 substrate near the bond coat surface. Previous studies [175, 197] have pointed out that

higher Pt concentration near the bond coat surface can promote the uphill diffusion of

aluminium to the bond coat/TGO interface. However, it is observed here that the Al

concentration near the bond coat surface is quite similar for the coatings on N5 and X4

a)

c)

Cracks along TGO/BC interface

d)

50 μm N5, 10 cycles X4, 10 cycles

Cracks within TGO

b)

Cracks along TGO/BC interface

EDX linescan


120

substrates. This suggests that Al depletion cannot explain the lifetime difference and there

should be some other reason for the shorter lifetime of the N5 superalloy system, which will

be discussed later.

Typical morphologies of the bond coat surface exposed by spallation of the top coat on the X4

and N5 substrate are shown in Fig. 4.6 a and b, and Fig. 4.6 c and d show the underside of a

spalled piece of the top coat. The backscattered electron (BSE) images can clearly demonstrate

the alumina (TGO), YSZ and exposed bond coat surface by different contrast. For TBC on the

N5 substrate, the exposed surface by failure of the coating showed mainly the bond coat (grey

area in Fig. 4.6 a), and correspondingly, the underside of the spalled top coat piece was covered

mainly by the alumina (dark area in Fig. 4.6 c), with only a small portion of YSZ presented

(bright area in Fig. 4.6 c). Therefore, the failure for the TBC on the N5 substrate was

predominantly along the bond coat/TGO interface, which agrees with the above cross-sectional

observations.

In contrast, TBC on the X4 substrate has failed in a different way. From Fig. 4.6 b, it can be

seen that the exposed surface for this system exhibited a mixture of alumina (dark area), YSZ

(mid-tone area, surrounded by alumina) and the metallic bond coat (bright area), suggesting

that the fracture for this system can take place at the TGO/bond coat interface and also within

the TGO. Sometimes, cracks can also extend laterally into the YSZ where TGO intrusions into

the bond coat existed (examples of this are shown in Fig. 4.5 b), thus leaving some YSZ

particles on the exposed bond coat surface. This is also consistent with the morphology of the

underside of spalled YSZ shown in Fig. 4.6 d.

In summary, the failure of TBC occurred predominantly along the bond coat/TGO interface for

TBCs on the René N5 substrate, whereas for TBCs on the CMSX-4 substrate, a mixed failure


121

path (at the bond coat/TGO interface and within TGO) was observed. These observations are

self-consistent and also in agreement with the failures reported by other researchers [130, 185]

for similar specimens.

Fig. 4.6 a) and b): bond coat surface BSE images exposed by spalling after 28 cycles at 1200 °C;

c) and d): BSE images of the back side of spalled YSZ coating after 28 cycles.

Fig. 4.7 a shows the average TGO thickness evolution during the cycling test at 1200 °C. It can

be seen that the average TGO thicknesses on the X4 and N5 substrate were quite similar up to

20 cycles, because microstructural studies confirmed that the bond coat on both superalloy

substrates (X4 and N5) was able to develop a pure alumina layer without formation of other

oxides (e.g. Ni-oxides) during the cycling test until failure. In addition, although the average

lifetime of TBC on the N5 substrate was 24 cycles, in comparison with 30 cycles for the X4

a) N5, bond coat surface

b) X4, bond coat surface

d) X4, spallaed YSZ

underside c) N5, spallaed YSZ underside

Alumina

Bond coat

YSZ

Bond coat

50 μm

alumina

Alumina

YSZ

Alumina

YSZ


122

substrate, both superalloy systems exhibited about the same TGO thickness near the time of

failure. This comparable TGO thickening rate indicates that the TBC lifetime difference of the

two superalloy substrates is not correlated to the TGO thickening evolution.

Fig. 4.7 b illustrates the root mean square roughness evolution of TGO/BC interface of the

TBCs on the X4 and N5 substrate, respectively. Clearly, similar interfacial roughness evolution

throughout the cycling test for different substrate systems indicates that the influence of TGO

rumpling on the observed lifetime difference can be neglected. B. A. Pint et al. [186] also

concluded that TGO rumpling had no effect on TBC lifetime by investigating the Pt-modified

aluminide coating on the X4 and N5 substrate, respectively.

Fig. 4.7 a) TGO thickness evolution during the cycling test; b) root mean square roughness

evolution of TGO/BC interface by processing of digitized profiles.

Overall, the cycling test showed that the TBC on the CMSX-4 superalloy substrate has a 25%

longer lifetime compared to that on the René N5 substrate. No significant difference regarding

interfacial morphology and TGO growth rate evolution has been observed for the TBCs on

different substrates, which rules out these effects on the observed lifetime difference. The

microstructural investigations showed that interfacial cracking occurred earlier for the TBC on

the N5 substrate than that on X4 substrate. In addition, the fracture was mainly found along the

a) b)


123

bond coat/TGO interface for the René N5 specimens, whereas for the TBC on the CMSX-4

substrate, a mixed failure path (at the bond coat/TGO interface and within TGO) was observed.

This strongly indicated that the bond coat/TGO interfacial degradation was faster and to a

greater extent for the N5 specimen, as compared to the X4 specimens under identical cyclic

exposure. To confirm this, a strain-to-fail test based on a previously established model was

employed to measure the bond coat/TGO interface fracture toughness evolution during

isothermal oxidation of TBCs with the N5 and X4 substrates, respectively. The results are

presented as follows.

4.3.3 Strain-to-fail compression test coupled with 3D-DIC

DIC residual fields were utilised to reveal the whole buckling propagation process during the

strain-to-fail test. Fig. 4.8 exhibits a) the reference image f (x), b) the deformed images

corrected by the DIC displacement field g (x + u (x)) with increasing loading time and c) the

corresponding residual field Φ (x) (differences between g (x + u (x)) and f (x)). Buckling can

hardly be seen on the deformed images (Fig. 4.8 b), but appeared clearly on the residual fields

(Fig. 4.8 c) due to the luminosity change induced by the buckling initiation. Buckling initiated

at the edge of the analysing area (not the sample edge, as indicated by the optical image of the

sample surface in Fig. 4.8 a), as shown by a red ellipse in Fig. 4.8 c, and then it grew stably.

Based on this image processing method, the buckling radius and the corresponding strain can

be correlated during the stable buckling propagation process. This provides a more accurate

buckling measurement compared to the previous work by X. Zhao et al. [191] as their work

only considered the final buckled stage.


124

Fig. 4.8 a) Optical image taken by the camera for DIC showing the sample surface prior to

applying the load. The region-of-interest (ROI) is highlighted with the red rectangle and is used


125

as the reference image f (x); b) deformed images at several loading scales corrected by a DIC

displacement field g (x + u (x)) and c) the corresponding residual field Φ (x) of a sample (as-

received TBC with N5 substrate) during the test. The red ellipse in c) highlights the occurrence

of buckling.

Buckles at intermediate propagation stage (e.g. the buckle shown by a red rectangle in Fig. 4.8

c) were considered for the calculation of interface toughness of all tested specimens. Fig. 4.9 a

shows the buckling radii of TBCs with N5 and X4 substrates, respectively, as a function of

oxidation time. It can be seen that the buckling radii for TBCs with two different substrates

were similar up to 45 h oxidation time. Fig. 4.9 b) shows the strains for corresponding buckling

radii calculated by DIC of TBCs with N5 and X4 substrates, respectively. The compressive

strains of both TBCs have decreased in magnitude with increasing oxidation time. This might

indicate degradation of coating adhesion during oxidation. These buckling strains and radii will

be used for the calculation of interface toughness.

Fig. 4.9 Evolution of a) average buckling radii, and b) corresponding strains calculated by DIC

as a function of oxidation time at 1150 °C for specimens with the X4 and N5 substrate,

respectively.

The compressive residual stresses in the TGO of isothermal oxidized specimens with N5 and

X4 substrates are shown in Fig. 4.10 a). The TGO stress on N5 substrate was ~ -2.2 GPa for

a) b)


126

as-received samples and reached a peak of ~ -2.8 GPa after 30 h oxidation, then the stress

slightly decreased to ~ -2.6 GPa after 45 h oxidation. This stress trend is the same as a previous

investigation on TBCs with Pt-diffused γ/γ’ bond coats under cyclic oxidation [198]. The TGO

stresses for the specimens with X4 substrates exhibited a similar trend but the stress values

were smaller than that of N5 substrates. The average TGO thicknesses are plotted in Fig. 4.10

b) as a function of oxidation time for two TBC systems. The TGO thicknesses for as-received

TBCs with N5 and X4 substrates were ~ 0.5 μm, and then the TGO growth followed a parabolic

law and grew to ~ 3.0 μm after 45 h.

Fig. 4.10 a) Residual stress of TGO and b) average TGO thickness as a function of isothermal

oxidation time at 1150 °C for TBCs with the X4 and N5 substrate, respectively.

The elastic moduli of YSZ top coat measured by the cross-sectional nano-indentation are listed

in Table 4.2. The elastic moduli increased slightly with increasing oxidation time. This is due

to the sintering of YSZ during the heat treatment.

Table 4-2 Elastic moduli of YSZ top coat measured by the cross-sectional nano-indentation

test

Oxidation time (h) 0 15 30 45

EYSZ (GPa) 25.6 + 1.7 27.1 + 2.0 30.1 + 2.3 32.8 + 3.1

a) b)


127

4.4 Discussion

4.4.1 Estimation of the interfacial toughness for TBCs on N5 and X4 substrates

So far, all parameters required for estimation of interfacial toughness are presented. Then the

interfacial toughness for TBCs on the N5 and X4 substrates can be estimated based on the

buckling scenario as described in Section 4.2.4.1. However, this buckling principle is for a

single layer coating system, while the coating here is actually a bilayer consists of a TGO layer

and a YSZ top coat. Therefore, some parameters will be modified to make the calculation

suitable for this bilayer. The (TGO + YSZ) bilayer can be approximated as a single layer, with

the effective Young’s modulus given by [146]:

Eeff = 𝐸𝑌𝑆𝑍ℎ𝑌𝑆𝑍/(1−𝜐𝑌𝑆𝑍)+𝐸𝑇𝐺𝑂ℎ𝑇𝐺𝑂/(1−𝜐𝑇𝐺𝑂)

ℎ𝑌𝑆𝑍+ℎ𝑇𝐺𝑂 (4.11)

where

𝐸𝑌𝑆𝑍, 𝐸𝑇𝐺𝑂 -Young’s modulus of the YSZ layer and the TGO layer,

ℎ𝑌𝑆𝑍, ℎ𝑇𝐺𝑂 - Average thickness of the YSZ and TGO layer, and

𝜐𝑌𝑆𝑍, 𝜐𝑇𝐺𝑂 - Poisson’s ratio of the YSZ and TGO layer, respectively.

The effective stress of this bilayer can be written as [146]:

σ = 𝜎𝑌𝑆𝑍ℎ𝑌𝑆𝑍+𝜎𝑇𝐺𝑂ℎ𝑇𝐺𝑂

ℎ𝑌𝑆𝑍+ℎ𝑇𝐺𝑂 (4.12)

where σ is the stress of each layer, given by [191]:

σYSZ = σYSZ,0 + σYSZ, applied = 𝐸𝑌𝑆𝑍∆𝛼∆𝑇

1−𝜐𝑌𝑆𝑍 +

𝐸𝑌𝑆𝑍𝜀

1−𝜐𝑌𝑆𝑍 (4.13a)

σTGO = σTGO,0 + σTGO, applied = σTGO,0 +𝐸𝑇𝐺𝑂𝜀

1−𝜐𝑇𝐺𝑂 (4.13b)

where the subscripts ‘0’ and ‘applied’ refer to the stress before and after applying the

compressive strain. ε is the buckling strain, as illustrated in Fig. 4.9 b. ∆𝛼 is the difference of

coefficient of thermal expansion between the YSZ and the substrate (~ 4×10-6 K-1 [4]); ∆𝑇 is


128

the temperature drop from the oxidation temperature (1150 °C) to the room temperature (25 °C).

σTGO,0 is the residual stress of the TGO scale, measured by luminescence spectrum (Fig. 4.10

a), and the residual stress of the YSZ layer is calculated by assuming the stress to be thermally

elastic, i.e. ε = 0 in Eq. 4.13a.

For the calculation of Γic (mode Ⅰ interfacial toughness), the buckling strains and corresponding

radii were taken from Fig. 4.9, and the stress of the TGO and YSZ bilayer was calculated

according to Eqs. 4.12 and 4.13. The TGO stresses and average thicknesses were taken from

Fig. 4.10. The other parameters were 𝐸𝑇𝐺𝑂 = 380 GPa, υTGO = υYSZ =0.22, hYSZ= 180 µm and

elastic moduli of YSZ coating shown in Table 4.2.

The calculated interfacial toughness for TBCs on the N5 (black squares) and X4 (red circles)

substrates as a function of oxidation time is shown in Fig. 4.11. The mode Ⅰ interfacial

toughness values of TBCs reported by other researchers [132, 146, 150] are also indicated,

which were comparable to the Γic values in this study when considering the phase angle effect

in different test methods. There are two noteworthy characteristics of the toughness values of

the two different TBCs. Firstly, for both TBC systems, the interfacial toughness decreased with

increasing oxidation, indicating that the bond coat/TGO interface was weakened by oxidation.

This behaviour agrees with some previous studies [146, 188]. However, it is noticed that a

recent study [132] has found that no decrease of interfacial fracture toughness with thermal

cycles up to 100 cycles at 1150 °C. The cause of this difference between different test methods

is unclear at this stage and further work is needed to clarify this. Another feature is that although

the mode I interfacial toughness of as-received TBCs were very similar for the two substrates,

the interfacial toughness of N5 samples decreased much faster with the oxidation time than that

of the X4 samples. For instance, the mode Ⅰ toughness of TGO/bond coat interface for the as-


129

received N5 specimen was ~ 33 J/m2 and it dramatically decreased to ~ 12 J/m2 after 15 h

oxidation. While for TBCs with X4 substrates, the interfacial toughness for as-received

samples was ~ 30 J/m2 and reduced to ~ 21 J/m2 after 15 h. The reason for this will be given in

Section 4.4.2.

Fig. 4.11 Mode Ⅰ interface toughness of TBCs on N5 (black square) and X4 substrates (red

circle) as a function of oxidation time, respectively. Data of other TBC systems are also

included for comparison: Pt diffusion bond coat (1150 °C, X. Wang et al. [132]); NiCoCrAlY

bond coat (Yu-Fu Liu et al. [150]); β-NiPtAl bond coat (Vasinonta and Beuth [146]). Note that

the data in all references has been replotted by considering the different phase angles in

different test methods.

4.4.2 Interface degradation of TBCs on different substrates

In order to find out the reason for the faster degradation of TGO/bond coat interface on the N5

substrate, high resolution STEM/EDX mapping analysis was carried out. The results for TBCs

with the N5 and X4 substrate after 3 1-h cycles at 1200 °C are shown in Fig. 4.12 and 4.13,

respectively. As can be seen from Fig. 4.12 a, pre-failure cracking was observed for TBC on


130

the N5 substrate. Moreover, sulfur was found to segregate at the bond coat/TGO interface, as

shown by Fig. 4.12 c. This sulphur segregation has been widely reported to be detrimental to

the coating performance because sulphur can significantly degrade the interfacial toughness

[74, 199]. In contrast, no cracks existed on the specimen for X4 substrate after 3 cycles (Fig.

4.13 a), and no sulfur segregation was observed at the TGO/bond coat interface (Fig. 4.13 c).

The STEM/EDX results can explain the faster degradation of TGO/bond coat interface

toughness on the N5 substrate, compared to that of X4 substrate [200]. Further research is

required to find out the reason for this different sulphur segregation behaviour for these two

superalloy substrates.


131

Fig. 4.12 High resolution STEM/EDX analysis of the TGO/bond coat interface: a) and b)

HAADF (high angle angular dark field)/STEM image, b) is the red box area shown in a); c) -

g) STEM/EDX mapping of TGO on N5 substrate after 3 cycles at 1200 °C.

Fig. 4.13 High resolution STEM/EDX of the TGO/bond coat interface: a) and b)

HAADF/STEM image, b) is the red box area shown in a); c) - g) STEM/EDX mapping of TGO

on X4 substrate after 3 cycles at 1200 °C.

H. M. Tawancy et al. [180] pointed out that for TBCs on Ti-containing substrates, the

interfacial adhesion can be degraded by the segregation of TiO2 particles at the TGO/bond coat


132

interface, resulting in poor TBC performance. However, in this study no Ti-oxide was found

within the TGO or at the bond coat/TGO interface for the Ti-containing CMSX-4 substrates.

This indicates that Ti in the superalloy substrate is not necessarily detrimental to the

performance of Pt-containing bond coats, as also suggested by other researchers [167, 201]. It

has been proposed that Pt in the γ/γ’ bond coat can serve as a sink for Ti [202], which might

explain this annihilation of Ti precipitates observed here.

4.5 Summary

In this contribution, cyclic oxidation tests at 1200 °C have been carried out on TBCs with the

CMSX-4 and the René N5 substrate, respectively. And a strain-to-fail method combined with

the 3D-DIC technique was employed to measure the bond coat/TGO interface toughness of

TBCs with different substrates. Main findings can be summarized as follows:

• TBCs based on the CMSX-4 superalloy have a 25% longer average lifetime compared

to that on the René N5 superalloy during the cyclic test. No significant difference

regarding interfacial morphology and TGO growth rate has been observed for these

TBCs with different substrates until failure of the coating.

• Spallation of the TBC occurred mainly along the bond coat/TGO interface for TBC

with the René N5 substrate, whereas for TBC with the CMSX-4 substrate, a mixed

failure path (along the bond coat/TGO interface and within TGO) is observed.

• The strain-to-fail test showed that although the mode I interfacial toughness (Γic) values

were almost identical for the two substrates in the as-deposited state, the interfacial

toughness of René N5 specimens decreased faster with increasing oxidation time.

• This faster degradation of TGO/bond coat interfacial toughness for the N5 substrate can

be ascribed to the sulfur segregation at this interface, which has been confirmed by high

resolution STEM/EDX.

CHAPTER 5 THE AL-ENRICHED γ’-BASE BOND COAT FOR TBC APPLICATIONS

133

Chapter 5 The Al-enriched γ’-Ni3Al-base bond coat for thermal

barrier coating applications

5.1 Introduction

The quest for increased operating temperatures of gas-turbine engines, in order to achieve a

higher engine efficiency, has driven the development of thermal barrier coatings (TBCs) for

protections of superalloy components in the hot section of gas-turbine engines [51, 73]. A TBC

system consists of a thermally-insulating ceramic top coat, an intermediate metallic bond coat

and the underlying superalloy substrate. The bond coat has been considered as the most crucial

part of a TBC system which has two primary functions. Firstly, it can provide adhesion between

the ceramic top coat and the metallic substrate. Maintaining the adhesion of top coats to the

superalloy substrates is vital for the lifetime of superalloys in the high temperature combustion

environment [203]. Secondly, the bond coat serves as an aluminium reservoir from which the

coating can form and maintain a slow-growing α-Al2O3 thermally grown oxide (TGO) layer.

This TGO layer can protect the underlying superalloy substrates from being oxidized at high

temperatures.

The diffusion bond coats based on the Pt-modified β-NiAl intermetallic compound, have been

optimized to form and maintain a dense α-Al2O3 layer. This ensures the coating to provide

good oxidation resistance for the underlying superalloys. However, the Pt-modified β-NiAl

coating (denoted as the β-NiPtAl coating hereinafter) exhibits inferior mechanical properties,

especially at high temperatures. For example, its strength decreases sharply at high


134

temperatures, thus makes it vulnerable to the deformation (referred to as rumpling) during

cyclic oxidation [204]. Rumpling of the bond coat surface can cause the detachment of the top

coats from the bond coat, leading to the subsequent spallation. Because the ceramic top coat

has low out-of-plane compliance which prevents it from deforming together with the bond coat

[203].

Another type of diffusion bond coat, the Pt-diffused γ-Ni/γ’-Ni3Al coating, has attracted

increasing attention. Early work by Tatlock et al. [205] confirmed that Pt can improve the

oxidation resistance of alloys with γ+γ’ compositions. Recent progress [175, 206] showed that

the Pt-diffused γ/γ’ coating exhibits a significant reduction of rumpling after high temperature

cyclic exposures, compared to the β-NiPtAl coating. This benefit can be ascribed to the high

strength of the γ’- Ni3Al structure at elevated temperatures [207, 208]. Despite these advantages,

the low Al content (~ 20 at. %) of the Pt-diffused γ/γ’ coatings has raised concerns regarding

their long-term oxidation behaviour [209]. Several studies [30, 52, 210] have reported the

spinel (NiAl2O4) formation in the TGO of this coating during prolonged oxidation because of

the insufficient Al source. The brittle spinel can significantly compromise the interface

adhesion between the TGO and the top coat, leading to spallation of the top coat.

Overall, there are both advantages and disadvantages for the two types of Pt diffusion coatings.

Although the β-NiPtAl coating can provide effective oxidation protection, it degrades

mechanically by rumpling during cyclic oxidation. The Pt-diffused γ/γ’ coatings exhibit

excellent rumpling resistance during cyclic exposures but has inferior oxidation resistance due

to low Al contents. Another critical issue with these two diffusion coatings is the coating-

substrate interdiffusion, which can also result in coating failure [74]. Pt addition was initially

intended to mitigate this interdiffusion and promote the outward Al diffusion to the


135

coating/TGO interface [43]. However, Chen and Little [211] have studied the degradation of

β-NiPtAl coatings on the CMSX-4 substrate and found that Pt neither inhibited coating-

substrate interdiffusion nor promoted the outward Al diffusion. Other studies have also

reported that Al depletion due to interdiffusion can result in the coating degradation for the Pt-

diffused γ/γ’ coatings [30].

Therefore, the objective of the present study is to fabricate a Pt diffusion coating that can resist

high-temperature rumpling while maintaining adequate oxidation resistance and exhibiting

mitigated coating-substrate interdiffusion. The proposed Al-enriched pure γ’-phase coating

(denoted as the Al-enriched γ’-phase coating) not only maintains the rumpling resistance of the

γ’-Ni3Al structure, it also shows comparable oxidation performance to the β-NiPtAl coating.

In addition, this new γ’-phase coating exhibits much less pronounced coating-substrate

interdiffusion during oxidation compared to the γ/γ’ two-phase coating, which contributes to

its superior TGO spallation resistance. There are two steps for fabricating the Al-enriched pure

γ’-phase coating. Firstly, a Pt-diffused intermediate coating will be fabricated on the CMSX-4

superalloy by a selective γ-etching process and subsequent Pt electroplating [24]. Then a pack

cementation aluminizing process will be carried out on this Pt-diffused intermediate coating to

obtain the Al-enriched pure γ’-phase coating. In addition, industry-standard Pt-diffused γ/γ’

coatings and β-NiPtAl coatings will be investigated and compared to this new Al-enriched γ’-

phase coating in terms of both isothermal and cyclic rumpling behaviours. The focus of this

manuscript will be on the coating-substrate interdiffusion, TGO microstructure & spallation

and the rumpling behaviour of these coatings.


136


5.2.1 Sample preparation

The Al-enriched γ’-phase coatings were fabricated at the University of Manchester. Standard

Pt-diffused γ/γ’ coatings and β-NiPtAl coatings were supplied by Rolls-Royce®. The next

section will describe the fabrication of Al-enriched γ’-phase coatings in details, followed by a

brief introduction of the fabrication of the industry-standard coatings.

5.2.1.1 Fabrication of Pt-diffused intermediate coatings

The Pt-diffused intermediate coatings were applied to CMSX-4 single crystal Ni-based

superalloy (Table 5.1, Rolls-Royce plc) substrates. Cylindrical CMSX-4 bars (20 mm in

diameter) with the <001> orientation aligned with the cylinders’ long axis were cut into buttons

of 5 mm thickness by a SiC blade using a precision machine (Accutom 5, Struers). All buttons

were ground using 400# SiC paper, and then cleaned by acetone in an ultrasonic bath for 20

min. Subsequently, the buttons were selectively electrolyte etched to remove the γ matrix of

the single crystal superalloy. The selective electrolyte γ-etching parameters are shown in Table

5.2 [24]. Fig. 5.1 exhibits a) the as-etched cross section by the focus ion beam (FIB; FEI Quanta

3D) system coupled with SEM, b) the as-etched surface morphology and c) etching time -

thickness plot, respectively. It can be seen from a) and b) that γ phases have been totally

removed and the cuboidal γ’ precipitates were retained after the selective γ-etching. Fig. 5.1 c)

shows a linear relationship of etching time vs. thickness (~ 5.0 µm/min).


137

Table 5-1 Composition of CMSX-4 superalloy

Element Ni Al Cr Co Ta Ti W Re

Wt. % 61.4 5.6 6.4 9.6 6.6 1.0 6.4 2.9

Fig. 5.1 a) cross-sectional FIB/SEM image and b) surface image of as-etched samples; c)

etching time-thickness plot and d) cross-sectional FIB/SEM image after Pt electroplating on

the etched substrate.

The as-etched substrates were Pt-electroplated where Pt filled the gaps left by removing the γ

matrix and then formed an overlay layer above the sample surface (Fig. 5.1 d). After Pt

electroplating, all samples were immersed in hot distilled water (80°C, 1h) to remove the

remaining salts (from the electroplating bath) on the sample surface, followed by vacuum heat

treatment at 1150°C for 2 h to form the Pt-diffused intermediate coatings.


138

Table 5-2 Electrolyte γ-etching bath parameters

Chemical formula 10 vol.% H3PO4 + 90 vol.% H2O

pH 0.04-0.06

Temperature Room temperature

Cathode current density 120 mA/cm2

Magnetic stirring speed 120/min

5.2.1.2 Fabrication of Al-enriched γ’-phase coatings by pack cementation

A low-temperature pack cementation aluminizing process was carried out on the Pt-diffused

intermediate coatings to increase the Al content. Firstly, the powder mixture of aluminium (The

Aluminium Powder Co. Ltd), CrCl3 activator (Sigma-Aldrich) and Al2O3 inert filler powder

(Honeywell FlukaTM) were manually ground with a mortar and a pestle, followed by

mechanical mixing of the powder mixture for 4 h. Then the Pt-diffused intermediate coating

samples were buried in a crucible with the powder mixture and sealed by the cement. The

following heat treatment was conducted in a tube furnace with Ar gas flowing at temperatures

650 to 950 °C to aluminize. Finally, samples were annealed in vacuum at 1150°C for 4 h to

form the Al-enriched γ’-phase coatings. Different pack cementation parameters (powder

mixture composition, heat treatment temperature & time, etc.) were investigated in order to

obtain Al-enriched pure γ’-phase coatings, which will be detailed in section 5.3.1. The

fabrication process of the Al-enriched γ’-phase coating is summarized in Fig. 5.2. It is noted

that the average Al concentration of the Pt-diffused intermediate coating (~ 10 - 12 at. %) is

lower than that of the Pt-diffused γ/γ’ coatings (~ 16 - 19 at. %) due to the etching process

which removes all Al contents of the γ phase, thus ‘diluting’ the Al content of the coating.


139

Fig. 5.2 Two steps to fabricate the Al-enriched pure γ’-phase coating: Ⅰ. Fabricate Pt-diffused

intermediate coatings; Ⅱ. Pack cementation aluminizing on the intermediate coating.


140

The two industry-standard coatings were also applied to the CMSX-4 substrates. The Pt-

diffused γ/γ’ coatings were produced on grit blasted substrates by Pt electroplating followed

by a diffusion anneal in vacuum [196], and β-NiPtAl coatings were produced by Pt

electroplating followed by a low-activity chemical vapour deposition (CVD) aluminizing [35].

5.2.2 Thermal treatment

Isothermal oxidation was performed at 1150ºC in laboratory air in a box furnace (CM™). The

samples were oxidized for different periods of time up to 100 h, followed by air quenching.

Cyclic oxidation was performed in laboratory air between room temperature and 1150°C in a

cycle furnace (CM™) in order to investigate the rumpling behaviour of the coatings. Each

cycle consisted of 10 min ramping period, 10 min holding time at 1150°C and 10 min fan-

assisted air quenching.

5.2.3 Characterization methods

To identify the as-diffused coating phases, electron backscatter diffraction (EBSD,

NordlysNano, Oxford Instruments) was used in addition to X-ray diffraction (XRD, Philips

X'Pert) with Cu Kα radiation (λ=0.154 nm).

After oxidation, the optical microscope (Olympus BH2-UMA) was used to examine the surface

morphology of the oxides. XRD with Cu Kα radiation (λ=0.154 nm) was used to examine the

oxide phases. Then the site-specific cross-sectional images of the oxides were captured by SEM

coupled with FIB (FEI Quanta 3D). Finally, the cross-sections of the samples were ground by

SiC paper up to 1200# and then polished using diamond paste to 0.25 μm finish. SEM (FEI

Quanta 650) equipped with an energy dispersive X-ray spectrometer (EDS) was used to

examine the cross-sections for elemental diffusion profiles of the coatings.


141

To investigate the rumpling behaviour of the coatings, firstly, Vickers micro-hardness

indentations (Duramin hardness tester, Struers) were placed into the surfaces of each coating

before the cyclic oxidation. These indentations can act as markers so that the rumpling

evolution of the same surface area on each sample could be tracked during the cyclic oxidation.

The coatings were removed from the cyclic furnace at specific cycles for surface roughness

characterization using an optical profilometer (Bruker). With the help of the indentation

markers, the digital images of the same area in the form of surface height, Z, as a function of

position x and y were recorded for each coating during the cyclic oxidation. The surface

rumpling magnitude was characterized by the root mean square roughness Rq:

Rq= √1

𝑛∑ (𝑍𝑖− 𝑍)

2𝑛𝑖=1 (5.1)

where Zi is the height of each point, 𝑍̅ is the average height of all points and n is the number of

total points.

5.3 Results

5.3.1 Synthesis of Al-enriched γ’-phase coatings by pack cementation

To find the appropriate pack cementation parameters for obtaining the Al-enriched γ’-coatings,

different pack cementation parameters have been tried with the guidance of Ni-Pt-Al phase

diagram to find the target concentration for the Al-enriched γ’-phase coating. Different pack

cementation parameters (powder composition, temperature and holding time) of 5 trial samples

(number 1-5) are given in Table 5.3. The coating compositions corresponding to sample 1-5

were measured by SEM/EDX and Table 5.3 lists the average Al, Pt and Ni concentrations of

the as-fabricated coating for each sample.


142

Table 5-3 Different pack cementation parameters and the resulting coating composition by

EDS after vacuum anneal

Sample number 1 2 3 4 5

Powder

composition

(wt. %)*

CrCl3 (4.0)

+ Al (5.0)

CrCl3 (4.0) +

Al (5.0)

CrCl3 (1.0) +

Al (1.0)

CrCl3 (1.0) +

Al (1.0)

CrCl3 (1.0) +

Al (1.0)

Temperature

(°C)

950 800 800 650 650

Hold time at

temperature

(min)

<0.5 <0.5 <0.5 15 5

Average Al

concentration

(at.%) of γ’-

coating by EDS

35.9 23.3 31.0 33.5 19.2

Ni concentration

(at.%) of γ’-

coating by EDS

26.3 32.6 30.5 29.7 40.1

Pt concentration

(at.%) of γ’-

coating by EDS

29.8 29.5 30.6 31.2 29.2

*Al2O3 powder with balance weight.

Fig. 5.3 exhibits the Ni-Pt-Al phase diagram at 1150°C [202], and different dots represent the

coating compositions of sample 1-5. Compositions of the sample 1-4 are in the three-phase

region (Fig. 5.3) including the β-NiAl phase, which indicates that these samples were over-

aluminized. The inset SEM image in Fig. 5.3 shows the coating microstructure of the sample

3. It can be seen that the interdiffusion zone of this coating had some precipitates, which is

characteristic for a β-NiAl-base diffusion coating [108]. Sample 5 (aluminized below 700°C)

located at the γ-Ni+α-NiPt two-phase region, indicating this sample was under-aluminized. In

order to find the target concentrations for the Al-enriched γ’-phase coating, the Ni-Pt-Al phase


143

diagram at the annealing temperature (1150°C) [202] was utilised, as shown in Fig. 5.4. Since

the Pt electroplating process was identical for all samples, the coatings exhibited similar Pt

concentration (~ 29.0 - 32.0 at. %), as presented by the two blue lines in Fig. 5.4. The region

defined by the two blue lines has intersected with the pure γ’-phase region, thus the upper and

lower limit of the target Al concentration for the Al-enriched pure γ’-phase coating can be

determined by the highest and lowest point of this intersection region, respectively. The upper

(~ 27 at. %) and lower (~ 22 at. %) limits of the target Al concentration for the Al-enriched

pure γ’-phase coatings were marked by the two horizontal red lines in Fig. 5.4. Recalling the

Al concentrations of different packs in Table 3, the upper and lower Al concentrations located

between that of sample 4 and sample 5. Therefore, a holding time between sample 4 and sample

5 can fabricate the Al-enriched pure γ’-phase coating.

Fig. 5.3 Ni-Pt-Al phase diagram at 1150°C [202]. The compositions of sample 1-5 are marked

by the different dots, respectively. The inset SEM image shows the as-fabricated cross-sectional

microstructure of sample 3.


144

Fig. 5.4 Ni-Pt-Al phase diagram at 1150°C [202]. The two horizontal red lines represent the

upper and lower limit of Al concentration for the pure γ’-phase coating, respectively.

5.3.2 Microstructure of the as-received coatings

By 10 min holding time at 650 °C and the same pack powder composition as sample 4, the Al-

enriched pure γ’-phase coating was fabricated. The average Ni, Pt and Al concentration of this

coating is shown by the red triangle in Fig. 5.4, which is located in the single γ’-phase region.

Fig. 5.5 shows the XRD results of the three as-fabricated coatings. The γ phase is a disordered

face centre cubic (fcc) solid solution, and the γ’ phase is an ordered fcc crystal structure (L12

superlattice). The β-NiAl phase is an ordered body centred cubic (bcc) structure (B2

superlattice). According to Fig. 5.5, the XRD patterns for as-fabricated Al-enriched γ’-phase

coating and the Pt-diffused γ/γ’ coating were quite similar. Because the diffraction peaks of γ

and γ’ locate very close to each other except for the additional peaks of γ’. It is noted that there

was no β-phase in the Al-enriched γ’-phase coating, which confirms that the pack cementation


145

process was not over-aluminized. In addition, the pattern of the β-NiPtAl coating only showed

β peaks, while the other two coatings also exhibited peaks of Pt.

Fig. 5.5 XRD patterns of as-fabricated Al-enriched γ’-phase coatings, Pt-diffused γ/γ’ and β-

NiPtAl coatings.

Fig. 5.6 a - c exhibit the cross-sectional SEM (backscattered electron, BSE) images of three as-

fabricated coatings. Only γ’-phase presented in the Al-enriched γ’-phase coating (Fig. 5.6 a),

whereas γ (dark contrast) and γ’ phase (bright contrast) were intermixed with each other in the

Pt-diffused γ/γ’ coating (Fig. 5.6 b). The β-NiPtAl coating exhibited an outer β-NiPtAl layer

and an inner interdiffusion zone (IDZ) (Fig. 5.6 c). A large number of white contrast

precipitates rich in refractory elements (Ta, Mo, W) can be seen, especially in the IDZ [185].

Moreover, numerous pores were observed in the Al-enriched γ’-phase coating, especially in

the area near the surface, while the other two coatings exhibited a dense microstructure. The

formation of these pores can be explained by the Kirkendall effect as a result of the

interdiffusion of Al, Ni and Pt during the fabrication process.

The EBSD phase mappings of each coating are shown in Figs. 5.6 d - i. Fig. 5.6 d confirmed

that no β or martensitic phases were observed in the Al-enriched γ’-phase coating. A previous


146

attempt to aluminize the Pt-diffused γ/γ’ based coatings on nickel superalloys has failed due to

the over-aluminizing resulting in the formation of β or martensitic phase on the coating surface

[59]. However, this study suggests that the low-temperature pack cementation process is

capable of aluminizing the γ’-base coating without altering the γ’-phase microstructure. It is

also noted that the lower part of the Al-enriched γ’-phase coating had the same crystal

orientation as the superalloy substrate (Fig. 5.6 g), indicated limited Pt solid solution in this

lower part. The grains of both the Pt-diffused γ/γ’ coating (Fig. 5.6 h) and β-NiPtAl coating

(Fig. 5.6 i) showed random orientations. In addition, it was observed that the average grain size

of the Pt-diffused γ/γ’ coating and the Al-enriched γ’-phase coating was comparable, while the

grain size of the β-NiPtAl coating was much smaller compared to the other coatings, especially

the surface part (nano-grains). This can be attributed to the large amount of recrystallization

during the coating manufacturing process of the β-NiPtAl coating.


147

Fig. 5.6 Microstructure of as-fabricated coatings: SEM (BSE) images of: a) γ’ coating, b) γ/γ’

coating and c) β-NiPtAl coating; EBSD phase contrast map of the red box area in a - c: d) γ’

coating, e) γ/γ’ coating and f) β-NiPtAl coating; and corresponding color-coded inverse pole

figure (IPF) mapping g) - i) showing the different grain sizes of three coatings.

5.3.3 Isothermal oxidation performance of three Pt-diffused coatings

5.3.3.1 Elemental diffusion of three coatings

The elemental diffusion (Ni, Al and Pt) profiles of three as-fabricated coatings during

isothermal oxidation are shown in Fig. 5.7 - 5.9. For the Pt-diffused γ/γ’ coating, Pt showed

the same fluctuating trend as Al along the distance from the coating surface to the inner

superalloy, whereas Ni showed a contrary fluctuating trend (Fig. 5.8 a), indicating different

partition behaviours of these elements in γ and γ’ phase. Conversely, Pt (and Al) was uniformly

distributed throughout the Al-enriched γ’-phase coating (Fig. 5.7 a), which again confirmed the

pure γ’-phase microstructure. The Al distribution was relatively uniform throughout the β-

NiPtAl coating except for some small bumps (Fig. 5.9 a), which is attributed to the numerous

refractory precipitates (as also shown in Fig. 5.6 c). In summary, for the as-received coatings,

the average Al concentration: Pt-diffused γ/γ’ coating (~ 16.9 at. %) < the Al-enriched γ’-phase


148

coating (~ 26.7 at. %) < β-NiPtAl coating (~ 41.8 at. %). While the Pt concentration of the β-

NiPtAl coating was considerably lower than that of other two coatings, which aligns with a

previous study [212].

Fig. 5.7 Ni, Pt and Al concentration evolution of the Al-enriched γ’ coating by EDX linescan

after a) 0 h, b) 20 h and c) 50 h oxidation.

After 20 h isothermal oxidation, the Pt and Al concentrations of the Al-enriched γ’-phase

coating near the surface only showed a little drop compared to that of as-fabricated coatings

(e.g. Pt concentration dropped from ~ 32 at. % to ~ 28 at. %, Fig. 5.7 a and b). On the other

hand, the Pt-diffused γ/γ’ coating exhibited a progressive reduction of both Pt and Al

concentrations near the surface, compared to the as-fabricated coating (Pt dropped from ~ 40

at. % to ~ 10 at. %, Fig. 5.8 a and b). Moreover, Pt concentration of the β-NiPtAl coating was

c)

a) b)

Ni

Al

Pt


149

severely depleted after 20 h oxidation (Fig. 5.9 b) to an average concentration of ~ 5.0 at. %.

After 50 h oxidation, the β-NiPtAl coating showed the most severe Pt (and Al) depletion among

the three coatings. For example, Pt has diffused to a maximum depth of ~ 100 μm (the red

arrow in Fig. 5.9 c) for this coating. Conversely, for the Al-enriched γ’-phase coating, Pt only

diffused to a maximum depth of ~ 30 μm after 50 h (the red arrow in Fig. 5.7 c) and both Al

and Pt still remained high concentrations near the coating surface. In a word, the Pt and Al

interdiffusion with the substrate were much less pronounced for the Al-enriched γ’-phase

coating compared to the other coatings during isothermal oxidation.

Fig. 5.8 Ni, Pt and Al concentration evolution of the Pt-diffused γ/γ’ coating by EDX linescan:

a) 0 h, b) 20 h and c) 50 h oxidation.

b) a)

c)

Ni

Pt

Al


150

Fig. 5.9 Ni, Pt and Al concentration evolution of the β-NiPtAl coating by EDX linescan: a) 0

h, b) 20 h and c) 50 h oxidation.

5.3.3.2 Oxide microstructure and growth kinetics

Fig. 5.10 shows the oxide phases (by XRD) on three coatings after different oxidation time at

1150 °C. Both α-Al2O3 and spinel (NiAl2O4) grew on the Pt-diffused γ/γ’ coating after 50 h

oxidation (Fig. 5.10 b). However, for the Al-enriched γ’-phase coating, spinel was only

identified after 100 h oxidation (Fig. 5.10 a). For the β-NiPtAl coating, only α-alumina was

detected in the oxide scale up to 100 h oxidation (Fig. 5.10 c). This demonstrates that although

the Al-enriched γ’-phase coating did not outperform the β-NiPtAl coating regarding the

selective oxidation of aluminium due to the much lower Al concentration (26.7 at. % compared

to 41.8 at. % for both as-fabricated coatings), it exhibited a better oxidation performance by

retarding Ni-oxide growth compared to the Pt-diffused γ/γ’ coating.

b)

c)

a)

Pt

Ni

Al


151

Fig. 5.10 Glancing angle (3°) XRD patterns of the oxides on a) γ’ coating, b) γ/γ’ coating and c) β-NiPtAl coating after different oxidation time at 1150 °C.

a)

c)

b)


152

Fig. 5.11 a) - c) cross-sectional SEM images of three coatings after 20 h oxidation at 1150 °C;

d) the magnified image of the red box area in c).

Fig. 5.11 shows the cross-sectional BSE images of the three coatings after 20 h isothermal

oxidation at 1150 °C. For the Al-enriched γ’-phase coating and the β-NiPtAl coating (Fig. 5.11

a and c), a pure α-Al2O3 layer was observed after 20 h oxidation, which agrees well with XRD

results (Fig. 5.10 a and c). A large number of alumina particles were observed near the

TGO/coating interface in the Al-enriched γ’-phase coating (red arrows in Fig. 5.11 a) because

of the internal oxidation. For the Pt-diffused γ/γ’ coating, local spinel formation above the α-

Al2O3 layer was observed, as shown by the red arrows in Fig. 5.11 b. It is noticed that XRD

did not detect any spinel in the oxide scale of the Pt-diffused γ/γ’ coating after 20 h oxidation

(Fig. 5.10 b), which is likely due to the small amount of spinel at this stage. Moreover, as can

be seen in Fig. 5.11 a, the Al-enriched γ’-phase coating remained γ’-phase structure near the


153

coating surface after 20 h oxidation, as confirmed by EDX spectrum. The Pt-diffused γ/γ’

coating, on the other hand, exhibited a progressive γ’ to γ phase transformation near the

TGO/coating interface (Fig. 5.11 b), which is attributed to the Al depletion resulting from TGO

growth and interdiffusion with the substrate [185]. The β-NiPtAl coating maintained the β-

phase structure, but occasionally, γ’ precipitates can be seen at the grain boundaries of the β-

NiPtAl coating (red arrows in Fig. 5.11 d), in addition to some refractory metal precipitates

(black arrows in Fig. 5.11 d) [122].

Fig. 5.12 shows the TGO morphology of three coatings after isothermal oxidation for 50 h. At

this stage, some spinel formed locally on top of the continuous alumina layer on the Al-enriched

γ’-phase coating, as indicated by the red arrows in Fig. 5.12 a. And for this coating, no γ’ to γ

transformation took place near the TGO/bond coat interface. The TGO scale on the Pt diffused

γ/γ’ coating was essentially duplex in structure after 50 h (Fig. 5.12 b), with outer spinel layer

and inner α-Al2O3 layer. A thin layer consisting of predominantly γ-phase existed just below

the bond coat surface, which is due to γ’ to γ transformation as a result of Al depletion. For the

β-NiPtAl coating, the TGO scale still consisted of pure alumina after 50 h oxidation.

Furthermore, this coating maintained β-phase structure except for a small amount of γ’-phase

at the coating grain boundaries (Fig. 5.12 c). These observations were also consistent with XRD

results in Fig. 5.10.


154

Fig. 5.12 Cross-sectional SEM backscattered electron (BSE) images of three coatings after 50

h oxidation at 1150 °C.

Fig. 5.13 Oxide thickness vs. isothermal oxidation time (at 1150 °C) for the three bond coat

systems.


155

The TGO thickness vs. isothermal oxidation time are plotted in Fig. 5.13 for the three coatings.

The TGO on the Pt-diffused γ/γ’ coating grew much faster than the other two coatings up to

100 h oxidation time, which is attributable to the spinel formation of this coating. The Al-

enriched γ’-phase coating exhibited a slightly higher TGO growth rate, compared to the β-

NiPtAl coating due to the local spinel formation after 50 h oxidation (Fig. 5.12 a).

5.3.3.3 TGO spallation

Fig. 5.14 gives the optical surface images of three coatings after different isothermal oxidation

time at 1150 °C. The TGO scale remained intact for the Al-enriched γ’-phase coating and the

β-NiPtAl coating after 100 h oxidation (Fig. 5.14 a and c). Conversely, the oxide scale on the

Pt-diffused γ/γ’ coating exhibited noticeable spallation (bright areas in Fig. 5.14 b) after 50 h

oxidation. This indicates that the TGO spallation resistance of this new Al-enriched γ’-phase

coating is comparable to the conventional β-NiPtAl coating. Moreover, the Al-enriched γ’-

phase coating has exhibited significant improvement in terms of TGO lifetime compared to the

Pt-diffused γ/γ’ coating.


156

Fig. 5.14 Optical surface images of three coatings after different isothermal oxidation time at

1150 °C.

5.3.4 Rumpling behaviour of three bond coats under cyclic oxidation

Fig. 5.15 - 5.17 illustrate the evolution of surface topography of three coatings under cyclic

oxidation, respectively. The surface topography recorded from an identical region of the Al-

enriched γ’-phase coating did not show significant changes from as-received condition up to

50 10-min cycles (Fig. 5.15 a - e). For detailed evaluation of the surface topography, the

roughness profiles along a line segment (the white line in Fig. 5.15 a) across the probed region

have been recorded at different stages of cyclic oxidation, as shown in Fig. 5.15 f. From the

statistical analysis, the roughness parameter for this line remained nearly constant during cyclic

oxidation: Rq (root mean square of roughness, calculated by Eq. 5.1) = 1.4 μm after 5 cycles

and Rq = 1.9 μm after 50 cycles. The small increase of Rq (~ 0.5 μm) was within the margins


157

of measurement error. The Pt-diffused γ/γ’ coating exhibited similar behaviours, as illustrated

in Fig. 5.16. The surface topography for the as-received coating (Fig. 5.16 a) was almost the

same as that after 50 10-min cycles (Fig. 5.16 e). This is further confirmed by the roughness

profile evolution of a line segment (the white line in Fig. 5.16 a), as shown in Fig. 5.16 f where

Rq only slightly increased from 2.9 μm (5 cycles) to 3.4 μm (50 cycles). These investigations

indicate that both the Al-enriched γ’-phase coating and the Pt-diffused γ/γ’ coating did not

show detectable rumpling during the cyclic oxidation. Other researchers have also concluded

that Pt-diffused γ/γ’ coatings were resistant to the rumpling deformation during cyclic

oxidation [59, 213].

Conversely, the profilometer images recorded from an identical region of the β-NiPtAl coating

showed progressive changes during cycling test (Fig. 5.17). Clearly this coating has

demonstrated a tendency to roughen with thermal cycling. In addition, once rumpling was

initiated, the regions above the average surface persisted to bow up, whereas the regions below

the average surface continued to depress down with further cycling. This observation is in

agreement with the result reported by Chen et al. [118] on the NiCoCrAlY bond coat. Fig 5.17

f exhibits the roughness profile evolution of the white line in Fig. 5.17 a. These profiles clearly

showed that the corresponding positions along this line have moved either up or down

progressively with respect to the original surface plane. As a result, the roughness Rq has

increased from 0.9 μm after 5 cycles to 2.3 μm after 50 cycles. This suggests that the β-NiPtAl

coating showed significant rumpling increment during cyclic oxidation, which agrees with

some previous studies on β-NiPtAl bond coats [121, 206, 214].


158

Fig. 5.15 Profilometer images of (a) as-fabricated Al-enriched γ’-phase coating surface and

after (b) 5 10-minute cycles, (c) 10 cycles, (d) 25 cycles and (e) 50 cycles recorded at the

identical area; (f) the surface profiles and the corresponding Rq of the line shown in (a).

d

)

c

)

e 50 × 10-minute

10 × 10-minute 25 × 10-minute

f

a

)

b

)

5 × 10-minute As-received

Profile line in Fig. (f)


159

Fig. 5.16 Profilometer images of (a) as-fabricated Pt-diffused γ/γ’ coating surface and after (b)

5 10-minute cycles, (c) 10 cycles, (d) 25 cycles and (e) 50 cycles recorded at the identical area;

(f) the surface profiles of the line shown in (a).

As-fabricated

e

)

d

) c

)

b

)

a

)

f

50 × 10-minute

25 × 10-minute

5 × 10-minute

10 × 10-minute


160

Fig. 5.17 Profilometer images of (a) as-fabricated β-NiPtAl coating surface and after (b) 5 10-

minute cycles, (c) 10 cycles, (d) 25 cycles and (e) 50 cycles recorded at the identical area; (f)

the surface profiles of the line shown in (a).

e

)

d

) c

)

50 × 10-minute

25 × 10-minute 10 × 10-minute

5 × 10-minute As-fabricated

f

a

) b

)


161

5.4 Discussion

5.4.1 Pt and Al depletion of three coatings

The elemental diffusion profiles in Section 5.3.3.1 show that Pt depletion (due to interdiffusion

with the substrate) was more severe in the Pt-diffused γ/γ’ coating than that in the Al-enriched

γ’-phase coating during isothermal oxidation. In other words, Pt can remain relatively stable in

the pure γ’-phase microstructure. This is in agreement with Bai’s work [24], which has shown

that Pt is more stable in the γ’ phase than in the γ phase by thermodynamic calculations.

As for the Al depletion, firstly, the β-NiPtAl coating had the most severe Al depletion due to

interdiffusion with the substrate. While the Pt-diffused γ/γ’ coating exhibited a reduction of Al

depletion compared to the β-NiPtAl coating. Because the Pt-diffused γ/γ’ coating has the

chemical compatibility with the substrate that can mitigate the interdiffusion, as also reported

in [209]. Furthermore, the Al-enriched γ’-phase coating has experienced even less Al depletion

compared to the Pt-diffused γ/γ’ coating. This can be interpreted by the comparison of the local

interdiffusion flux of the Al-enriched γ’-phase coating/CMSX-4 diffusion couple and the Pt-

diffused γ/γ’ coating/CMSX-4 couple at specific positions during the oxidation. For simplic ity ,

we considered the Ni-Pt-Al ternary system, in which Ni was taken to be dependent and Al and

Pt were independent variables. Thus, the local interdiffusion flux of Al (𝐽𝐴𝑙) can be described

by Fick’s first law in terms of the two independent concentration gradients [54]:

𝐽𝐴𝑙 = −𝐷𝐴𝑙𝐴𝑙𝑁𝑖 𝜕𝐶𝐴𝑙

𝜕𝑥 − 𝐷𝐴𝑙𝑃𝑡

𝑁𝑖 𝜕𝐶𝑃𝑡

𝜕𝑥 (5.2)

where 𝐷𝐴𝑙𝐴𝑙𝑁𝑖 is the main-term interdiffusion coefficient of Al which relates the flux of Al to its

own concentration gradient; 𝐷𝐴𝑙𝑃𝑡𝑁𝑖 is the cross-term interdiffusion coefficient which accounts

for the chemical interaction between Pt and Al; 𝜕𝐶𝐴𝑙

𝜕𝑥 and

𝜕𝐶𝑃𝑡

𝜕𝑥 are the local concentration

gradients of Al and Pt, respectively.


162

To calculate 𝐽𝐴𝑙 of the two couples at specific position 𝑥𝑖, the data for 𝐷𝐴𝑙𝐴𝑙𝑁𝑖 can be found in

[54] for different Ni-Pt-Al concentrations of the γ and γ’ phase, respectively. Specifically, 𝐷𝐴𝑙𝐴𝑙𝑁𝑖

for the Al-enriched γ’-phase coating/CMSX-4 diffusion couple can be found directly in [54],

while 𝐷𝐴𝑙𝐴𝑙𝑁𝑖 for the γ/γ’ coating was calculated according to the rule of mixture based on the

ratio of the γ and γ’ phase from the cross-sectional SEM image processing. As for the cross-

term interdiffusion coefficient, according to Hayashi et al. [54], there is no dependence of 𝐷𝐴𝑙𝑃𝑡𝑁𝑖

on the Pt and Al contents within the composition range studied (up to 25 at.% Pt addition in

the γ-Ni and γ’-Ni3Al alloys). Thus 𝐷𝐴𝑙𝑃𝑡𝑁𝑖 was taken as -2.4×10-10 and -0.9×10-10 cm2/s for the

γ and γ’ phase respectively in this study. 𝜕𝐶𝐴𝑙

𝜕𝑥 and

𝜕𝐶𝑃𝑡

𝜕𝑥 of the two couples can be calculated

from the elemental concentration profiles (Fig. 5.7 and 5.8). The measured concentration

profiles of Al and Pt were fitted using a cubic spline interpolation method, then 𝜕𝐶𝐴𝑙

𝜕𝑥 and

𝜕𝐶𝑃𝑡

𝜕𝑥

can be obtained at any 𝑥𝑖 of the diffusion couple. The interdiffusion coefficient in Eq. 5.2 has

the unit cm2/s, and the concentration has the unit mol/cm3. The elemental concentration with

the unit of atomic percent (at. %) in Fig. 5.7 and 5.8 can be converted into mol/cm3 by

introducing the molar volume (𝑉𝑚, unit is m3/mol). The average molar volume (𝑉𝑚) for the γ’

phase was estimated to be 6.83×10-6 m3/mol [215]. Since the molar volume varies very slightly

with the change of composition for the γ solid solution, the average 𝑉𝑚 for this phase was taken

as 6.68×10-6 m3/mol [215]. Then 𝑉𝑚 for the γ/γ’ coating was calculated based on the rule of

mixture. The local interdiffusion flux of Al (𝐽𝐴𝑙, calculated by Eq. 5.2) at specific positions

(𝑥𝑖=1, 5, 10 and 20 μm respectively) for these two couples after 20 h oxidation at 1150 ℃ is

compared in Fig. 5.18.

As can be seen from Fig. 5.18, the arrows indicate the direction of the local interdiffusion flux.

‘←’ represents the uphill diffusion from the inner part to the coating/TGO interface, in which


163

the interdiffusion flux of Al (𝐽𝐴𝑙) has the opposite sign with the Al concentration gradient (𝜕𝐶𝐴𝑙

𝜕𝑥).

‘→’ represents the inward diffusion from the coating to the inner part of superalloy, in which

the interdiffusion flux has the same sign with the concentration gradient. Fig. 5.18 showed that

for 20 h oxidation, the Al-enriched γ’-phase coating exhibited the uphill Al diffusion at 𝑥𝑖=1,

5 and 10 μm, whereas the Pt-diffused γ/γ’ coating only exhibited uphill Al diffusion at 𝑥𝑖=1

μm. The cross-term interdiffusion coefficient 𝐷𝐴𝑙𝑃𝑡𝑁𝑖 was found to be negative in sign for both

coatings, suggesting that Pt has a negative chemical interaction with Al. In other words, Pt can

reduce the chemical activity of Al and promote the uphill diffusion of Al when the

concentration gradients of Al and Pt have opposite sign in the interdiffusion zone [216]. This

uphill diffusion can mitigate Al depletion during the oxidation process. Since the Al-enriched

γ’-phase coating exhibited uphill Al diffusion to a greater extent as shown in Fig. 5.18, the Al

depletion was less pronounced for this coating. This also coincided with the significant ly

reduced Al depletion of the γ’-phase coating compared to the γ/γ’ coating, as shown in Section

5.3.3.1.

The above calculations confirmed that the new γ’-phase coating exhibited less pronounced Al

depletion due to coating-substrate interdiffusion, which is beneficial to its lifetime. However,

there are a number of sources of error with this method to calculate 𝐽𝐴𝑙. For instance, the

concentration measurements by EDX are believed to be within the accuracy of ~ + 1.0 at. %

for each element. The cubic spline can reduce the concentration fluctuations measured by EDX.

But when the concentration gradient is really low, the fitting error can be relatively significant.

The assumption that the partial molar volume for each phase is concentration-independent

within the composition range studied can also be the error source. However, the trends and

agreements with experimental results are reasonable for the 𝐽𝐴𝑙 as shown in Fig. 5.18, which

provides an illustration for the elemental diffusion evolutions in Section 5.3.3.1.


164

Fig. 5.18 The local interdiffusion flux of Al (𝐽𝐴𝑙) for the Al-enriched γ’-phase coating/CMSX-

4 diffusion couple and the Pt-diffused γ/γ’ coating/CMSX-4 diffusion couple at specific

positions (𝑥𝑖= 1, 5, 10 and 20 μm) after 20 h oxidation. The arrows represent the diffusion

direction. ←: uphill diffusion from the inner part to the TGO/coating interface; →: from the

coating to the inner part of superalloy.

5.4.2 Effect of bond coat composition on the selective oxidation of aluminium

The microstructural investigations (Section 5.3.3.2) confirmed that the oxide scale on the β-

NiPtAl coating was composed of exclusive alumina (without any Ni-oxides) up to 100 h

oxidation at 1150 °C. This is expected due to its high Al concentration (~ 41.8 at. %) which

ensures the selective oxidation of aluminium. The Al-enriched γ’-phase coating can promote

an exclusive growth of alumina and retard the Ni-oxide growth at initial oxidation compared

to the Pt-diffused γ/γ’ coating, although the Al concentration of the Al-enriched γ’-phase (~

26.7 at. %) is under the Al concentration minimum (~ 40.0 at. %) predicted in [24] for a

selective oxidation of aluminium. Two factors could be considered to explain this phenomenon.


165

First, Chen et al. [55] proposed that large number of nanostructured grain boundaries in the

surface region of the NiCoCrAlY coating can provide numerous fast diffusion paths of

aluminium at initial stage of oxidation and therefore contributes to the establishment of an

exclusive alumina scale. The Al-enriched γ’-phase coating in this study, however, did not have

any nanostructured grains in the coating surface, as can be seen from the inverse pole figure

map of the as-fabricated coating (Fig. 5.6 a). This excluded this microstructural factor for

promoting the selective oxidation of aluminium. Second, Section 5.3.3.1 has shown that the Pt

concentration of the Al-enriched γ’-phase coating sustained at a relatively high level during

isothermal oxidation, especially at the coating surface. The Pt enrichment in the coating surface

can reduce the chemical activity of aluminium at the oxide/coating surface (Section 5.4.1),

which facilitates the aluminium diffusion flux from the inner part to the TGO/coating interface,

thus promoting the exclusive growth of alumina at initial stage of oxidation [68].

5.4.3 Rumpling behaviour of three coatings

Results in Section 5.3.4 suggest that both Pt-diffused γ/γ’ coatings and the Al-enriched γ’-phase

coatings showed negligible rumpling during cyclic oxidation, whereas the β-NiPtAl coating

experienced a significant higher degree of rumpling than the other two coatings. While the

substantial rumpling deformations of the β-NiPtAl coating are expected at this temperature

(1150 °C), the mechanisms responsible for the absence of rumpling in the new Al-enriched γ’-

phase coating are worth consideration. The following sections will discuss the possible

mechanisms by introducing a classical rumpling model firstly.

5.4.3.1 Balint and Hutchinson rumpling model

Rumpling in a TBC system is characterized by viable material and geometric parameters

including TGO thickening, lateral TGO growth strain, CTE mismatches, high temperature

strength of TGO, bond coat strength & phase transformation, etc. Balint and Hutchinson (B&H)


166

[125] have proposed a comprehensive rumpling model that resolves these competing factors

and can be used to clarify the rumpling behaviour of coatings in experiments. In the B&H

model, the rumpling is driven by the lateral growth strain of TGO and occurs at a rate governed

by factors including the power-law creep of the bond coat and the plastic yielding of the TGO.

The power-law creep of the bond coat can be described by the following temperature-

dependent equation for the steady-state creep [204]:

휀̇ = 𝐶(𝜎

𝐸)𝑛exp (

−𝑄𝑐𝑟𝑒𝑒𝑝

𝑅𝑇) (5.3)

where 휀̇ is the creep rate; C is a constant; σ is the temperature-dependent equi-biaxial stress in

the coating; E is the temperature-dependent Young’s modulus; R is the gas constant and T is

the absolute temperature; 𝑄𝑐𝑟𝑒𝑒𝑝 is the activation energy for creep.

The equi-biaxial stress of the coating imposed by the coating/substrate CTE mismatch can

promote the coating creep, thus promoting the rumpling through the interaction between this

equi-biaxial stress in the coating and the normal traction imposed on the coating surface by

TGO [204]. At high temperatures, when the coating stress is relaxing, the undulation growth

of the coating is expected which is driven by the normal stress applied on the coating surface

by the compressed TGO layer. When the coating stress decays completely by creep, the

undulation growth is prohibited effectively. Because the cyclic oxidation scheme periodically

redefines the stress of the coating, the rumpling thus grows cycle-by-cycle [5].

5.4.3.2 B&H model applied to the Al-enriched γ’-phase coating

The first consideration regarding the absence of rumpling of the Al-enriched γ’-phase coating

compared to the β-NiPtAl coating is CTE mismatch. The γ’-phase coating has a smaller CTE

mismatch with the CMSX-4 substrate (CTE: ~ 15.0×10-6 °C-1 [217]) than that of the β-NiPtAl

coating [217]. This can result in the lower rumpling amplitude of the Al-enriched γ’-phase


167

coating for a given thermal history according to the B&H model [125], because CTE mismatch

also contributes to the rumpling driving force.

The second consideration is the TGO growth rate. The B&H model has proposed that coatings

with thicker TGOs tend to have a faster rumpling rate because the thicker TGO imposes a larger

traction on the coating. However, the present experiments demonstrate that the Al-enriched γ’-

phase coating has a significantly lower rumpling rate in spite of its larger TGO thickness

(Section 5.3.3.2). This discrepancy can be explained by the improved creep strength of the γ’-

phase coating, as suggested by Jorgensen et al. [203]. In their simulation work, they have shown

that increasing the coating creep strength drastically inhibits rumpling even with thicker TGOs.

These simulations corroborate with the present study: the γ’-phase coating resists rumpling to

a large degree with only a slight increase in the total amplitude over the thicker TGO layer.

5.5 Summary

An Al-enriched γ’-phase coating was fabricated on the CMSX-4 substrate by the selective γ-

phase etching and a subsequent low temperature pack cementation aluminizing process. The

elemental diffusion profiles, TGO microstructure & spallation, and rumpling of the Al-enriched

γ’-phase coating have been investigated and compared to the industry-standard β-NiPtAl

coating and the Pt-diffused γ/γ’ coating. The following conclusions can be drawn:

1. The Al-enriched γ’-phase coating exhibited a comparable TGO spallation lifetime to

the β-NiPtAl coating and a significantly improved TGO lifetime compared to the Pt-

diffused γ/γ’ coating.

2. The Al-enriched γ’-phase coating exhibited a less pronounced coating-substrate

interdiffusion (Pt and Al depletion) compared to the other two coatings, which is

beneficial to the coating lifetime.


168

3. The Al-enriched γ’-phase coating was more resistant to rumpling than the β-NiPtAl

coating, which mainly arose from the improved creep strength of the γ’-phase.

In summary, the new Al-enriched γ’-phase coating can resist high temperature rumpling

deformation while maintaining adequate oxidation properties and mitigating coating-substrate

interdiffusion, which is promising for TBC applications.

CHAPTER 6 CONCLUSIONS AND FUTURE WORK

169

Chapter 6 Conclusions and Future Work

6.1 Conclusions

Overall, this work has provided a new mechanistic understanding on the effects of the

underlying metalllic bond coat or substrate microstructure and chemical composition on the

TBC oxidation behaviour. It is concluded that the chemical composition of the underlying

coating or superalloy substrate has a direct effect on the oxidation behaviour of a TBC system,

from the early oxidation stage (Chapter 3) until the prolonged oxidation stage (Chapter 4 and

5). Specifically, the early stage oxidation of γ/γ’-based Ni-Al coatings with different Pt addition

(Chapter 3), the substrate composition effect on prolonged TBC lifetimes (Chapter 4) and a

new Pt diffusion bond coat design (Chapter 5) have been studied in this thesis. The main

conclusions are:

A. Pt addition has three effects on the early stage oxidation of γ/γ’-based Ni-Al coatings: 1)

retard θ-Al2O3 to α-Al2O3 transformation thus extending the transient θ-Al2O3 lifetime; 2)

promote the growth of Al2O3 and inhibit the growth of Ni-oxides; 3) significantly reduce

the TGO stress during the early stage of oxidation. Crystal orientation mapping results

show that the nucleation of α-Al2O3 is inhomogeneous along the oxide/coating interface

and might be related to the variation of coating compositions due to Pt additions. Spatially

resolved PLPS studies of the Ni-Pt-Al alloy with different Pt additions show that where Pt

stabilises the γ’ structure in nickel, the suppression of θ-Al2O3 to α-Al2O3 transition is

observed. Based on these findings, a new mechanism has been proposed to explain this Pt

effect on θ-Al2O3 to α-Al2O3 transformation: γ grains near the coating/oxide interface

promote the θ-Al2O3 to α-Al2O3 transformation while γ’ grains retard this transformation.


170

B. Cyclic oxidation tests at 1200 °C have been carried out on TBCs with the CMSX-4 and the

René N5 substrate respectively to study the effect of substrate composition on TBC

lifetimes. It was found that TBCs based on the CMSX-4 superalloy had a 25% longer

average cyclic lifetime compared to that on the René N5 superalloy. Meanwhile, the failure

occurred mainly along the bond coat/TGO interface for TBC with the René N5 substrate,

whereas for TBC with the CMSX-4 substrate, a mixed failure path (along the bond

coat/TGO interface and within TGO) was observed. This indicated that the different

cracking behaviour at the TGO/bond coat interface for the two TBCs may originate from

the difference in the intrinsic interface toughness evolution. To confirm this, a strain-to-fail

method combined with the 3D-DIC technique was employed to measure the bond

coat/TGO interface toughness of TBCs with different substrates. Although the mode I

interfacial toughness (Γic) values were almost identical for the two substrates in the as-

deposited state (~ 30 J/m2), the interfacial toughness of René N5 specimens decreased faster

with oxidation time. This faster degradation of TGO/bond coat interfacial toughness for the

N5 substrate can be ascribed to the sulfur segregation at this interface.

C. An Al-enriched γ’-Ni3Al-base bond coat was successfully fabricated on the CMSX-4

substrate by the selective γ-phase etching and a subsequent low temperature pack

cementation aluminizing process. The elemental diffusion profiles, TGO microstructure &

spallation, and rumpling of this Al-enriched γ’-phase coating have been investigated and

compared to the industry-standard β-NiPtAl coating and the Pt-diffused γ/γ’ coating.

Compared to the two conventional diffusion coatings, the Al-enriched γ’-phase coating can

resist high temperature rumpling deformation while maintaining adequate oxidation


171

properties and mitigating coating-substrate interdiffusion, which is promising for TBC

applications.

6.2 Future work

A. Study the early stage oxidation of NiCoCrAlY bond coats by scanning diffraction analysis

The findings in Chapter 3 indicated that the phase structure near the interface of the underlying

coating plays a key role in the transient Al2O3 to stable α-Al2O3 phase transformation rate

during early stage oxidation. This motivates further research by the scanning diffraction

technique to investigate the early stage oxidation behaviour of NiCoCrAlY bond coats with a

focus on the transient alumina transformation. Since the NiCoCrAlY bond coat is also two-

phase microstructure (β + γ), the alumina phase transformation mechanism relating to the

underlying coating phase structure might also be applied to this bond coat.

B. Study the effect of substrate composition on the lifetime of TBCs with β-NiPtAl bond coats

In Chapter 4, we have shown that substrate compositions can affect the lifetimes of TBCs with

the Pt-diffused γ/γ’ bond coats and illustrated the mechanisms behind this substrate effect.

However, the substrate composition effect on the TBCs with the β-NiPtAl bond coats is still

not fully understood. Further study is required to identify if the substrate composition plays a

role in the lifetime and failure mechanisms of this TBC system. Also, the interface toughness

measurement based on the strain-to-fail method combined with 3D-DIC technique can also be

utilised on this TBC system to replenish the current data base on this critical issue.

C. Prolonged cyclic oxidation tests on the Al-enriched γ’-phase coatings

Chapter 5 has demonstrated that the Al-enriched γ’-phase coating exhibited superior rumpling

resistance during short-term cyclic oxidation, indicating this new bond coat has high creep

strength. Thus, this new bond coat is expected to show promising cyclic oxidation performance


172

compared to the β-NiPtAl bond coats and Pt-diffused γ/γ’ bond coats because of its

combination of rumpling resistance and improved Al content. In order to validate this and

further explore the application of this new bond coat at elevated temperatures, prolonged cyclic

oxidation tests (1-h holding time, 1150 ℃) are required. The cyclic oxidation lifetime of this

new bond coat can be compared to that of the conventional Pt diffusion coatings reported in

the literature. In addition, some high temperature mechanical properties (e. g. yield strength)

also have significant influence on the durability of coatings during service. To obtain a better

understanding of its high temperature oxidation behaviour, the evolution of mechanical

properties of the Al-enriched γ’-phase bond coat at high temperatures needs to be

experimentally determined.

REFERENCE

173

Reference

[1] J.H. Perepezko, The Hotter the Engine, the Better, Science 326(5956) (2009) 1068-1069.

[2] D.R. Clarke, M. Oechsner, N.P. Padture, Thermal-barrier coatings for more efficient gas-turbine

engines, Mrs Bull 37(10) (2012) 891-902.

[3] D.R. Clarke, C.G. Levi, Materials design for the next generation thermal barrier coatings, Annu Rev

Mater Res 33 (2003) 383-417.

[4] A.G. Evans, D.R. Clarke, C.G. Levi, The influence of oxides on the performance of advanced gas

turbines, J Eur Ceram Soc 28(7) (2008) 1405-1419.

[5] Y. Chen, Study of bond coats for thermal barrier coating applications, The University of Manchester,

Manchester, 2015, pp. 1-161.

[6] D.R. Mumm, A.G. Evans, Mechanisms controlling the performance and durability of thermal barrier

coatings, Key Eng Mater 197 (2001) 199-230.

[7] L.T. Wu, R.T. Wu, P. Xiao, T. Osada, K.I. Lee, M. Bai, A prominent driving force for the spallation

of thermal barrier coatings: Chemistry dependent phase transformation of the bond coat, Acta Mater

137 (2017) 22-35.

[8] D.J. Jorgensen, Design and characterization of high-strength bond coats for improved thermal

barrier coating durability, University of California, Santa Barbara, 2016, pp. 1-302.

[9] S. Kramer, J. Yang, C.G. Levi, C.A. Johnson, Thermochemical interaction of thermal barrier

coatings with molten CaO-MgO-Al2O3-SiO2 (CMAS) deposits, J Am Ceram Soc 89(10) (2006) 3167-

3175.

[10] M.J. Stiger, N.M. Yanar, M.G. Topping, F.S. Pettit, G.H. Meier, Thermal barrier coatings for the

21st century, Z Metallkd 90(12) (1999) 1069-1078.

[11] N.P. Padture, M. Gell, E.H. Jordan, Materials science - Thermal barrier coatings for gas-turbine

engine applications, Science 296(5566) (2002) 280-284.

[12] A. Feuerstein, J. Knapp, T. Taylor, A. Ashary, A. Bolcavage, N. Hitchman, Technical and

economical aspects of current thermal barrier coating systems for gas turbine engines by thermal spray

and EBPVD: A review, J Therm Spray Techn 17(2) (2008) 199-213.

[13] P.G. Klemens, M. Gell, Thermal conductivity of thermal barrier coatings, Mat Sci Eng a-Struct

245(2) (1998) 143-149.

[14] J. Chevalier, L. Gremillard, A.V. Virkar, D.R. Clarke, The Tetragonal-Monoclinic Transformation

in Zirconia: Lessons Learned and Future Trends, J Am Ceram Soc 92(9) (2009) 1901-1920.

[15] S. Deville, G. Guenin, K. Chevalier, Martensitic transformation in zirconia - Part I. Nanometer

scale prediction and measurement of transformation induced relief, Acta Mater 52(19) (2004) 5697-

5707.

[16] M.S. Morsi, S.A. Abd El Gwad, M.A. Shoeib, K.F. Ahmed, Effect of Air Plasma Sprays

Parameters on Coating Performance in Zirconia-Based Thermal Barrier Coatings, Int J Electrochem Sc

7(4) (2012) 2811-2831.

[17] V. Lughi, V.K. Tolpygo, D.R. Clarke, Microstructural aspects of the sintering of thermal barrier

coatings, Mat Sci Eng a-Struct 368(1-2) (2004) 212-221.

[18] C.J. Chan, F.F. Lange, M. Ruhle, J.F. Jue, A.V. Virkar, Ferroelastic Domain Switching in

Tetragonal Zirconia Single-Crystals Microstructural Aspects, J Am Ceram Soc 74(4) (1991) 807-813.

[19] X.J. Jin, Martensitic transformation in zirconia containing ceramics and its applications, Curr Opin

Solid St M 9(6) (2005) 313-318.

[20] K.A. Khor, Y. Murakoshi, M. Takahashi, T. Sano, Plasma Spraying of Titanium Aluminide

Coatings - Process Parameters and Microstructure, J Mater Process Tech 48(1-4) (1995) 413-419.

REFERENCE

174

[21] U. Schulz, C. Leyens, K. Fritscher, M. Peters, B. Saruhan-Brings, O. Lavigne, J.M. Dorvaux, M.

Poulain, R. Mevrel, M.L. Caliez, Some recent trends in research and technology of advanced thermal

barrier coatings, Aerosp Sci Technol 7(1) (2003) 73-80.

[22] D.E. Wolfe, J. Singh, Synthesis and characterization of TiBCN coatings deposited by ion beam

assisted, co-evaporation electron beam-physical vapor deposition (EB-PVD), J Mater Sci 37(17) (2002)

3777-3787.

[23] R. Vassen, M.O. Jarligo, T. Steinke, D.E. Mack, D. Stover, Overview on advanced thermal barrier

coatings, Surf Coat Tech 205(4) (2010) 938-942.

[24] M. Bai, Fabrication and characterization of thermal barrier coatings, The University of Manchester,

2015, pp. 1-158.

[25] D.R. Clarke, S.R. Phillpot, Thermal barrier coating materials, Mater Today 8(6) (2005) 22-29.

[26] R. Vassen, E. Traeger, D. Stover, New thermal barrier coatings based on pyrochlore/YSZ double-

layer systems, Int J Appl Ceram Tec 1(4) (2004) 351-361.

[27] N.P. Bansal, D.M. Zhu, Effects of doping on thermal conductivity of pyrochlore oxides for

advanced thermal barrier coatings, Mat Sci Eng a-Struct 459(1-2) (2007) 192-195.

[28] Z.H. Xu, L.M. He, R.D. Mu, F. Lu, S.M. He, X.Q. Cao, Thermal cycling behavior of YSZ and

La2(Zr0.7Ce0.3)(2)O7 as double-ceramic-layer systems EB-PVD TBCs, J Alloy Compd 525 (2012) 87-96.

[29] K.A. Marino, B. Hinnemann, E.A. Carter, Atomic-scale insight and design principles for turbine

engine thermal barrier coatings from theory, P Natl Acad Sci USA 108(14) (2011) 5480-5487.

[30] X. Zhao, I.P. Shapiro, P. Xiao, Spinel formation in thermal barrier systems with a Pt-enriched

gamma-Ni+gamma'-Ni3Al bond coat, Surf Coat Tech 202(13) (2008) 2905-2916.

[31] T.M. Pollock, D.M. Lipkin, K.J. Hemker, Multifunctional coating interlayers for thermal-barrier

systems, Mrs Bull 37(10) (2012) 923-931.

[32] M.J. Pomeroy, Coatings for gas turbine materials and long term stability issues, Mater Design 26(3)

(2005) 223-231.

[33] Y. Matsuoka, Y. Matsunaga, K. Nakagawa, Y. Tuda, S. Taniguchi, Growth behavior of coatings

formed by vapor phase aluminizing using Fe-Al pellets of varying composition, Mater Trans 47(9)

(2006) 2341-2347.

[34] M. Zielinska, J. Sieniawski, M. Yavorska, M. Motyka, Influence of Chemical Composition of

Nickel Based Superalloy on the Formation of Aluminide Coatings, Arch Metall Mater 56(1) (2011)

193-197.

[35] B.M. Warnes, D.C. Punola, Clean diffusion coatings by chemical vapor deposition, Surf Coat Tech

94-5(1-3) (1997) 1-6.

[36] T. Lu, D.Z. Yao, C.G. Zhou, Low-temperature Formation of Aluminide Coatings on Ni-base

Superalloys by Pack Cementation Process, Chinese J Aeronaut 23(3) (2010) 381-385.

[37] Z.D. Xiang, P.K. Datta, Pack cementation process for the formation of refractory metal modified

aluminide coatings on nickel-base superalloys, J Mater Sci 38(18) (2003) 3721-3728.

[38] Z.D. Xiang, P.K. Datta, Relationship between pack chemistry and aluminide coating formation for

low-temperature aluminisation of alloy steels, Acta Mater 54(17) (2006) 4453-4463.

[39] A. Biswas, S.K. Roy, K.R. Gurumurthy, N. Prabhu, S. Banerjee, A study of self-propagating high-

temperature synthesis of NiAl in thermal explosion mode, Acta Mater 50(4) (2002) 757-773.

[40] Z.D. Xiang, S.R. Rose, P.K. Datta, Low-temperature formation and oxidation resistance of nickel

aluminide/nickel hybrid coatings on alloy steels, Scripta Mater 59(1) (2008) 99-102.

[41] G.W. Goward, D.H. Boone, Mechanisms of Formation of Diffusion Aluminide Coatings on

Nickel-Base Superalloys, Oxid Met 3(5) (1971) 475-&.

[42] J. Angenete, K. Stiller, Comparison of inward and outward grown Pt modified aluminide diffusion

coatings on a Ni based single crystal superalloy, Surf Coat Tech 150(2-3) (2002) 107-118.

[43] H.M. Tawancy, N.M. Abbas, T.N. Rhysjones, Role of Platinum in Aluminide Coatings, Surf Coat

Tech 49(1-3) (1991) 1-7.

REFERENCE

175

[44] P.Y. Hou, V.K. Tolpygo, Examination of the platinum effect on the oxidation behavior of nickel-

aluminide coatings, Surf Coat Tech 202(4-7) (2007) 623-627.

[45] B.A. Pint, I.G. Wright, W.Y. Lee, Y. Zhang, K. Prussner, K.B. Alexander, Substrate and bond coat

compositions: factors affecting alumina scale adhesion, Mat Sci Eng a-Struct 245(2) (1998) 201-211.

[46] J.A. Haynes, B.A. Pint, K.L. More, Y. Zhang, I.G. Wright, Influence of sulfur, platinum, and

hafnium on the oxidation behavior of CVD NiAl bond coatings, Oxid Met 58(5-6) (2002) 513-544.

[47] J. Angenete, K. Stiller, Oxidation of simple and Pt-modified aluminide diffusion coatings on Ni-

base superalloys - II. Oxide scale failure, Oxid Met 60(1-2) (2003) 83-101.

[48] H. Svensson, M. Christensen, P. Knutsson, G. Wahnstrom, K. Stiller, Influence of Pt on the metal-

oxide interface during high temperature oxidation of NiAl bulk materials, Corros Sci 51(3) (2009) 539-

546.

[49] P.Y. Hou, K.F. McCarty, Surface and interface segregation in beta-NiAl with and without Pt

addition, Scripta Mater 54(5) (2006) 937-941.

[50] S.R.B. D.S. Rickerby, R.G. Wing, Method of applying a thermal barrier coating to a superalloy

article and a thermal barrier coating, US Patent No. 5,667,663 (1997).

[51] J.A. Haynes, B.A. Pint, Y. Zhang, I.G. Wright, The effect of Pt content on gamma-gamma' NiPtAl

coatings, Surf Coat Tech 203(5-7) (2008) 413-416.

[52] T. Izumi, B. Gleeson, Oxidation behavior of Pt plus Hf-modified gamma-Ni plus gamma'-Ni(3)Al

alloys, High-Temperature Oxidation and Corrosion 2005 522-523 (2006) 221-228.

[53] S. Hayashi, T. Narita, B. Gleeson, Early-stage oxidation behavior of gamma'-Ni(3)Al-based alloys

with and without Pt addition, High-Temperature Oxidation and Corrosion 2005 522-523 (2006) 229-

238.

[54] S. Hayashi, D.J. Sordelet, L.R. Walker, B. Gleeson, Interdiffusion in Pt-containing gamma-Ni and

gamma'-Ni3Al alloys at 1150 degrees C, Mater Trans 49(7) (2008) 1550-1557.

[55] Y. Chen, X.F. Zhao, P. Xiao, Effect of microstructure on early oxidation of MCrAlY coatings,

Acta Mater 159 (2018) 150-162.

[56] Y. Zhang, B.A. Pint, J.A. Haynes, I.G. Wright, A platinum-enriched gamma+gamma' two-phase

bond coat on Ni-based superalloys, Surf Coat Tech 200(5-6) (2005) 1259-1263.

[57] E.A.G. Shillington, D.R. Clarke, Spalling failure of a thermal barrier coating associated with

aluminum depletion in the bond-coat, Acta Mater 47(4) (1999) 1297-1305.

[58] J.P. Stacy, Y. Zhang, B.A. Pint, J.A. Haynes, B.T. Hazel, B.A. Nagaraj, Synthesis and oxidation

performance of Al-enriched gamma+gamma' coatings on Ni-based superalloys via secondary

aluminizing, Surf Coat Tech 202(4-7) (2007) 632-636.

[59] Y. Zhang, D.A. Ballard, J.P. Stacy, B.A. Pint, J.A. Haynes, Synthesis and oxidation behavior of

platinum-enriched gamma+gamma' bond coatings on Ni-based superalloys, Surf Coat Tech 201(7)

(2006) 3857-3861.

[60] O. Trunova, T. Beck, R. Herzog, R.W. Steinbrech, L. Singheiser, Damage mechanisms and lifetime

behavior of plasma sprayed thermal barrier coating systems for gas turbines - Part 1: Experiments, Surf

Coat Tech 202(20) (2008) 5027-5032.

[61] J.S. Jiang, Z.H. Zou, W.Z. Wang, X.F. Zhao, Y.Z. Liu, Z.M. Cao, Effect of internal oxidation on

the interfacial morphology and residual stress in air plasma sprayed thermal barrier coatings, Surf Coat

Tech 334 (2018) 215-226.

[62] F.H. Yuan, Z.X. Chen, Z.W. Huang, Z.G. Wang, S.J. Zhu, Oxidation behavior of thermal barrier

coatings with HVOF and detonation-sprayed NiCrAlY bondcoats, Corros Sci 50(6) (2008) 1608-1617.

[63] F. Tang, L. Ajdelsztajn, J.M. Schoenung, Characterization of oxide scales formed on HVOF

NiCrAlY coatings with various oxygen contents introduced during thermal spraying, Scripta Mater 51(1)

(2004) 25-29.

[64] M. Goral, S. Kotowski, K. Dychton, M. Drajewicz, T. Kubaszek, Influence of low pressure plasma

spraying parameters on MCrAlY bond coat and its microstructure, Materials Structure &

Micromechanics of Fracture Vii 592-593 (2014) 421-424.

REFERENCE

176

[65] N.M. Yanar, F.S. Pettit, G.H. Meier, Failure characteristics during cyclic oxidation of yttria

stabilized zirconia thermal barrier coatings deposited via electron beam physical vapor deposition on

platinum aluminide and on NiCoCrAlY bond coats with processing modifications for improved

performances, Metall Mater Trans A 37a(5) (2006) 1563-1580.

[66] E. Hejrani, D. Sebold, W.J. Nowak, G. Mauer, D. Naumenko, R. Vassen, W.J. Quadakkers,

Isothermal and cyclic oxidation behavior of free standing MCrAlY coatings manufactured by high-

velocity atmospheric plasma spraying, Surf Coat Tech 313 (2017) 191-201.

[67] M.H. Guo, Q.M. Wang, P.L. Ke, J. Gong, C. Sun, R.F. Huang, L.S. Wen, The preparation and hot

corrosion resistance of gradient NiCoCrAlYSiB coatings, Surf Coat Tech 200(12-13) (2006) 3942-3949.

[68] B. Gleeson, N. Mu, S. Hayashi, Compositional factors affecting the establishment and maintenance

of Al2O3 scales on Ni-Al-Pt systems, J Mater Sci 44(7) (2009) 1704-1710.

[69] B.A. Pint, K.L. More, Characterization of alumina interfaces in TBC systems, J Mater Sci 44(7)

(2009) 1676-1686.

[70] B.A. Pint, Experimental observations in support of the dynamic-segregation theory to explain the

reactive-element effect, Oxid Met 45(1-2) (1996) 1-37.

[71] Y. Chen, X. Zhao, Y. Dang, P. Xiao, N. Curry, N. Markocsan, P. Nylen, Characterization and

understanding of residual stresses in a NiCoCrAlY bond coat for thermal barrier coating application,

Acta Mater 94 (2015) 1-14.

[72] F.H. Stott, G.C. Wood, Growth and Adhesion of Oxide Scales on Al2O3-Forming Alloys and

Coatings, Mater Sci Eng 87(1-2) (1987) 267-274.

[73] X. Zhao, P. Xiao, Thermal barrier coatings on nickel superalloy substrates, Advances in Ceramic

Materials 606 (2009) 1-26.

[74] A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, F.S. Pettit, Mechanisms c ontrolling the

durability of thermal barrier coatings, Prog Mater Sci 46(5) (2001) 505-553.

[75] M. Gell, E. Jordan, K. Vaidyanathan, K. McCarron, B. Barber, Y.H. Sohn, V.K. Tolpygo, Bond

strength, bond stress and spallation mechanisms of thermal barrier coatings, Surf Coat Tech 120 (1999)

53-60.

[76] V.K. Tolpygo, D.R. Clarke, Competition between stress generation and relaxation during oxidation

of an Fe-Cr-Al-Y alloy, Oxid Met 49(1-2) (1998) 187-212.

[77] V.K. Tolpygo, J.R. Dryden, D.R. Clarke, Determination of the growth stress and strain in alpha-

Al2O3 scales during the oxidation of Fe-22Cr-4.8Al-0.3Y alloy, Acta Mater 46(3) (1998) 927-937.

[78] E. Schumann, C. Sarioglu, J.R. Blachere, F.S. Pettit, G.H. Meier, High-temperature stress

measurements during the oxidation of NiAl, Oxid Met 53(3-4) (2000) 259-272.

[79] P.Y. Hou, A.P. Paulikas, B.W. Veal, Strains in thermally growing alumina films measured in-situ

using synchrotron x-rays, High-Temperature Oxidation and Corrosion 2005 522-523 (2006) 433-440.

[80] J.A. Nychka, D.R. Clarke, Damage quantification in TBCs by photo-stimulated luminescence

spectroscopy, Surf Coat Tech 146 (2001) 110-116.

[81] S. Sridharan, L.D. Xie, E.H. Jordan, M. Gell, Stress variation with thermal cycling in the thermally

grown oxide of an EB-PVD thermal barrier coating, Surf Coat Tech 179(2-3) (2004) 286-296.

[82] R.J. Christensen, D.M. Lipkin, D.R. Clarke, K. Murphy, Nondestructive evaluation of the oxidation

stresses through thermal barrier coatings using Cr3+ piezospectroscopy, Appl Phys Lett 69(24) (1996)

3754-3756.

[83] D.M. Lipkin, D.R. Clarke, Measurement of the stress in oxide scales formed by oxidation of

alumina-forming alloys, Oxid Met 45(3-4) (1996) 267-280.

[84] J. He, D.R. Clarke, Determination of the Piezospectroscopic Coefficients for Chromium-Doped

Sapphire, J Am Ceram Soc 78(5) (1995) 1347-1353.

[85] D.R. Clarke, R.J. Christensen, V. Tolpygo, The evolution of oxidation stresses in zirconia thermal

barrier coated superalloy leading to spalling failure, Surf Coat Tech 94-5(1-3) (1997) 89-93.

[86] X. Wang, G. Lee, A. Atkinson, Investigation of TBCs on turbine blades by photoluminescence

piezospectroscopy, Acta Mater 57(1) (2009) 182-195.

REFERENCE

177

[87] P.T. Moseley, K.R. Hyde, B.A. Bellamy, G. Tappin, The Microstructure of the Scale Formed

during the High-Temperature Oxidation of a Fecralloy Steel, Corros Sci 24(6) (1984) 547-565.

[88] K.M.N. Prasanna, A.S. Khanna, R. Chandra, W.J. Quadakkers, Effect of theta-alumina formation

on the growth kinetics of alumina-forming superalloys, Oxid Met 46(5-6) (1996) 465-480.

[89] M.W. Brumm, H.J. Grabke, The Oxidation Behavior of NiAl 1. Phase-Transformations in the

Alumina Scale during Oxidation of Nial and Nial-Cr Alloys, Corros Sci 33(11) (1992) 1677-&.

[90] J. Doychak, M. Ruhle, Tem Studies of Oxidized Nial and Ni3al Cross-Sections, Oxid Met 31(5-6)

(1989) 431-452.

[91] J. Doychak, J.L. Smialek, T.E. Mitchell, Transient Oxidation of Single-Crystal Beta-Nial, Metall

Trans A 20(3) (1989) 499-518.

[92] J.C. Yang, E. Schumann, I. Levin, M. Ruhle, Transient oxidation of NiAl, Acta Mater 46(6) (1998)

2195-2201.

[93] S. Hayashi, B. Gleeson, Phase Transformation Behavior of Al2O3 Scale Formed on Pt-Modified

gamma '-Ni3Al-Based Alloys With and Without Hf Addition, Oxid Met 77(5-6) (2012) 237-251.

[94] J. Jedlinski, J.L.G. Poussard, J. Dabek, B. Gleeson, M. Nocun, R. Goldyn, The Effect of Pt Content

on the Scale Development on beta-NiAl at Very Early Oxidation Stages, Oxid Met 87(3-4) (2017) 311-

319.

[95] G.C. Rybicki, J.L. Smialek, Effect of the Theta-Alpha-Al2O3 Transformation on the Oxidation

Behavior of Beta-NiAl+Zr, Oxid Met 31(3-4) (1989) 275-304.

[96] S. Hayashi, B. Gleeson, Early-Stage Oxidation Behavior of Pt-Modified gamma'-Ni3Al-Based

Alloys with and without Hf Addition, Oxid Met 71(1-2) (2009) 5-19.

[97] J. Jedlinski, The Effect of Yttrium on the Early Stages of Oxidation of Alumina Formers -

Comments, Oxid Met 39(1-2) (1993) 55-60.

[98] Y. Cadoret, D. Monceau, M. Bacos, P. Josso, V. Maurice, P. Marcus, Effect of platinum on the

growth rate of the oxide scale formed on cast nickel aluminide intermetallic alloys, Oxid Met 64(3-4)

(2005) 185-205.

[99] G. Smola, W. Wang, J. Jedlinski, B. Gleeson, K. Kowalski, A. Bernasik, M. Nocun, Mechanistic

aspects of Pt-modified beta-NiAl alloy oxidation, Mater High Temp 26(3) (2009) 273-280.

[100] T.M. Pollock, S. Tin, Nickel-based superalloys for advanced turbine engines: Chemistry,

microstructure, and properties, J Propul Power 22(2) (2006) 361-374.

[101] C.T. Sims, N.S. Stoloff, W.C. Hagel, Superalloys II, Wiley, New York, 1987.

[102] P. Caron, T. Khan, Evolution of Ni-based superalloys for single crystal gas turbine blade

applications, Aerosp Sci Technol 3(8) (1999) 513-523.

[103] R.C. Reed, The superalloys : fundamentals and applications, Cambridge University Press,

Cambridge, UK ; New York, 2008.

[104] M.S.A. Karunaratne, S. Kyaw, A. Jones, R. Morrell, R.C. Thomson, Modelling the coefficient of

thermal expansion in Ni-based superalloys and bond coatings, J Mater Sci 51(9) (2016) 4213-4226.

[105] E.S. Huron, Serrated Yielding in a Nickel-Base Superalloy, Superalloys 1992 (1992) 675-684.

[106] A.C. Yeh, A. Sato, T. Kobayashi, H. Harada, On the creep and phase stability of advanced Ni-

base single crystal superalloys, Mat Sci Eng a-Struct 490(1-2) (2008) 445-451.

[107] X. Chen, R. Wang, N. Yao, A.G. Evans, J.W. Hutchinson, R.W. Bruce, Foreign object damage

in a thermal barrier system: mechanisms and simulations, Mat Sci Eng a-Struct 352(1-2) (2003) 221-

231.

[108] M. Gell, K. Vaidyanathan, B. Barber, J. Cheng, E. Jordan, Mechanism of spallation in platinum

aluminide/electron beam physical vapor-deposited thermal barrier coatings, Metall Mater Trans A 30(2)

(1999) 427-435.

[109] A.G. Evans, J.W. Hutchinson, M.Y. He, Micromechanics model for the detachment of residually

compressed brittle films and coatings, Acta Mater 47(5) (1999) 1513-1522.

[110] J.W. Hutchinson, M.Y. He, A.G. Evans, The influence of imperfections on the nucleation and

propagation of buckling driven delaminations, J Mech Phys Solids 48(4) (2000) 709-734.

REFERENCE

178

[111] I.T. Spitsberg, D.R. Mumm, A.G. Evans, On the failure mechanisms of thermal barrier coatings

with diffusion aluminide bond coatings, Mat Sci Eng a-Struct 394(1-2) (2005) 176-191.

[112] J.S. Wang, A.G. Evans, Measurement and analysis of buckling and buckle propagation in

compressed oxide layers on superalloy substrates, Acta Mater 46(14) (1998) 4993-5005.

[113] P.K. Wright, A.G. Evans, Mechanisms governing the performance of thermal barrier coatings,

Curr Opin Solid St M 4(3) (1999) 255-265.

[114] X.Y. Gong, D.R. Clarke, On the measurement of strain in coatings formed on a wrinkled elastic

substrate, Oxid Met 50(5-6) (1998) 355-376.

[115] K.W. Schlichting, N.P. Padture, E.H. Jordan, M. Gell, Failure modes in plasma-sprayed thermal

barrier coatings, Mat Sci Eng a-Struct 342(1-2) (2003) 120-130.

[116] D. Renusch, M. Schorr, M. Schutze, The role that bond coat depletion of aluminum has on the

lifetime of APS-TBC under oxidizing conditions, Mater Corros 59(7) (2008) 547-555.

[117] V.K. Tolpygo, D.R. Clarke, K.S. Murphy, Evaluation of interface degradation during cyclic

oxidation of EB-PVD thermal barrier coatings and correlation with TGO luminescence, Surf Coat Tech

188 (2004) 62-70.

[118] Y. Chen, X. Zhao, M. Bai, L. Yang, C. Li, L. Wang, J.A. Carr, P. Xiao, A mechanistic

understanding on rumpling of a NiCoCrA1Y bond coat for thermal barrier coating applications, Acta

Mater 128 (2017) 31-42.

[119] T. Xu, S. Faulhaber, C. Mercer, M. Maloney, A. Evans, Observations and analyses of failure

mechanisms in thermal barrier systems with two phase bond coats based on NiCoCrAlY, Acta Mater

52(6) (2004) 1439-1450.

[120] D.R. Mumm, A.G. Evans, I.T. Spitsberg, Characterization of a cyclic displacement instability for

a thermally grown oxide in a thermal barrier system, Acta Mater 49(12) (2001) 2329-2340.

[121] V.K. Tolpygo, D.R. Clarke, On the rumpling mechanism in nickel-aluminide coatings - Part I: an

experimental assessment, Acta Mater 52(17) (2004) 5115-5127.

[122] V.K. Tolpygo, D.R. Clarke, Rumpling of CVD (Ni,Pt)Al diffusion coatings under intermediate

temperature cycling, Surf Coat Tech 203(20-21) (2009) 3278-3285.

[123] V.K. Tolpygo, D.R. Clarke, Rumpling induced by thermal cycling of an overlay coating: the

effect of coating thickness, Acta Mater 52(3) (2004) 615-621.

[124] Y.H. Sohn, J.H. Kim, E.H. Jordan, M. Gell, Thermal cycling of EB-PVD/MCrAlY thermal barrier

coatings: 1. Microstructural development and spallation mechanisms, Surf Coat Tech 146 (2001) 70-

78.

[125] D.S. Balint, J.W. Hutchinson, An analytical model of rumpling in thermal barrier coatings, J

Mech Phys Solids 53(4) (2005) 949-973.

[126] H.E. Evans, Oxidation failure of TBC systems: An assessment of mechanisms, Surf Coat Tech

206(7) (2011) 1512-1521.

[127] D.R.C. V.K. Tolpygo, unpublished work, (2003).

[128] R.J. Christensen, V.K. Tolpygo, D.R. Clarke, The influence of the reactive element yttrium on

the stress in alumina scales formed by oxidation, Acta Mater 45(4) (1997) 1761-1766.

[129] L.R. Luo, X. Shan, Z.H. Zou, C.S. Zhao, X. Wang, A.P. Zhang, X.F. Zhao, F.W. Guo, P. Xiao,

A high performance NiCoCrAlY bond coat manufactured using laser powder deposition, Corros Sci

126 (2017) 356-365.

[130] A. Selcuk, A. Atkinson, The evolution of residual stress in the thermally grown oxide on Pt

diffusion bond coats in TBCs, Acta Mater 51(2) (2003) 535-549.

[131] X. Zhao, P. Xiao, Effect of platinum on the durability of thermal barrier systems with a gamma

plus gamma' bond coat, Thin Solid Films 517(2) (2008) 828-834.

[132] X. Wang, C.J. Wang, A. Atkinson, Interface fracture toughness in thermal barrier coatings by

cross-sectional indentation, Acta Mater 60(17) (2012) 6152-6163.

[133] J.J. Chen, S.J. Bull, Approaches to investigate delamination and interfacial toughness in coated

systems: an overview, J Phys D Appl Phys 44(3) (2011).

REFERENCE

179

[134] A.A. Volinsky, N.R. Moody, W.W. Gerberich, Interfacial toughness measurements for thin films

on substrates, Acta Mater 50(3) (2002) 441-466.

[135] A. Furuya, N. Hosoi, Y. Ohshita, Evaluation of Cu Adhesive Energy on Barrier Metals by Means

of Contactangle Measurement, J Appl Phys 78(10) (1995) 5989-5992.

[136] P.D. Warren, D.A. Hills, S.G. Roberts, Surface Flaw Distributions in Brittle Materials and

Hertzian Fracture, J Mater Res 9(12) (1994) 3194-3202.

[137] M.D. Thouless, Cracking and Delamination of Coatings, J Vac Sci Technol A 9(4) (1991) 2510-

2515.

[138] X.D. Li, D.F. Diao, B. Bhushan, Fracture mechanisms of thin amorphous carbon films in

nanoindentation, Acta Mater 45(11) (1997) 4453-4461.

[139] J.W. Hutchinson, Z. Suo, Mixed-Mode Cracking in Layered Materials, Adv Appl Mech 29 (1992)

63-191.

[140] J. Malzbender, G. de With, Energy dissipation, fracture toughness and the indentation load-

displacement curve of coated materials, Surf Coat Tech 135(1) (2000) 60-68.

[141] M.D. Drory, J.W. Hutchinson, Measurement of the adhesion of a brittle film on a ductile substrate

by indentation, P Roy Soc a-Math Phy 452(1953) (1996) 2319-2341.

[142] J.J. Chen, Indentation-based methods to assess fracture toughness for thin coatings, J Phys D Appl

Phys 45(20) (2012).

[143] G. Fargas, D. Casellas, L. Llanes, M. Anglada, Thermal shock resistance of yttria-stabilized

zirconia with Palmqvist indentation cracks, J Eur Ceram Soc 23(1) (2003) 107-114.

[144] K. Niihara, R. Morena, D.P.H. Hasselman, Evaluation of KIc of Brittle Solids by the Indentation

Method with Low Crack-to-Indent Ratios, J Mater Sci Lett 1(1) (1982) 13-16.

[145] J. Chen, S.J. Bull, Assessment of the toughness of thin coatings using nanoindentation under

displacement control, Thin Solid Films 494(1-2) (2006) 1-7.

[146] A. Vasinonta, J.L. Beuth, Measurement of interfacial toughness in thermal barrier coating systems

by indentation, Eng Fract Mech 68(7) (2001) 843-860.

[147] S.S. Kim, Y.F. Liu, Y. Kagawa, Evaluation of interfacial mechanical properties under shear

loading in EB-PVD TBCs by the pushout method, Acta Mater 55(11) (2007) 3771-3781.

[148] D.R. Mumm, K.T. Faber, Interfacial Debonding and Sliding in Brittle-Matrix Composites

Measured Using an Improved Fiber Pullout Technique, Acta Metall Mater 43(3) (1995) 1259-1270.

[149] S.Q. Guo, D.R. Mumm, A.M. Karlsson, Y. Kagawa, Measurement of interfacial shear mechanical

properties in thermal barrier coating systems by a barb pullout method, Scripta Mater 53(9) (2005)

1043-1048.

[150] Y.F. Liu, Y. Kagawa, A.G. Evans, Analysis of a "barb test" for measuring the mixed-mode

delamination toughness of coatings, Acta Mater 56(1) (2008) 43-49.

[151] D. Di Maio, S.G. Roberts, Measuring fracture toughness of coatings using focused-ion-beam-

machined microbeams, J Mater Res 20(2) (2005) 299-302.

[152] K. Matoy, T. Detzel, M. Muller, C. Motz, G. Dehm, Interface fracture properties of thin films

studied by using the micro-cantilever deflection technique, Surf Coat Tech 204(6-7) (2009) 878-881.

[153] T. Kitamura, T. Shibutani, T. Ueno, Crack initiation at free edge of interface between thin films

in advanced LSI, Eng Fract Mech 69(12) (2002) 1289-1299.

[154] Y. Chen, X. Zhang, X.F. Zhao, N. Markocsan, P. Nylen, P. Xiao, Measurements of elastic

modulus and fracture toughness of an air plasma sprayed thermal barrier coating using micro-cantilever

bending, Surf Coat Tech 374 (2019) 12-20.

[155] G. Katz, Adhesion of Copper-Films to Aluminum-Oxide Using a Spinel Structure Interface, Thin

Solid Films 33(1) (1976) 99-105.

[156] I. Hofinger, M. Oechsner, H.A. Bahr, M.V. Swain, Modified four-point bending specimen for

determining the interface fracture energy for thin, brittle layers, Int J Fracture 92(3) (1998) 213-220.

REFERENCE

180

[157] Y. Zhao, A. Shinmi, X. Zhao, P.J. Withers, S. Van Boxel, N. Markocsan, P. Nylen, P. Xiao,

Investigation of interfacial properties of atmospheric plasma sprayed thermal barrier coatings with four-

point bending and computed tomography technique, Surf Coat Tech 206(23) (2012) 4922-4929.

[158] K.L. Mittal, Adhesion Measurement of Thin-Films, Electrocomp Sci Tech 3(1) (1976) 21-42.

[159] Y. Chen, X. Zhao, M. Bai, A. Chandio, R. Wu, P. Xiao, Effect of platinum addition on oxidation

behaviour of gamma/gamma ' nickel aluminide, Acta Mater 86 (2015) 319-330.

[160] Y. Niu, W.T. Wu, D.H. Boone, J.S. Smith, J.Q. Zhang, C.L. Zhen, Oxidation Behavior of Simple

and Pt-Modified Aluminide Coatings on In738 at 1100-Degrees-C, J Phys Iv 3(C9) (1993) 511-519.

[161] H.J. Choi, J. Jedlinski, B. Yao, Y.H. Sohn, Transmission electron microscopy observations on

the phase composition and microstructure of the oxidation scale grown on as-polished and yttrium-

implanted beta-NiAl, Surf Coat Tech 205(5) (2010) 1206-1210.

[162] V.K. Tolpygo, D.R. Clarke, Microstructural study of the theta-alpha transformation in alumina

scales formed on nickel-aluminides, Mater High Temp 17(1) (2000) 59-70.

[163] H. Svensson, P. Knutsson, K. Stiller, Formation and Healing of Voids at the Metal-Oxide

Interface in NiAl Alloys, Oxid Met 71(3-4) (2009) 143-156.

[164] A. Garner, A. Gholinia, P. Frankel, M. Gass, I. MacLaren, M. Preuss, The microstructure and

microtexture of zirconium oxide films studied by transmission electron backscatter diffraction and

automated crystal orientation mapping with transmission electron microscopy, Acta Mater 80 (2014)

159-171.

[165] P.W. Trimby, Orientation mapping of nanostructured materials using transmission Kikuchi

diffraction in the scanning electron microscope, Ultramicroscopy 120 (2012) 16-24.

[166] X. Peng, D.R. Clarke, Piezospectroscopic analysis of interface debonding in thermal barrier

coatings, J Am Ceram Soc 83(5) (2000) 1165-1170.

[167] R.J. Bennett, R. Krakow, A.S. Eggeman, C.N. Jones, H. Murakami, C.M.F. Rae, On the oxidation

behavior of titanium within coated nickel-based superalloys, Acta Mater 92 (2015) 278-289.

[168] J. Mayer, L.A. Giannuzzi, T. Kamino, J. Michael, TEM sample preparation and FIB-induced

damage, Mrs Bull 32(5) (2007) 400-407.

[169] E. Husson, Y. Repelin, Structural studies of transition aluminas. Theta alumina, Eur J Sol State

Inor 33(11) (1996) 1223-1231.

[170] E.A. Owen, E.L. Yates, X ray measurement of the thermal expansion of pure nickel, Philos Mag

21(142) (1936) 809-819.

[171] L.W. Finger, R.M. Hazen, Crystal-Structure and Compression of Ruby to 46 Kbar, J Appl Phys

49(12) (1978) 5823-5826.

[172] S. Zaefferer, New developments of computer-aided crystallographic analysis in transmission

electron microscopy, J Appl Crystallogr 33 (2000) 10-25.

[173] E.F. Rauch, L. Dupuy, Rapid spot diffraction patterns identification through template matching,

Arch Metall Mater 50(1) (2005) 87-99.

[174] C.G. Levi, E. Sommer, S.G. Terry, A. Catanoiu, M. Ruhle, Alumina grown during depos ition of

thermal barrier coatings on NiCrAlY, J Am Ceram Soc 86(4) (2003) 676-685.

[175] B. Gleeson, W. Wang, S. Hayashi, D. Sordelet, Effects of platinum on the interdiffusion and

oxidation behavior of Ni-Al-based alloys, Mater Sci Forum 461-464 (2004) 213-222.

[176] H.C. Kao, W.C.J. Wei, Kinetics and microstructural evolution of heterogeneous transformation

of theta-alumina to alpha-alumina (vol 83, pg 362, 2000), J Am Ceram Soc 83(5) (2000) 1325-1325.

[177] B.A. Pint, M. Treska, L.W. Hobbs, The effect of various oxide dispersions on the phase

composition and morphology of Al2O3 scales grown on beta-NiAl, Oxid Met 47(1-2) (1997) 1-20.

[178] J. Jedlinski, Defect-diffusion-stress relationships in modeling the oxidation and degradation

processes of alumina formers: A brief survey, Defect Diffus Forum 237-240 (2005) 911-921.

[179] D.D. Lee, H.S. Seung, Algorithms for non-negative matrix factorization, Adv Neur In 13 (2001)

556-562.

REFERENCE

181

[180] H.M. Tawancy, L.M. Al-Hadhrami, Influence of Titanium in Nickel-Base Superalloys on the

Performance of Thermal Barrier Coatings Utilizing gamma-gamma ' Platinum Bond Coats, J Eng Gas

Turb Power 133(4) (2011).

[181] U. Schulz, M. Menzebach, C. Leyens, Y.Q. Yang, Influence of substrate material on oxidation

behavior and cyclic lifetime of EB-PVD TBC systems, Surf Coat Tech 146 (2001) 117-123.

[182] K. Bouhanek, O.A. Adesanya, F.H. Stott, P. Skeldon, D.G. Lees, G.C. Wood, High temperature

oxidation of thermal barrier coating systems on RR3000 substrates: Pt aluminide bond coats, High

Temperature Corrosion and Protection of Materials 5, Pts 1 and 2 369-3 (2001) 615-622.

[183] M. Levy, P. Farrell, F. Pettit, Oxidation of Some Advanced Single-Crystal Nickel-Base

Superalloys in Air at 2000-F (1093-C), Corrosion 42(12) (1986) 708-717.

[184] R. Molins, I. Rouzou, P. Hou, Chemical and morphological evolution of a (NiPt)Al bondcoat,

Oxid Met 65(3-4) (2006) 263-283.

[185] R.T. Wu, K. Kawagishi, H. Harada, R.C. Reed, The retention of thermal barrier coating systems

on single-crystal superalloys: Effects of substrate composition, Acta Mater 56(14) (2008) 3622-3629.

[186] B.A. Pint, J.A. Haynes, Y. Zhang, Effect of superalloy substrate and bond coating on TBC lifetime,

Surf Coat Tech 205(5) (2010) 1236-1240.

[187] D.R. Mumm, A.G. Evans, On the role of imperfections in the failure of a thermal barrier coating

made by electron beam deposition, Acta Mater 48(8) (2000) 1815-1827.

[188] Y. Yamazaki, A. Schmidt, A. Scholz, The determination of the delamination resistance in thennal

barrier coating system by four-point bending tests, Surf Coat Tech 201(3-4) (2006) 744-754.

[189] D.B. Marshall, A.G. Evans, Measurement of Adherence of Residually Stressed Thin-Films by

Indentation .1. Mechanics of Interface Delamination, J Appl Phys 56(10) (1984) 2632-2638.

[190] Y.C. Zhou, T. Hashida, C.Y. Jian, Determination of interface fracture toughness in thermal barrier

coating system by blister tests, J Eng Mater-T Asme 125(2) (2003) 176-182.

[191] X. Zhao, J. Liu, D.S. Rickerby, R.J. Jones, P. Xiao, Evolution of interfacial toughness of a thermal

barrier system with a Pt-diffused gamma/gamma' bond coat, Acta Mater 59(16) (2011) 6401-6411.

[192] J.W. Hutchinson, M.D. Thouless, E.G. Liniger, Growth and Configurational Stability of Circular,

Buckling-Driven Film Delaminations, Acta Metall Mater 40(2) (1992) 295-308.

[193] X. Wang, A. Atkinson, Piezo-spectroscopic mapping of the thermally grown oxide in thermal

barrier coatings, Mat Sci Eng a-Struct 465(1-2) (2007) 49-58.

[194] W.C. Oliver, G.M. Pharr, An Improved Technique for Determining Hardness and Elastic-

Modulus Using Load and Displacement Sensing Indentation Experiments, J Mater Res 7(6) (1992)

1564-1583.

[195] S. Roux, J. Rethore, F. Hild, Digital image correlation and fracture: an advanced technique for

estimating stress intensity factors of 2D and 3D cracks, J Phys D Appl Phys 42(21) (2009).

[196] O.A. Adesanya, K. Bouhanek, F.H. Stott, P. Skeldon, D.G. Lees, G.C. Wood, Cyclic oxidation

of two bond coats in thermal barrier coating systems on CMSX-4 substrates, High Temperature

Corrosion and Protection of Materials 5, Pts 1 and 2 369-3 (2001) 639-646.

[197] C. Jiang, B. Gleeson, Surface segregation of Pt in gamma'-Ni3Al: A first-principles study, Acta

Mater 55(5) (2007) 1641-1647.

[198] X. Wang, A. Atkinson, L. Chirivi, J.R. Nicholls, Evolution of stress and morphology in thermal

barrier coatings, Surf Coat Tech 204(23) (2010) 3851-3857.

[199] T. Gheno, D. Monceau, D. Oquab, Y. Cadoret, Characterization of Sulfur Distribution in Ni-

Based Superalloy and Thermal Barrier Coatings After High Temperature Oxidation: A SIMS Analysis,

Oxid Met 73(1-2) (2010) 95-113.

[200] P.Y. Hou, Impurity segregation to scale/alloy interfaces and its effect on interfacial properties,

High Temperature Corrosion and Protection of Materials 5, Pts 1 and 2 369-3 (2001) 23-38.

[201] K. Shirvani, S. Firouzi, A. Rashidghamat, Microstructures and cyclic oxidation behaviour of Pt-

free and low-Pt NiAl coatings on the Ni-base superalloy Rene-80, Corros Sci 55 (2012) 378-384.

REFERENCE

182

[202] S. Hayashi, S.I. Ford, D.J. Young, D.J. Sordelet, M.F. Besser, B. Gleeson, alpha-NiPt(Al) and

phase equilibria in the Ni-Al-Pt system at 1150 degrees C, Acta Mater 53(11) (2005) 3319-3328.

[203] D.J. Jorgensen, A. Suzuki, D.M. Lipkin, T.M. Pollock, Bond coatings with high rumpling

resistance: Design and characterization, Surf Coat Tech 300 (2016) 25-34.

[204] D. Pan, M.W. Chen, P.K. Wright, K.J. Hemker, Evolution of a diffusion aluminide bond coat for

thermal barrier coatings during thermal cycling, Acta Mater 51(8) (2003) 2205-2217.

[205] G.J. Tatlock, T.J. Hurd, Platinum and the Oxidation Behavior of a Nickel Based Superalloy, Oxid

Met 22(5-6) (1984) 201-226.

[206] R.T. Wu, X. Wang, A. Atkinson, On the interfacial degradation mechanisms of thermal barrier

coating systems: Effects of bond coat composition, Acta Mater 58(17) (2010) 5578-5585.

[207] K.J. Hemker, W.D. Nix, High-Temperature Creep of the Intermetallic Alloy Ni3Al, Metall Trans

A 24(2) (1993) 335-341.

[208] R. Darolia, W.S. Walston, M.V. Nathal, NiAl alloys for turbine airfoils, Superalloys 1996 (1996)

561-570.

[209] J.A. Haynes, B.A. Pint, Y. Zhang, I.G. Wright, Comparison of the cyclic oxidation behavior of

beta-NiAl, beta-NiPtAl and gamma-gamma' NiPtAl coatings on various superalloys, Surf Coat Tech

202(4-7) (2007) 730-734.

[210] E. Schumann, M. Ruhle, Microstructural Observations on the Oxidation of Gamma-Ni3Al at High

Oxygen Partial-Pressure, Acta Metall Mater 42(4) (1994) 1481-1487.

[211] J.H. Chen, J.A. Little, Degradation of the platinum aluminide coating on CMSX4 at 1100 degrees

C, Surf Coat Tech 92(1-2) (1997) 69-77.

[212] Y. Zhang, J.A. Haynes, W.Y. Lee, I.G. Wright, B.A. Pint, K.M. Cooley, P.K. Liaw, Effects of Pt

incorporation on the isothermal oxidation behavior of chemical vapor deposition aluminide coatings,

Metall Mater Trans A 32(7) (2001) 1727-1741.

[213] V. Deodeshmukh, N. Mu, B. Li, B. Gleeson, Hot corrosion and oxidation behavior of a novel Pt

plus Hf-modified gamma'-Ni3Al + gamma-Ni-based coating, Surf Coat Tech 201(7) (2006) 3836-3840.

[214] A.V. Put, D. Oquab, J.R. Nicholls, D. Monceau, Hf Addition by Sputtering in beta-NiPtAl Bond

Coating for TBC Systems and its Effect on Thermal Cycling Behaviour, Tms 2009 138th Annual

Meeting & Exhibition - Supplemental Proceedings, Vol 1: Materials Processing and Properties (2009)

185-194.

[215] V.D. Divya, U. Ramamurty, A. Paul, Effect of Pt on interdiffusion and mechanical properties of

the gamma and gamma' phases in the Ni-Pt-Al system, Philos Mag 92(17) (2012) 2187-2214.

[216] S. Hayashi, W. Wang, D.J. Sordelet, B. Gleeson, Interdiffusion behavior of Pt-modified gamma-

Ni+gamma'-Ni3Al alloys coupled to Ni-Al-based alloys, Metall Mater Trans A 36a(7) (2005) 1769-

1775.

[217] X. Zhao, B. Cernik, C.C. Tang, S.P. Thompson, P. Xiao, Stress evolution in a Pt-diffused

gamma/gamma' bond coat after oxidation, Surf Coat Tech 247 (2014) 48-54.

study of bond coats and failure mechanisms of thermal

Documents