Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
AN EXPERIMENTAL AND COMPUTATIONAL
STUDY OF DAMAGE AND CRACK GROWTH FOR A
NICKEL-BASED SUPERALLOY UNDER FATIGUE-
OXIDATION CONDITIONS
ALKISTIS KARABELA
BENG (HONS), MSC, MIET, AMIMECHE
The thesis is submitted in partial fulfilment of the requirements for the award of the degree of
Doctor of Philosophy
Of the
University of Portsmouth
This research was conducted in the Mechanical Behaviour of Materials laboratory, in the School of Engineering at the University of Portsmouth. The research was funded by the Faculty of Technology of the University of Portsmouth and the Engineering and Physical Sciences Research Council (EPSRC) of the UK. The test samples were provided by Rolls-Royce plc in Derby, the UK. The work was in a close collaboration with Cranfield University of the UK.
30 September 2011
2
ABSTRACT
Oxidation damage, in conjunction with fatigue, is a concern for nickel-based superalloys
utilised as disc rotors in high pressure compressor and turbine of aero-engines. A combined
experimental and numerical study has been carried out for alloy RR1000, developed at Rolls-
Royce plc through a powder metallurgy route to meet the demands for higher overall
pressure ratios, compressor discharge temperatures and rotational speeds for the latest
aero-engines.
Cyclic experiments have been carried out for waisted specimens at selected temperatures
(700C-800C), followed by microscopy examination using Focused Ion Beam (FIB). The
results suggest that the major mechanism of oxidation damage consists of the formation of
surface oxide scales and internal micro-voids and oxide particles beneath the oxide scales,
which becomes more severe with the increase of temperature. Applying a cyclic stress does
not change the nature of oxidation damage but tends to enhance the extent of oxidation
damage for temperatures at 750°C and 800°C. Further energy dispersive X-ray (EDX) analyses
show the enrichment of Cr and Ti, together with lower Ni and Co levels, in the surface oxide
scales, suggesting the formation of brittle Cr2O3, TiO2, NiO and Co3O4 oxides on the specimen
surface. Penetration of oxygen into the material and associated internal oxidation, which
leads to further material embrittlement and associated failure, are evidenced from both
secondary ion imaging and EDX analyses.
Scanning electron microscopy (SEM) studies of fracture surfaces have been performed for
dwell crack growth, and the results confirmed the transition from transgranular to
3
predominant intergranular cracking in alloy RR1000 for increased dwell time. This change in
fracture mode supports the oxidation-assisted crack growth mechanism via grain boundary
embrittlement. Oxidation embrittlement has also been supported by the FIB analyses of
fracture surfaces which confirmed the oxidation reaction for alloy RR1000 at high
temperatures.
A stress-assisted diffusion approach has been used to model oxygen penetration in the
waisted specimen based on Fick’s first and second laws. Grain microstructures were
considered explicitly in the model using a finite element submodelling technique, and the
grain boundary was taken as the primary path for oxygen diffusion. The material constitutive
behaviour was described by a crystal plasticity model to consider the effects of
heterogeneous deformation at grain level on oxygen diffusion. Two essential diffusion
parameters, i.e., oxygen diffusivity and pressure factor, have been obtained from the
simulated oxygen penetration as well as the FIB measurements of internal oxidation depth.
Using the obtained diffusion parameters, finite element analyses of a compact tension
specimen have been carried out to model oxygen diffusion, coupled with viscoplastic
deformation, near a fatigue crack tip. A failure curve for crack growth has been constructed
based on the consideration of both oxygen concentration and accumulated inelastic strain
near the crack tip. The failure curve was then utilized to predict crack growth rates under
fatigue-oxidation conditions for selected loading frequencies and dwell periods, with
comparison against the experimental results and those predicted from the viscoplastic model
alone.
4
TABLE OF CONTENTS
List of figures ............................................................................................................................................................ 9
Chapter 1 ................................................................................................................................................................ 18
1.1. Introduction .............................................................................................................................................. 18
1.2. References ................................................................................................................................................ 22
Chapter 2 ................................................................................................................................................................ 26
2.1. Introduction .............................................................................................................................................. 26
2.2. Oxidation of nickel and its alloys ............................................................................................................... 28
2.3. Damage caused by oxidation .................................................................................................................... 31
2.4. Effect of mechanical stress on oxidation .................................................................................................. 35
2.5. Oxidation-assisted crack growth ............................................................................................................... 37
2.5.1. Time-dependent crack growth due to oxidation .............................................................................. 37
2.5.2. Crack tip embrittlement mechanism—short range diffusion .......................................................... 39
2.5.3. Crack tip embrittlement mechanism—long range diffusion ............................................................ 41
2.5.4. Interaction between crack tip field and oxidation ........................................................................... 42
2.5.5. Crack growth prediction models ...................................................................................................... 44
2.6. Dynamic embrittlement mechanism ........................................................................................................ 47
2.7. Review of research on alloy RR1000 ......................................................................................................... 51
2.7.1. Crack growth study ........................................................................................................................... 52
2.7.2. Cyclic constitutive behaviour ........................................................................................................... 52
2.7.3. Modelling of crack tip deformation and growth .............................................................................. 55
2.7.4. Oxidation work ................................................................................................................................. 57
2.8. Conclusions ............................................................................................................................................... 60
2.9. References ................................................................................................................................................ 61
Chapter 3 ................................................................................................................................................................ 74
3.1. Introduction .............................................................................................................................................. 74
3.2. Material, specimen and experimental details........................................................................................... 76
5
3.2.1. Material ............................................................................................................................................ 76
3.2.2. Specimen geometry.......................................................................................................................... 76
3.2.3. Fatigue tests ..................................................................................................................................... 77
3.2.4. FIB analyses ...................................................................................................................................... 78
3.2.5. EDX analyses of the oxides ............................................................................................................... 80
3.3. Results ....................................................................................................................................................... 81
3.3.1. Stress analysis ................................................................................................................................... 81
3.3.2. Fatigue experiments ......................................................................................................................... 84
3.3.3. FIB examinations .............................................................................................................................. 86
3.3.4. Effect of stress level on oxidation damage ....................................................................................... 91
3.3.5. Composition of oxides ...................................................................................................................... 93
3.4. Discussions ................................................................................................................................................ 95
3.5. Conclusions ............................................................................................................................................... 99
3.6. References .............................................................................................................................................. 101
Chapter 4 .............................................................................................................................................................. 105
4.1. Introduction ............................................................................................................................................ 105
4.2. Microscopy analyses of CT specimens .................................................................................................... 106
4.2.1. SEM fractography ........................................................................................................................... 108
4.2.2. FIB examination .............................................................................................................................. 110
4.3. Microscopy analyses of CN specimens .................................................................................................... 114
4.3.1. SEM fractography ........................................................................................................................... 116
4.3.2. FIB examination .............................................................................................................................. 118
4.4. Conclusions ............................................................................................................................................. 124
4.5. References .............................................................................................................................................. 126
Chapter 5 .............................................................................................................................................................. 130
5.1. Introduction ............................................................................................................................................ 130
5.2. Modelling of oxygen diffusion................................................................................................................. 131
6
5.2.1. Governing equations ...................................................................................................................... 131
5.2.2. Oxygen flux boundary conditions ................................................................................................... 132
5.2.3. Crystal plasticity ............................................................................................................................. 132
5.2.4. Model parameters and UMAT ........................................................................................................ 133
5.2.5. Finite element model ..................................................................................................................... 134
5.3. Results and discussions ........................................................................................................................... 136
5.3.1. Natural diffusion ............................................................................................................................. 136
5.3.2. Effect of mechanical stress on oxygen diffusion ............................................................................ 140
5.3.3. Determination of pressure factor................................................................................................... 142
5.4. Conclusion ............................................................................................................................................... 145
5.5. References .............................................................................................................................................. 146
Chapter 6 .............................................................................................................................................................. 150
6.1. Introduction ............................................................................................................................................ 150
6.2. Viscoplasticity deformation analysis ....................................................................................................... 151
6.2.1. The material model ........................................................................................................................ 152
6.2.2. Finite element analysis ................................................................................................................... 154
6.2.3. Near-tip inelastic deformation ....................................................................................................... 156
6.2.4. Prediction of crack growth ............................................................................................................. 158
6.3. Coupled viscoplasticity-oxygen diffusion analysis .................................................................................. 162
6.3.1. Near-tip oxygen concentration ...................................................................................................... 163
6.3.2. Failure curve ................................................................................................................................... 165
6.3.3. Prediction of crack growth under fatigue-oxidation condition ...................................................... 167
6.4. Discussion ................................................................................................................................................ 174
6.5. Conclusions ............................................................................................................................................. 177
6.6. References .............................................................................................................................................. 178
7. Conclusions & Further Work ....................................................................................................................... 182
7.1. Conclusions ......................................................................................................................................... 182
7
7.2. Further work ....................................................................................................................................... 183
Appendix A............................................................................................................................................................186
Appendix B............................................................................................................................................................187
8
DECLARATION
Whilst registered as a candidate for the above degree, I have not been registered for any
other research award. The results and conclusions embodied in this thesis are the work of
the named candidate and have not been submitted for any other academic award. Some of
the results presented in this thesis have already been published.
Signed:
Date:
9
LIST OF FIGURE
Figure 2. 1: External oxide and internal voids formed in Ni 270 after air exposure at 1000°C for 200 hours [58]. 32
Figure 2. 2: Cracking of sub-surface carbides due to oxidation swelling [62]. ....................................................... 33
Figure 2. 3: Fatigue crack propagation behaviour in IN718 at ΔK=27.5MPa√m *74+. ........................................... 38
Figure 2. 4: Fatigue crack growth rate in air and vacuum at 650°C [87]. ............................................................... 38
Figure 2. 5: (A) Intergranular and (B) Transgranular fracture surfaces for crack growth in air and vacuum for Alloy
718 at 650°C [87]. ................................................................................................................................................... 39
Figure 2. 6: Fatigue crack growth rate as function of oxygen partial pressure [90]. .............................................. 40
Figure 2. 7: A schematic of the kinetic model [106]. .............................................................................................. 47
Figure 2. 8: At 650°C, cracking in 1atm of oxygen was completely arrested in about 10s, the time necessary to
evacuate the testing chamber below 10-2Pa. Cracking resumed when oxygen was re-admitted to the chamber
[107]. ...................................................................................................................................................................... 49
Figure 2. 9: Comparison of model simulations and experimental results: (a). Stress-strain loop at saturation
(dε/dt=0.05%/s and Δε=2%) (b). Stress relaxation behaviour during 100-second dwells at maximum and
minimum strain levels (dε/dt=0.5%/s and Δε=2%) *132+. ...................................................................................... 54
Figure 2. 10: Contour plot of the von Mises equivalent strain at the maximum strain level of the 2nd cycle (strain
rate dε/dt=0.005%/s, strain range Δε=20%, strain ratio =-1) [132]. ...................................................................... 54
Figure 2. 11: (a). The evolution of stress-strain loops over 30 cycles ahead of a stationary crack tip. (b). The
evolution of stress-strain loops registered at an integration point O as the crack grew towards it [133]. ........... 55
Figure 2. 12: Focused ion beam image of an oxidised RR1000 surface, exposed to 700°C for 200 hours [48]. .... 58
Figure 2. 13: Oxide layer on the fracture surface during dwell fatigue tests at 725°C [98]. .................................. 59
Figure 3. 1: Waisted specimen geometry and the three positions “A”, “B” and “C” selected for FIB examinations.
Drawing provided by Rolls-Royce Plc. Internal communication. ........................................................................... 77
Figure 3. 2: Waisted specimen surface, the Pt strip applied to protect the oxide scale and the first stage of rough
milling. .................................................................................................................................................................... 79
Figure 3. 3: (a). Secondary electron image (b). Secondary ion image. ................................................................... 80
Figure 3. 4: (a). Contour plot of the normal stress in the y direction for the waisted specimen under an applied
load of 10kN, where d is the distance from the narrowest section of the specimen (b). Stress concentration
factor, defined as the ratio of the normal stress in the y-direction (σyy) to the applied stress (σapp = 39.0MPa),
against the distance from the narrowest section of the specimen. ...................................................................... 83
10
Figure 3. 5: (a). The fracture surface of the specimen fatigue tested at 750°C. (b). Surface cracks on the side
surface of the specimen fatigue tested at 800°C. .................................................................................................. 85
Figure 3. 6: FIB images for position "A" at 700°C (a) secondary electron and (b) secondary ion, where the dashed
line in red indicates the original specimen surface. ............................................................................................... 88
Figure 3. 7: FIB images for position "A" at 750°C (a) secondary electron and (b) secondary ion, where the dashed
line in red indicates the original specimen surface. ............................................................................................... 89
Figure 3. 8: FIB images for position "A" at 800°C (a) secondary electron and (b) secondary ion, where the dashed
line in red indicates the original specimen surface. ............................................................................................... 90
Figure 3. 9: (a). Comparison of the thickness of the oxide scale and depth of internal damage, for positions A, B
and C at 700°C. (b). Comparison of thickness of oxide scale and depth of internal damage for positions A, B and
C at 750°C. (c). Comparison of thickness of oxide scale and depth of internal damage for positions A, B and C at
800°C. ..................................................................................................................................................................... 92
Figure 3. 10: Five spectrums selected for EDX analyses, over a depth of 2.2μm from the oxide scale surface and
with equal spacing (0.54μm) between each other. ............................................................................................... 93
Figure 3. 11: (a). EDX element profiles showing weight% of Cr, Ni, Ti and Co against the distance from the oxide
scale surface (750°C). (b). EDX element profiles showing weight% of O, Mo, Al and Ta against the distance from
the oxide scale surface (750°C). ............................................................................................................................. 94
Figure 3. 12: (a). EDX element profiles showing weight% of Cr, Ni, Ti and Co against the distance from the oxide
scale surface (800°C). (b). EDX element profiles showing weight% of O, Mo, Al and Ta against the distance from
the oxide scale surface (800°C). ............................................................................................................................. 94
Figure 4. 1: Fracture surface of a CT specimen tested at T=650°C for variable dwell times. ............................... 107
Figure 4. 2: SEM micrographs of the fractured surface of a CT specimen tested at 650°C under (a) 20s, (b) 300s
and (c) 1000s dwell. ............................................................................................................................................. 109
Figure 4. 3: SEM micrographs of the fractured surface of a CT specimen tested at 725°C under (a) 20s, (b) 300s
and (c) 1000s dwell. ............................................................................................................................................. 110
Figure 4. 4: Micrograph of the selected FIB location, which also shows the severe surface roughness. ............ 111
Figure 4. 5: The secondary electron image for FIB analysis which reveals a very thin and light oxide film built on
the surface. .......................................................................................................................................................... 112
Figure 4. 6: The secondary ion image for FIB analysis, showing the surface oxide scale and penetration of oxides
into the material along the path of a secondary crack. ....................................................................................... 113
Figure 4. 7: The secondary ion image for FIB analysis, which reveals a thicker oxide scale, built on the fracture
surface (Testing temperature at 725°C). .............................................................................................................. 113
Figure 4. 8: Fractured surface of the CN specimen tested at 650°C, with indications of the three FIB positions.
.............................................................................................................................................................................. 115
11
Figure 4. 9: Fractured surface of the CN specimen tested at 700°C, with indications of the three FIB positions.
.............................................................................................................................................................................. 115
Figure 4. 10: Fractured surface of the CN specimen tested at 750°C, with indications of the three FIB positions.
.............................................................................................................................................................................. 116
Figure 4. 11: SEM pictures of the fractured surface of the CN specimen R134 (a) room temperature fatigue and
(b) 1 hour dwell at 650°C. .................................................................................................................................... 117
Figure 4. 12: SEM pictures of the fractured surface of the CN specimen R134 (a) room temperature fatigue and
(b) 1 hour dwell at 700°C. .................................................................................................................................... 117
Figure 4. 13: SEM pictures of the fractured surface of the CN specimen tested at 750°C: (a) 0.25Hz followed by 1
hour dwell and (b) 1 hour dwell. .......................................................................................................................... 118
Figure 4. 14: Secondary Ion images at FIB position 1 for the specimens tested under one-hour dwell at (a) 650°C,
(b) 700°C and (c) 750°C. The colour lines indicate the surface oxide boundaries. .............................................. 119
Figure 4. 15: Secondary Ion images at FIB position 2 for the specimens tested under one-hour dwell at (a) 650°C,
(b) 700°C and (c) 750°C. The colour lines indicate the surface oxide boundaries. .............................................. 120
Figure 4. 16: Secondary Ion images at FIB position 3 for the specimens tested under one-hour dwell at (a) 650°C,
(b) 700°C and (c) 750°C. The coloured lines indicate the surface oxide boundaries. .......................................... 121
Figure 5. 1: (a) The 150-grain submodel with random orientation (b) Finite element mesh for the global model
and (c) Inverse pole figure for the 150 grains. ..................................................................................................... 135
Figure 5. 2: For 650°C the diffusion value has been determined using the Arrhenius equation and the
experimental values obtained for 700, 750 and 800°C, as shown on the graph. ................................................ 137
Figure 5. 3: Contour plots of oxygen concentration along the grain boundaries for natural diffusion: (a) 650°C
and (b) 800°C. ....................................................................................................................................................... 138
Figure 5. 4: Oxygen concentration against the distance from the specimen surface for four different
temperatures. ...................................................................................................................................................... 140
Figure 5. 5: Contour plot of the hydrostatic stress (peak load level) for the submodel after 33473 cycles at 800°C
for an applied load of 8kN. ................................................................................................................................... 141
Figure 5. 6: Contour plot of the oxygen concentration in the submodel at 800°C after 33473 cycles: (a) natural
diffusion and (b) with a pressure factor of 0.3MPa-1
. .......................................................................................... 142
Figure 5. 7: Depth of oxygen penetration against the pressure factor for T=750°C, where the penetration depth
(δ) under fatigue is normalized against that (δ0) for natural diffusion. ............................................................... 143
Figure 5. 8: Depth of oxygen penetration against the pressure factor for T=800°C, where the penetration depth
(δ) under fatigue is normalized against that (δ0) for natural diffusion. ............................................................... 144
12
Figure 6. 1:(a) Finite element mesh for a CT specimen and (b) refined mesh for the crack tip area (element size is
6.35μm). ............................................................................................................................................................... 156
Figure 6. 2: Contour plot of the accumulated inelastic strain near the crack tip after ten loading cycles for a
stress intensity factor range ΔK=30MPa√m and a loading frequency f=0.25Hz. ................................................. 157
Figure 6. 3: Contour plot of the accumulated inelastic strain near the crack tip after ten loading cycles for a
stress intensity factor range ΔK=30MPa√m and a dwell period of 1000s............................................................ 158
Figure 6. 4: Effects of loading frequency on crack growth rate for a triangular loading waveform with ΔK =
30MPa√m and load ratio R = 0.1 at 650°C, comparison of model prediction and experimental results *11, 12+.
.............................................................................................................................................................................. 160
Figure 6. 5: Effects of dwell periods on crack growth rate for a trapezoidal loading waveform with ΔK =
30MPa√m and load ratio R = 0.1 at 650°C, comparison of model prediction and experimental results [11, 12].
.............................................................................................................................................................................. 161
Figure 6. 6: (a) Contour plot of the oxygen concentration near the crack tip after 10 cycles for a frequency of
0.001Hz. ............................................................................................................................................................... 164
Figure 6. 7: Failure curve in terms of the accumulated inelastic strain and the oxygen concentration near the
crack tip (ΔΚ=30ΜPa√m, R=0.1, T=650°C). .......................................................................................................... 167
Figure 6. 8: Illustration of determining the number of cycles for the crack to grow, i.e., when the combination of
the accumulated inelastic strain and the oxygen concentration reached the failure curve, for different
frequencies. .......................................................................................................................................................... 169
Figure 6. 9: Illustration of determining the number of cycles for the crack to grow, i.e., when the combination of
the accumulated inelastic strain and the oxygen concentration reached the failure curve, for different dwell
time. ..................................................................................................................................................................... 170
Figure 6. 10: The effects of loading frequency on the crack growth rate for a triangular waveform with
ΔK=30MPa√m and load ratio R = 0.1 at 650°C; comparison of the model prediction and the experimental results
[11, 12]. ................................................................................................................................................................ 172
Figure 6. 11: The effects of dwell period on the crack growth rate for a trapezoidal loading waveform with
ΔK=30MPa√m and load ratio R = 0.1 at 650°C; comparison of the model prediction and the experimental results
[11, 12]. ................................................................................................................................................................ 173
13
DISSEMINATION
A. Karabela, L.G. Zhao, J. Tong, N.J. Simms, J.R. Nicholls and M.C. Hardy. Effects of
cyclic stress and temperature on oxidation damage of a nickel-based superalloy.
Materials Science and Engineering: A, 528(2011), 6194-6202.
Tong, J., Cornet, C., Lupton, C. and Karabela, A. (2010). Developing improved service
propagation lives in arduous cyclic environment (DISPLACE) project. TSB Project.
University of Portsmouth, Portsmouth, UK.
"Fatigue-oxidation interaction for a nickel-based superalloy", at EUROCORR'2010
(European Corrosion Congress, Moscow, Russia, 13-17 September 2010) was awarded
a Young Author's prize by the Russian Scientific Committee, with 500EUR prize.
14
ACKNOWLEDGEMENTS
My sincere gratitude to my supervisor Dr Liguo Zhao, who’s guidance and support have been
of great value and without whom this PhD thesis wouldn’t have been possible.
My gratitude also for my second supervisor Prof Jie Tong, who has been my reference point
as a woman and as an engineer, since I first arrived in Portsmouth.
Many thanks to Mr Colin Lupton, for his help and friendship, his advices and interest were
always very much appreciated.
Thank also to all the members of the Mechanical Behaviour of Materials Research group for
their friendship and support whenever it was needed, especially Dr Bing Lin with his help
with ABAQUS when it was needed.
The work has been funded by the EPSRC (Grant EP/E062180/1) of the UK and in a close
collaboration with Rolls-Royce plc. (Dr M.C. Hardy) and Cranfield University (Professor J.R.
Nicholls and Dr N.J. Simms) of the UK. The bursary and tuition fees have been provided by
the Faculty of Technology of the University of Portsmouth.
The waisted specimens were provided by Rolls-Royce plc of the UK.
Support has been provided by Dr Adriana Encinas-Oropesa and Mr Andrew Dyer in the use of
the FIB facilities at Cranfield University of the UK.
The crystal plasticity UMAT was originally developed by Professor Esteban Busso and his
associates, while he was with the Imperial College London of the UK.
15
Finally many thanks to all my friends here in Portsmouth, especially Dr Eleni Noussi, Dr Nick
Savage and Mr Gianluca Tozzi for their help, advices, and constant moral support on personal
matters that occurred these past few months of my PhD.
16
DEDICATION
To my parents Alexios and Anastasia
17
CHAPTER 1
INTRODUCTION
18
CHAPTER 1
1.1. INTRODUCTION
Increasing turbine entry temperatures and rotational speeds in order to increase fuel
efficiency and to reduce NOx/CO2 emissions are the pursuits of modern development and
design of aero engines. As a consequence, demand for increased performance has been
critically placed upon the nickel based superalloy materials, which are used in the hot section
of the engine, particularly the critical rotating turbine discs. RR1000 is one of the new
generation nickel alloys, developed by Rolls-Royce plc of the UK through the powder
metallurgy process, to meet such demands. Very recently, RR1000 has been put into
application for the turbine discs of the latest Rolls-Royce aircraft engines, which were
designed towards higher performance and increased failure resistance, to power the
dreamliner Boeing 787 aircrafts.
During a typical flight cycle, turbine discs are subjected to variable centrifugal and thermal
stresses with operation temperatures reaching up to 700C at the rim. Hence, failure of
turbine discs could result from the interaction of fatigue, creep and environmental effects.
Fatigue and creep failure behaviour of RR1000 at high temperatures was considerably
studied before its very recent application. Effects of load ratios, frequency, wave form and
dwell period on fatigue crack growth have been extensively studied in the Mechanical
Behaviour of Materials laboratory at the University of Portsmouth, and data and models are
already available for life assessment of RR1000 turbine discs [1-3]. In addition, experimental
and analytical studies on the constitutive behaviour (stress-strain relationship) of RR1000
19
have also been carried out in the MBM research group at Portsmouth [4-10] and applied for
the study of fatigue and creep crack growth behaviour [11-13]. However, environmental
effects on the crack growth behaviour for RR1000 [14, 15], which is a crucial factor for life
assessment of RR1000 turbine discs, have not been well studied yet. Crack growth
predictions in Zhao and Tong [16], based on viscoplastic deformation only, showed a great
discrepancy with experimental data for low frequencies and long dwell periods, which was
due to the neglect of oxidation effects at high temperature.
Recently, Encinas-Oropesa et al. [17] studied the oxidation of alloy RR1000 using circular disk
specimens that have been isothermally exposed to air at 700°C, 750°C and 800°C for 200
hours. Focused Ion Beam (FIB) examinations of oxidized samples for RR1000 revealed the
appearance of the oxide scale on the surface and the formation of recrystallized grains and
voids beneath the oxide scales [17]. The oxide scale, rich in Cr and Ti, has been found to grow
in a parabolic dependence with time. However, the work of Encinas-Oropesa et al. [17] was
carried out without considering the influence of mechanical loading such as fatigue, which
can have a significant effect on oxidation and corrosion damage for nickel and its alloys.
For polycrystalline turbine-disc made of nickel superalloys environmental effects, in
particular oxidation, were frequently observed to modify the fracture morphology from
transgranular to intergranular during fatigue crack growth, as well as to increase the growth
rates significantly at a given crack-tip stress intensity [18-21]. Several prediction models have
been developed for different nickel alloys based on different views of the fatigue-oxidation
interaction process, such as the oxide-depth based model [21], the oxide-passivation
(formation of protective Cr2O3) based model [22] and the multiplicative mechanical-
20
environmental damage model [23]. However, these life prediction models for crack growth
are highly empirical.
It has been widely acknowledged that oxygen-accelerated crack propagation behaviour in
nickel alloys is a result of oxygen diffusion into a local area with increasing tensile stress, such
as crack tips. The oxidising elements, by themselves or through chemical reactions with alloy
elements, attack the grain boundaries by lowering boundary cohesion and tend to introduce
accelerated intergranular cracking [e.g. 18, 20, 21, 24-26]. Oxygen diffusion at the crack tip is
a dynamic phenomenon, a combined effect of time, temperature and local deformation [24-
26]. This approach has been shown encouraging to predict crack growth rate for a nickel alloy
under fatigue oxidation conditions [27].
The aim of this PhD program was to study the oxidation damage under fatigue, and to model
crack growth under fatigue-oxidation conditions using a dynamic embrittlement approach for
alloy RR1000. Funded by the EPSRC [28], the work has been in close collaboration with Rolls-
Royce plc and Cranfield University. The objectives were:
To carry out a literature review on oxidation and its effect on crack growth for nickel
alloys (Chapter 2).
To carry out fatigue tests on waisted specimens at elevated temperature at
Portsmouth and examine fatigue-oxidation damage using the FIB and EDX facilities at
Cranfield (Chapter 3).
21
To carry out post-mortem SEM and FIB studies on fracture surfaces of compact
tension (CT) and corner notch (CN) specimens that have been tested under fatigue at
Portsmouth (Chapter 4).
To calibrate the diffusion parameters from oxygen diffusion analysis of waisted
specimen at grain level via a coupled crystal plasticity-diffusion approach (Chapter 5).
To carry out oxygen diffusion analysis, coupled with viscoplasticity deformation, near
the crack tip, and develop a model for crack growth prediction under fatigue-
oxidation conditions (Chapter 6).
To conclude the work and make suggestions for future work (Chapter 7).
The work is of great practical importance for “Damage Tolerance” lifing of critical rotating
turbine discs [29]. The investigation will provide a fundamental and scientific understanding
of oxidation damage and associated fast crack growth in the fatigue-creep regime for alloy
RR1000. RR1000 is representative of the best powder metallurgy, fine grained nickel-based
superalloys for service as aero-gas turbine discs. Such studies will provide support to Rolls-
Royce in ensuring the structural integrity of aero-engines through a close collaboration.
22
1.2. REFERENCES
[1]. Dalby, S. Tong, J. (2005). Crack growth in a new nickel-based superalloy at elevated
temperature, Part I: Effects of loading waveform and frequency on crack growth. Journal
of Materials Science, 40(5), 1217-1228.
[2]. Tong, J., Dalby, S. and Byrne, J. (2005). Crack growth in a new nickel-based superalloy
at elevated temperature, Part III: Characterisation. Journal of Materials Science, 40(5),
1237-1243.
[3]. Dalby, Sara. (2002). The effects of frequency and waveshape on the fatigue crack
growth of an advanced nickel based superalloy at elevated temperatures. PhD Thesis.
University of Portsmouth, Portsmouth, UK.
[4]. Zhao, L.G., Tong, J., Vermeulen, B. and Byrne, J. (2001). On the uniaxial mechanical
behaviour of an advanced nickel base superalloy at high temperature. Mechanics of
Materials, 33(10), 593-600.
[5]. Tong, J. and Vermeulen, B. (2003). The description of cyclic plasticity and
viscoplasticity of Wasploy using unified constitutive equations. International Journal of
Fatigue, 25(5), 413-420.
[6]. Tong, J., Zhan, Z.-L. and Vermeulen, B. (2004). Modelling of cyclic plasticity and
viscoplasticity of a nickel-based alloy using Chaboche constitutive equations, International
Journal of Fatigue, 26, 829-837 (2004).
[7]. Zhan, Z.-L. (2004). A study of cyclic plasticity and viscoplasticity in a new nickel-based
superalloy using unified constitutive equations. PhD Thesis. University of Portsmouth,
Portsmouth, UK.
[8]. Zhan, Z.-L., Tong J. (2007). A study of cyclic plasticity and viscoplasticity in a new
nickel-based superalloy using unified constitutive equations. Part I: Evaluation and
determination of material parameters. Mechanics of Materials, 39(1), 64-72.
[9]. Zhan, Z.-L., Tong J. (2007). A study of cyclic plasticity and viscoplasticity in a new
nickel-based superalloy using unified constitutive equations. Part II: simulation of cyclic
stress relaxation. Mechanics of Materials, 39(1), 73-80.
[10]. C. Cornet, L.G. Zhao and J. Tong. (2011). A study of cyclic behaviour of a nickel-
based superalloy at elevated temperature using a viscoplastic-damage model.
International Journal of Fatigue, 33(2), 241-249.
23
[11]. Zhao, L.G., Tong, J. and Byrne, J. (2001). Finite element simulation of creep-crack
growth in a nickel base superalloy. Engineering Fracture Mechanics, 68(10), 1157–1170.
[12]. Zhao, L.G., Tong, J. and Byrne, J. (2004). The evolution of the stress-strain fields near
a fatigue crack tip and plasticity-induced crack closure revisited. Fatigue and Fracture of
Engineering Materials and Structures, 27(1), 19-29.
[13]. Cornet, C., Zhao, L.G. and Tong, J. (2009). Ratchetting strain as a damage
parameter in controlling crack growth at elevated temperature. Engineering Fracture
Mechanics, 76(16), 2538-2553.
[14]. Tong, J., Dalby, S., Byrne, J., Henderson, M.B. and Hardy, M.C. (2001). Creep,
fatigue and oxidation in crack growth in advanced nickel base superalloys. International
Journal of Fatigue, 23(10), 897-902.
[15]. Knowles, D.M. and Hunt, D.W. (2002). The influence of microstructure and
environment on the crack growth behaviour of powder metallurgy nickel superalloy
RR1000. Metallurgical and Materials Transactions A, 33(10), 3165-3172.
[16]. Zhao, L.G., Tong, J. (2008). A viscoplastic study of crack-tip deformation and crack
growth in a nickel-based superalloy at elevated temperature. Journal of the Mechanics
and Physics of Solids, 56(12), 3363-3378.
[17]. Encinas-Oropesa, A., Drew, G.L., Hardy, M.C., Leggett, A.J., Nicholls, J.R., Simms,
N.J. (2008). Effects of oxidation and hot corrosion in a nickel disc alloy. Champion,
Pennsylvania, USA : Wiley, Superalloys 2008. 609-618.
[18]. Andrieu, E., Hoshstetter, G., Molins, R. and Pineau, A. (1997). Oxidation and
intergranular cracking behaviour of two high strength Ni base superalloys. Corrosion
Deformation Interaction (EFC 21), edited by T. Magnin, 461-475.
[19]. Molins, R., Hochstetter, G., Chassaigne, J.C. and Andrieu, E. (1997). Oxidation
effects on the fatigue crack growth behaviour of alloy 718 at high temperature. Acta
Materialia, 45(2), 663-674.
[20]. Andrieu, E., Molins, R., Ghonem, H. And Pineau, A. (1992). Intergranular crack tip
oxidation mechanism in a nickel-based superalloy. Materials Science and Engineering: A,
154(1), 21-28.
[21]. Ghonem, H., Zheng, D. (1992). Depth of intergranular oxygen diffusion during
environment-dependent fatigue crack growth in alloy 718. Materials Science and
Engineering: A, 150(2), 151-160.
24
[22]. Ghonem, H. and Zheng, D. (1991). Oxidation-assisted fatigue crack growth behavior
in alloy 718-Part I. Quantitative modeling. Fatigue & Fracture of Engineering Materials &
Structures, 14(7), 749-760.
[23]. Miller, M. P., McDowell, D. L. and Oehmke, R. L. T. (1992). A creep-fatigue-
oxidation microcrack propagation model for thermomechanical fatigue. Journal of
Engineering Materials and Technology. 114, 282-288.
[24]. Krupp, U. (2005). Dynamic embrittlement: time-dependent quasi-brittle
intergranular fracture at high temperatures. International Materials Reviews, 50(2), 83-
97.
[25]. Krupp, U., Kane, W.M., Laird, C., McMahon Jr., C.J. (2004). Brittle intergranular
fracture of a Ni-base superalloy at high temperatures by dynamic embrittlement.
Materials Science and Engineering A, 13th International Conference on the Strength of
Materials, 387-389, 409-413.
[26]. Pfaendtner, J.A., McMahon Jr., C.J. (2001). Oxygen-induced intergranular cracking
of a Ni-base alloy at elevated temperatures—an example of dynamic embrittlement. Acta
Materialia, 49(16), 3369-3377.
[27]. Zhao, L.G., Tong, J., and Hardy, M.C. (2010). Prediction of crack growth for a nickel
base superalloy under fatigue-oxidation conditions. Engineering Fracture Mechanics,
77(6), 925-938.
[28]. Zhao, L.G. EPSRC grant (EP/E062180/1), “A micro-mechanistic study of oxygen-
diffusion-assisted crack growth in a polycrystalline nickel-based superalloy”, £196k,
11/2007-10/2010.
[29]. King, J.E. (1987). Fatigue crack propagation in nickel-base superalloys — effects of
microstructure, load ratio and temperature. Materials Science and Technology, 3(9), 750-
764.
25
CHAPTER 2
LITERATURE REVIEW
26
CHAPTER 2
2.1. INTRODUCTION
Due to their superior mechanical properties at elevated temperature, nickel based superalloy
are favoured for applications in rotating blades and discs in gas turbine engines. It has been
more than 70 years since their initial use as blade materials in aeroengines [1]. Gas turbine
discs are subjected to extremely severe operating conditions during the repeated start-up,
take-off and climb, cruising, landing and shut-down cycles. These include variable centrifugal
and thermal stress as well as the prolonged application of load in the presence of corrosive
exhaust gases at elevated temperature up to 700C. Consequently fatigue, creep and
environmental factors, as well as their interaction, are the major concerns for gas turbine
safety.
Investigation of fatigue [2-4], creep [5, 6] and fatigue-creep interaction [7-11] of nickel alloys
has been carried out using a variety of experimental, analytical and numerical approaches. In
particular, unified constitutive models, with both isotropic and kinematic hardening internal
variables, have been rapidly developed to facilitate the safe life assessment of nickel alloy
discs under creep-fatigue conditions [7-17]. On the other hand, considerable experimental
and modelling work has also been carried out to study the fatigue/creep crack growth in
nickel alloys. It is well known that the time-independent crack growth (also known as cycle-
dependent crack growth or fatigue crack growth) is predominantly transgranular and
depends on load ratio, maximum load and temperature [1, 18-31]. Material microstructure
including grain orientation exerts a strong influence on transgranular crack propagation and
27
the crack growth rate [e.g., 1, 18, 21-25, 32]; While for creep crack growth the principal
mechanisms are void growth and coalescence at grain boundaries under applied stresses. In
this case the fracture morphology is dominantly intergranular [33, 34]. Floreen [35] found
from a variety of nickel-base superalloys that creep crack growth rate is proportional to a
power of the stress intensity factor. Furthermore, the presence of a dwell (i.e., creep
deformation) at maximum load in fatigue cycle was shown to increase crack growth rates.
For example, Wei [36] conducted experiments on nickel-based superalloys in high purity
argon and oxygen at 873-973K, and showed that crack growth rates increased linearly with
increasing dwell time. Nicholas and Weerasooriya [37] showed that not only the dwell load
level played a significant effect on crack growth rate, but also the frequency at which the
dwell was imposed. The R-ratio was also shown to be an influencing factor for dwell time
effects. The effect of dwell periods at higher R-ratio values is greater than that at lower R-
ratio values [38].
These crack growth studies are of particular importance for damage tolerance assessment of
nickel alloy discs. Apart from creep and fatigue, oxidation damage, which is also the focus of
this work, is another crucial factor that contributes to the failure of turbine discs. Oxidation
affects the fracture morphology by inducing a passage from slow transgranular to fast
intergranular fracture [e.g., 39, 40, 41], which might lead to premature failure of turbine
discs. In this chapter, a literature review has been carried out regarding the existing research
work on oxidation damage and oxidation-assisted crack growth in nickel and its alloys.
28
2.2. OXIDATION OF NICKEL AND ITS ALLOYS
In 1936, Smithells and Ransley [42] studied nickel oxidation between 900°C and 1050°C using
thin-wall tubes. Their micrographs appeared to confirm grain-boundary oxygen penetration.
The results demonstrated that oxygen can penetrate large distances into materials by
preferential grain-boundary diffusion. Giggins and Pettit [43] studied the oxidation of Ni-Cr-Al
alloys and suggested that the mechanism of oxidation in these alloys could be divided into
three categories, based on the amount of Cr and Al that they contain, as follows:
Type I: An external layer of NiO is formed, with a subscale of Cr2O3, Al2O3 and /or
Ni(Cr,Al)2O4 when the alloy contain small amounts of Cr and Al.
Type II: An external layer of Cr2O3, with a subscale of Al2O3, when the alloy
contains higher Cr but lower Al.
Type III: The only scale formed is Al2O3 with no signs of internal oxidation when
the alloy contains sufficient Al.
These are rather generic categories for oxidation of nickel alloys, and they give an idea of the
oxide scale formation based on the contents of Cr and Al.
Literature [44-47] agrees that there is a preferential oxidation of the various alloying
elements (Nb, Ti, Al, Cr and Ni) in nickel alloys. The formed oxide layers are usually an
external layer of NiO and Cr2O3, a sub layer of Al2O3 and TiO2 and various carbides (TixNby)C.
Encinas-Oropesa et al. [48] studied the oxidation of a powder metallurgy nickel alloy in air at
700°C, 750°C and 800°C. Energy Dispersive X-ray (EDX) analyses showed a progressive
29
enrichment of Cr and Ti in the oxide scales with increasing time and temperature, coupled
with lower Ni and Co levels. In Geng et al. [45], a study of the surface oxidation behaviour of
Alloy 718 was conducted, by hot extruding and isothermally exposing the alloy at 1100°C.
The oxidation time was varied from 10 to 180h. Three distinct surface layers were observed
depending on the oxidation time. An outside layer of a single phase Nb-rich Cr oxides with
some voids and microcracks was formed after an exposure of 80h. A second layer with a
small amount of discontinuous phase of Al2O3 was observed after 10h, and a third layer with
complicated shapes and carbides, composed of various (TixNby)C with different shapes, was
observed after 40h of exposure.
Chen et al. [44] studied the oxidation of three different alloys, Astroloy, Waspaloy and
Udimet720 (the percentage of Cr for these alloys is less than 20wt %). The alloys were
exposed isothermally in air at a temperature varying between 750-1000°C for up to
1000hours. Oxidation kinetics was determined and scales were analysed by electron
microscopy and X-ray diffraction. Waspaloy showed the lowest weight gain, but the deepest
internal corrosion due to oxidation of the grain-boundary carbides. Energy Dispersive X-ray
Microanalysis (EDX) was used to identify the oxides composition. The major oxides detected
on the external scale were Cr2O3, TiO2 and NiCr2O4 spinel, with NiO occasionally detected
after 20hr at 750°C. Titanium in the alloys also oxidised, diffusing through the chromia
(Cr2O3) scale and formed rutile (TiO2) grains at the surface as well as TiO2 and TiN internally.
Alumina (Al2O3) formed as discrete internal oxides below the chromia scale. The internal
oxides developed predominantly along grain boundaries since they were short-circuit paths
for inward oxygen diffusion.
30
Miller et al. [46] determined the depth of internal oxides at a crack tip for nickel-alloys using
X-ray photoelectron spectroscopy (XPS), and found that the oxygen affected region and the
depth of internal oxides varied according to the alloy. Andrieu et al. [49] suggested two
oxidation mechanisms at a crack tip, namely, short range and long range oxygen diffusion
processes. For short range diffusion, an oxide layer is formed at the crack tip and the
thickness of the oxide layer depends on the operating and material parameters. For long
range diffusion, oxygen can penetrate into the material along grain boundaries and slip
planes. As a result, internal oxidation and cavity formation take place and the oxidation of
various alloying elements results in the release of embrittling agents into the material.
Oxidation is generally quantified by measuring the weight gain of the alloy and the thickness
or depth of oxide scale built on the surface [43-45, 47, 48]. Both can be described using the
parabolic law for either the oxide scale thickness or the weight gain due to oxidation:
where X is the oxide scale thickness or depth, ΔW the mass gain, t the time. The parabolic
oxidation factor follows the Arrhenius law:
where is a constant, Q the activation energy, R universal gas constant and T the
temperature. Materials with low rates of mass gain and growth of oxide scales are more
resistant to oxidation.
31
2.3. DAMAGE CAUSED BY OXIDATION
High temperature exposure in oxygen-containing environments leads to a degradation of the
mechanical properties for nickel alloys, often termed as oxygen embrittlement. Douglass [50]
showed the severe loss in tensile and creep ductility at 600°C following the prior exposure at
1200°C. Chang [51] showed a drastic loss in tensile ductility for a nickel alloy René-80
following air exposure for 150hr at 982°C. He also demonstrated similar effects in two other
nickel base superalloys, René77 and René100. The detrimental effect of prior high
temperature exposure on subsequent stress rupture life and fatigue response has been also
demonstrated for nickel alloys by Woodford [52] and Antolovich et al. [53].
Oxygen embrittlement has been attributed to the penetration of oxygen along the grain
boundaries. Direct evidence of the presence of oxygen on embrittled grain boundaries has
been demonstrated in nickel and its alloys by Auger microscopy [54], as also represented by
intergranular fracture morphology. There are a number of mechanisms by which a boundary
may be embrittled by oxygen penetration. These include oxide formation, cavity formation
and solute segregation. It is also possible for oxygen to take part in chemical reactions
releasing known embrittling agent into a grain boundary. Each or all of these may be
operative in any particular alloy under a given set of conditions.
32
Figure 2. 1: External oxide and internal voids formed in Ni 270 after air exposure at 1000°C
for 200 hours [58].
For instance, the experiments of Woodford and Bricknell [55] showed the presence of small
oxides on nickel grain boundaries. Another consequence of oxygen penetration which has
been studied in nickel is the formation of voids. Such voids have been previously attributed
to vacancy injection from the growing oxides [50, 56 and 57]. While in Bricknell and
Woodford [58, 59], it showed that these voids are in fact gas bubbles as seen in Figure 2.1.
Mrowec et al. [60] studied the oxidation kinetics of pure Ni as well as two binary alloys
composed of Ni-Cr and Ni-Na, respectively. It was shown that the predominant defects in
non stoichiometry Ni oxide are double ionized cation vacancies and electron holes, and the
oxide scale on Ni grows by outward volume diffusion of cations. They also showed that Ni-Cr
alloy has higher oxidation rate, compared to the oxidation rate of pure Ni. Apart from oxides
and void formation, Fournier et al. [61] as well as Reed [62] agreed that oxidised niobium
33
carbides are also responsible for the damage caused in nickel alloys due to their swelling, as
shown in Figure 2.2.
Figure 2. 2: Cracking of sub-surface carbides due to oxidation swelling [62].
In the work of Garat et al. [47], thin tensile specimens of alloy 718 were oxidised under
synthetic air, at 1000°C for different durations, then heat treated according to aeronautical
heat treatment before being tested on a tensile machine at room temperature. Mechanical
tests were performed in order to highlight the influence of the damaged zone upon a global
tensile behaviour. A significant drop in flow stress was seen for oxidised specimens. The
tensile curves suggest that the flow stress generally decreases as the oxidation duration
before mechanical testing increases. The oxide scale and the intergranular oxygen
penetration (a mean depth of about 7μm) were observed on specimen’s cross-sections. The
EDX analysis showed that the external oxide scale is Cr and Ti rich. Close to the oxide-alloy
34
interface Nb enrichment and Cr depletion were observed. The intergranular oxygen
penetration was found to contain Al and O2 as major chemical elements.
It was shown in Garat et al. [47] that the OAZ (oxidation affected zone), consisting of the
oxide scale and the intergranular oxygen penetration, is responsible for initiation of several
types of in-service damaging processes. By assuming that the OAZ is unable to support any
mechanical load, they derived the apparent yield stress from the applied load using the
formula:
where σ is the yield stress, F0, Fi, S0, Si are the corresponding loads and specimen sections
before and after oxidation, e is the thickness, l is the width of the specimen and δe is the
depth affected by oxidation.
Calvarin et al. [63] investigated the interaction between the oxidation mechanism and the
mechanical behaviour of a Ni-20Cr alloy. Tensile tests showed a decrease of ductility with
temperature increase, accompanied by more severe intergranular decohesion. While for
creep tests, the environmental effect only appears at 500°C and beyond this temperature the
behaviour seems to be the same in air and vacuum. The alloy creep resistance is improved in
air at high temperatures and low stresses; on the contrary at high stresses and low
temperatures, the creep behaviour is decreased in air. After creep tests in air, whatever the
temperature is, an external NiO scale systematically developed on the surface substrate. At
500°C-600°C an internal oxidation zone was observed under the oxidised surface. Beyond
35
these temperatures, this zone did not appear but a continuous chromia layer developed at
the oxide-alloy interface. The environmental effect in relation to the oxidation process was
described by two competing mechanisms that explain the effect of an oxide layer on the
mechanical behaviour:
The presence of internal oxides may strengthen the material, which then behaves as
oxide dispersion strengthened alloy with improved creep behaviour [64].
The formation of a scale by outward cationic diffusion towards the external surface of
the substrate (provided that the metal-oxide interface cannot move) might induce
vacancy injection into the alloy, resulting in an increase in the creep rate of the
material.
2.4. EFFECT OF MECHANICAL STRESS ON OXIDATION
In addition, it has been well acknowledged that the mechanical stress can further influence
oxidation and corrosion damage in nickel and its alloys. For instance, Moulin et al. [65]
studied the influence of various mechanical loading (fatigue, creep and creep-fatigue) on
oxygen diffusion during nickel oxidation. It was shown that a relative O enrichment was
observed in the oxide scale under mechanical loading, although with a decrease of oxygen
diffusivity in the oxide scale. In the substrate beneath the oxide scale, the ingress of oxygen
became easier with a constant tensile load (creep) and oxygen diffusion took place along
grain boundaries. The intergranular oxygen diffusion coefficient was increased by a factor of
10 with respect to other samples. A short fatigue period during creep-fatigue decreased the
sensitivity of nickel to intergranular oxygen diffusion.
36
In the work of Berger et al. [66, 67], stress-oxidation of nickel has been studied using a
nuclear microprobe method. The application of a constant load (up to 75MPa) induced
multiple cracks at the surface of oxide scale as well as an increase of oxygen diffusivity by
two orders of magnitude (typically from 10-15cm2/s to around 10-13cm2/s in NiO thermally
grown on nickel monocrystals), although the enhancement of oxygen diffusivity decreased
with the further increase of load level. Zhou et al. [68] investigated the oxidation kinetics and
the growth of oxide scales on pure nickel oxidized under tensile and compressive stress at
973 K. The oxidation kinetics curves gained by thermogravimetric analysis (TGA) indicate that
the oxidation rate of pure nickel was accelerated by the external stress. Compared with the
tensile stress, the effect of compressive stress on the oxidation kinetics was more
pronounced. Surface morphologies examined by scanning electron microscopy (SEM) reveal
that granular oxides formed on the stressed specimen, which increase the short-circuit
diffusion the oxidation [68]. A study of cast nickel superalloy IN100 by Reger and Remy [69]
showed that matrix oxidation was strongly enhanced by fatigue cycling.
Effects of applied tensile stress on the growth kinetics of oxide scales has also been studied
for Ni-20Cr alloys at 600°C and 900°C to understand the simultaneous effect of the
environment and mechanical loading [70]. Applying a tensile load did not modify the oxide
layer nature but oxidation rate was increased due to the increase of oxygen diffusion via fast
diffusion paths generated when the strain was higher than a critical value. Consequently, the
internal oxidation was promoted and the oxide layers were shown to be much thicker than
that formed without any load. At 900°C, for short oxidation time, the spinel oxide layer
(NiCr2O4) was enlarged; while for longer time, the spinel layer progressively disappeared and
37
the excess of oxygen diffusing inwards reacted with Cr to form a thicker Cr2O3 layer.
Mechanical stress was also found to promote internal nitridation of nickel alloy Inconel 718
[71].
2.5. OXIDATION-ASSISTED CRACK GROWTH
2.5.1. TIME-DEPENDENT CRACK GROWTH DUE TO OXIDATION
The service experience of gas turbines involves low cyclic frequencies, often with long hold
times at maximum load (dwells) and temperature, in an aggressive environment. Under such
conditions cyclic crack growth rate is greatly accelerated and the crack length is a function of
both cycle count and period as shown in various studies [21, 40, 41, 46, 61, 72-74]. The
processes that contribute to time dependent crack propagation are thermally activated and
therefore become increasingly important with increasing temperature [73] as illustrated in
Figure 2.3 for crack growth in IN718 [74]. At 400°C the alloy shows pure cycle dependent
crack growth at all frequencies, and air tests give the same propagation rates as vacuum
tests. But with increasing temperature and cycle period, the propagation rates in air become
more time dependent.
In early work, the time dependence was assumed to stem from the nature of creep
deformation at the crack tip and the term creep-fatigue was applied [75, 76]. However, for
nickel-based superalloys, it was demonstrated that the test environment in fact played the
dominant role [77-79], especially oxygen in air [80]. Many studies showed that nickel alloys
are sensitive to oxidation-assisted crack growth mechanisms [21, 39, 40, 41, 46, 61, 72-74,
38
81-87], as shown in Figure 2.4. The loading frequency as well as microstructural effects
appear only in air, and disappear under vacuum [86].
Figure 2. 3: Fatigue crack propagation behaviour in IN718 at ΔK=27.5MPa√m *74+.
Figure 2. 4: Fatigue crack growth rate in air and vacuum at 650°C [87].
39
Experiments show that the process of oxygen embrittlement during fatigue crack
propagation occurs at the crack tip. The role of oxygen is simply to reduce the grain boundary
energy in such a way that the crack growth path becomes intergranular [20, 21, 40, 41, 49,
61, 86-89], as shown in Figure 2.5. Miller et al. [46] concluded that environmentally
enhanced crack growth in IN718 occurs by the stress-enhanced oxygen diffusion ahead of the
crack tip, followed by the oxidation of Ti, Al, Cr and Nb if present at the grain boundaries. The
oxidation of all these elements was considered to affect the environmentally enhanced crack
growth. As mentioned in Section 2.2, oxidation mechanisms can be identified in terms of
oxygen short - and long- range diffusion processes [49, 88].
Figure 2. 5: (A) Intergranular and (B) Transgranular fracture surfaces for crack growth in air
and vacuum for Alloy 718 at 650°C [87].
2.5.2. CRACK TIP EMBRITTLEMENT MECHANISM—SHORT RANGE DIFFUSION
Short-range diffusion refers to the formation of an oxide layer at the crack tip with a depth
that depends on cycle and material parameters. The formation of this layer may cause high
stresses to occur and transmit to the substrate. This mechanism could also result in the
40
formation of wedge-shaped oxide intrusions along the crack front. When these wedges
rupture at the intersection of grain boundaries, accelerated intergranular crack growth may
be observed.
Figure 2. 6: Fatigue crack growth rate as function of oxygen partial pressure [90].
Oxygen partial pressure is an important factor associated with short-range oxygen diffusion
at the crack tip. Molins et al. work [90] showed a sharp transition in fatigue crack growth rate
at an oxygen pressure of 10-1Pa, as can be seen from Figure 2.6. Crack propagation mode is
transgranular with low growth rate when oxygen pressure is below 10-1Pa and becomes
intergranular with accelerated growth rate when oxygen pressure is over 10-1Pa. The results
of Andrieu et al. [91] also confirmed the strong pressure dependence for crack growth.
Between pressures of about 10-1Torr and 10-3Torr, the crack propagation rate dropped by
three orders of magnitude. This is because the pressure of oxygen controls the preferential
formation of particular species of oxides which may either 'shield or contribute' to the crack-
41
tip damage process through the passivation or the enhancement of the oxidation process,
respectively [88].
In particular, Andrieu et al. [49] conducted tests and analyses of oxide formation on
deformed and non-deformed free surfaces of alloy IN718 as a function of O2 partial pressure.
They suggested that the oxidation process starts with the formation of selective oxides of
FeO and NiO and their spinels, once the atmospheric O2 partial pressure is available at the
crack tip. These oxides are porous and thus permit the oxidation process to continue into the
base material. However, over a period of time (equal to or less than the oxide passivation
time), the O2 partial pressure tends to decrease at the oxide-substrate interface, which
provides the required reaction conditions for the formation of a less porous, denser sub-layer
of Cr2O3. This chromium layer, which is known to be thermodynamically stable at high
temperatures, limits the diffusion of oxygen and acts to protect the grain boundary material
from further oxidation. Fatigue crack growth experiments conducted by Ghonem [88]
showed that the delay in the formation of the protective Cr2O3 layer resulted in a higher
crack growth rate.
2.5.3. CRACK TIP EMBRITTLEMENT MECHANISM—LONG RANGE DIFFUSION
Long-range diffusion refers to the penetration of oxygen into the crack-tip material along
rapid diffusion paths, for example slip planes [88] and grain boundaries. This process is
characterised by internal oxide sites, cavity formations and/or solute segregation. Woodford
and Bricknell [80] showed that oxygen could also take part in chemical reactions, thus
releasing embrittlement agents onto a grain boundary. These processes prevent the sliding
and migration of boundaries, and therefore their ability to relieve the build-up of local
42
stresses during deformation is reduced and intergranular cracking is promoted. Either or
both of these oxidation processes could be operative at the same time under any particular
set of conditions. As discussed by Ghonem [88], it is acknowledged that oxidation
mechanisms due to short and long-range diffusion are not completely separable.
2.5.4. INTERACTION BETWEEN CRACK TIP FIELD AND OXIDATION
It is extremely difficult to study micro structurally the interaction of oxygen with the tip of a
propagating crack because of the small scale penetration involved at temperatures of
interest (i.e., 400°C-650°C). This process could occur over distances of just a few atom
spacing in the limit, and also the near-tip stress levels may become uncertain as the material
creeps. Nevertheless, there have been several attempts to examine the oxidation near crack
tip region in detail. For instance, Wei and co-workers [92] have examined the region ahead of
sustained load cracks and showed a correlation between environmental sensitivity and
niobium content for several nickel alloys. However, significant oxygen enhanced cracking was
also seen in a niobium free alloy in their later work [93]. Thus, it remains unclear the role that
niobium plays in the oxidation-embrittlement mechanism near the crack tip. It should be
noted that their work evidenced internal oxidation ahead of the crack tip during oxygen-
enhanced crack growth, with the depth of internal oxides determined using X-ray
photoelectron spectroscopy (XPS) [46, 94 and 95].
Both strain rate and stress and strain fields [18, 19 and 61] within the plastic zone ahead of
the crack tip play a significant role in the mechanism of oxygen-induced intergranular
fracture. Intergranular oxygen diffusion depends on the stress and strain states along the
affected grain boundaries [96-98]. Molins et al. [90] carried out experiments to study the
43
local interaction between oxidation and deformation rate at the crack tip, to identify the
damaging part in a mechanical cycle. It was found that the oxidation effects occurred only
between load ratios of R = 0.3 and R = 0.9, corresponding to the loading and un-loading
threshold. This suggests that the embrittlement process is a dynamic coupling between
oxidation and surface mechanical behaviour and controlled by crack-tip stress states.
Conditions that promote rapid stress relaxation near the crack tip significantly reduce the
extent of environmental damage and result in a decrease in the crack growth rate [65].
Studies have also been attempted to define an effective time tox for the damaging effect of
oxygen penetration to occur in a fatigue cycle, as the loading and unloading parts may play
different roles in the oxygen diffusion process. Molins et al. [90] studied the oxidation
assisted fatigue crack growth in nickel based superalloys by varying oxygen partial pressure
during a loading cycle. They found that oxidation effects occur primarily during the loading
part of the cycle and part (about 10%) of the unloading cycle when the crack tip is in traction.
Consequently, an oxidation sensitive time was proposed by Tong et al. [99] and correlated
reasonably with crack growth data, where only 10% of the unloading duration was used.
Other work proposed that it was half of the tensile part of the cycle [100] or all the loading
part [101]. On the contrary, an experimental attempt to measure tox concluded that, for the
same frequency, varying the ratio of loading and unloading portions of the cycle did not
influence the fatigue crack growth rate of IN718 at 650°C [87]. It was therefore concluded
that, for this alloy under these conditions, the cycle effective oxidation time was equal to the
total time of the cycle. Using this observation, and a derived relationship between K and the
intergranular oxygen diffusion rate, estimates were made of crack growth rates for several
44
test frequencies. Although the comparison is quite good, the range of test is limited and
dwell tests were not included.
2.5.5. CRACK GROWTH PREDICTION MODELS
Life prediction models were proposed through literature on how to characterise crack
growth by introducing an oxidation parameter, which have been mainly based on
phenomenological approaches rather than analytical or theoretical methods. The simplest
model for the effect of oxidation on fatigue involves the addition of an oxidation contribution
to a mechanical increment of crack growth:
where
is the total crack growth rate,
and
are the crack growth rates
contributed by mechanical deformation and oxidation, respectively.
King and Coterill [102] gave the crack growth as a result of oxidation as the time per cycle
multiplied by the average oxidation rate P:
where f is the loading frequency and P the oxidation rate1.
1 The oxidation rate P refers to the growth rate of oxide scale thickness, and can be expressed as P =
, where X
is oxide scale thickness and t is the time. The frequency f is defined as
, where T is the time for one fatigue
cycle.
45
Reuchet and Remy [103] have developed an equation that includes the oxidation in the
fatigue crack growth, which was expressed as:
where is the material constant for the oxidation constant of the material matrix and t the
cycle time.
The oxidation activation energy Q has also been used by many authors to quantify the effect
of oxidation in crack growth. For instance, Gao et al. [104] used Arrhenius relationship to
describe time-dependent crack growth response:
where and m are fatigue crack growth constants, K is the stress intensity factor, R the
universal gas constant and T the absolute temperature. Miller et al. [105] suggested that if
the crack is part of the oxidised material then the crack propagation should be directly
related to the cycle time of oxygen penetration:
where , m and ξ are constants, t is the cycle time and J is the J integral (see Appendix B).
Ghonem and Zheng [87] proposed that, in the environmental-dependent stage, crack growth
increment per cycle is equal to the intergranular oxygen diffusion depth X occurring during
the cycle effective oxidation time tox and expressed the fatigue crack greater as:
46
Where is the rate of intergranular oxygen diffusion depth at the crack tip, as a function of
stress intensity factor range K and time t.
Another interesting model [106] is shown schematically in Figure 2.7. For a given material at
a given K level, there exists a corresponding fatigue crack increment per cycle af which is
caused solely by the fatigue process without environmental effects. For an effective cycle
time period t the damaging species must move with the crack tip to be detrimental. During
this time the intergranular penetration distance is ap which may be increased by increasing
the temperature or reducing the frequency. If af is much greater than ap crack growth is
caused primarily by the fatigue process alone. In this case, da/dN is frequency independent
and given by the limit line of no environmental effect as shown in Figure 2.7. If ap is greater
than or equal to af the environment has its full effect on da/dN. Any further increase in ap
by increasing the temperature or by reducing the frequency will not increase da/dN. This
situation is thus depicted by the limit line of full environmental effect as shown in Figure 2.7.
The crack growth mode changes from transgranular without environmental effects to fully
intergranular with environmental effects. In the region between these two limits the rate
controlling environmental effect becomes important and da/dN is temperature and
frequency dependent. According to some crack growth experimental results on IN100 at
650C, the general applicability of the model appeared very good [106].
47
Figure 2. 7: A schematic of the kinetic model [106].
2.6. DYNAMIC EMBRITTLEMENT MECHANISM
Dynamic embrittlement was introduced to define a time-dependent intergranular brittle
cracking process in which the supply of embrittling elements is due to grain boundary
diffusion into an area of increasing tensile stress ahead of a crack tip [107-112]. The
embrittling element lowers the cohesion of grain boundaries, allowing for crack advance and
shifting crack tip, the stress and the diffusion gradient onward. The case of hydrogen
embrittlement of steels is one of the most important damage mechanisms that fit in the
dynamic embrittlement scheme [113]. Other established dynamic embrittlement cases
include sulphur-induced stress relief cracking in steels [112], S-induced cracking in Cu Cr
48
alloys [114], O-induced cracking in Cu Be alloys [115] and Ni3Al [111], and Sn-induced
cracking in Cu-Sn model alloys [112, 116].
The dynamic embrittlement has also been used to explain the oxidation-accelerated crack
growth in nickel alloy by Pfaendtner and McMahon [107], Krupp et al. [108] and Krupp [109]
in their studies of the intergranular crack propagation in Inconel 718 under four-point
bending. Firstly, the individual grain boundary facets were found to be smooth on the sub-
micrometer level, and oxidation of NbC precipitates was only observed on the fractured
surface where the crack front had passed. Secondly, tests found that the cracking process is
extremely sensitive to rapid changes under oxygen pressure. Crack growth could be stopped
in less than 10s, by alternately pumping oxygen from the chamber and re-admitting it, as
shown in Figure 2.8. Given the fact that oxidation kinetics for nickel alloys at 650°C is
extremely slow [117], there is insufficient time for oxide formation to happen ahead of the
crack tip within the order of seconds. Direct observations of internal oxidation and associated
damage were only available on smooth specimens or fractured surfaces, which have been
exposed to oxidizing environment for a sufficiently long time, mostly greater than 20 hours
[47, 48, 58, 70]. This mechanism seems to suggest that oxygen elements diffuse into the
crack tip along grain boundaries and attack the grain boundaries by themselves, which is
different from the internal oxide formation mechanism as discussed in Section 2.5. Here,
high-level stresses at the crack tip are regarded as the driving force to accelerate the
diffusion of embrittling elements into the crack tip and promote crack propagation by
weakening grain boundary cohesion, i.e., a dynamic process of embrittlement. It was further
shown by Krupp [109] that the character of grain boundaries plays a role in oxidation-
49
assisted crack growth of nickel alloy IN718. For instance, special grain boundaries with a high
fraction of coincident lattice sites exhibit a higher resistance to oxygen attack and reduced
crack growth rate, due to their low oxygen diffusivity, when compared to random high-angle
grain boundaries [109, 118].
Figure 2. 8: At 650°C, cracking in 1atm of oxygen was completely arrested in about 10s, the
time necessary to evacuate the testing chamber below 10-2Pa. Cracking resumed when
oxygen was re-admitted to the chamber [107].
An early model for dynamic embrittlement was developed analytically by Bika and McMahon
[110] to describe quantitatively the experimental results for S-induced stress relief cracking
of steel. They used diffusion differential equations for grain boundary diffusion under the
influence of stress, obtained using a fracture mechanics expression, ahead of the crack tip.
This allowed for a numerical solution that yields the concentration profiles of the embrittling
element ahead of the crack tip. An empirical fracture criterion, which provided a functional
50
connection between the embrittler concentration and the local critical tensile strength, was
proposed to predict the onset location and time of intergranular decohesion. Xu and Bassani
[119] developed a cohesive model for diffusion-controlled fracture. The constitutive
equations for the cohesive zone are coupled with stress-assisted diffusion of impurities into
the grain boundaries. A damage model was used to describe the effect of impurities
concentration on grain boundary strength and fracture.
Computational simulation of oxygen diffusion [108, 109, 120, 121] has been attempted based
on Fick’s first and second laws, where the effect of stress on oxygen diffusion is assumed to
be controlled by the gradient of hydrostatic stress [110, 122-124]. In this case, diffusion of
oxygen within the material is controlled by two essential parameters, namely, the oxygen
diffusivity and the pressure factor [108, 109, 120 and 121]. This approach has been applied to
study the oxygen embrittlement of engineering materials under mechanical loading
conditions, especially for prediction of crack growth under fatigue-oxidation conditions. For
instance, Carranza and Haber [120] presented a numerical investigation, based on finite
element calculation and cohesive interface model, of high temperature intergranular fracture
in nickel base superalloys subjected to oxygen embrittlement. The formulation includes
coupled models for the mechanical response, stress-assisted diffusion of oxygen and a
moving cohesive interface that determines crack initiation, growth and arrest. The cohesive
strength was expressed as a function of the concentration and both the strength and the
work of separation of the cohesive interface are reduced by the presence of oxygen. Zhao et
al. [121] carried out finite element analyses of oxygen diffusion, coupled with viscoplastic
deformation, near a fatigue crack tip for a nickel-based superalloy at elevated temperature.
51
Low frequencies and superimposed hold periods at peak loads significantly enhanced oxygen
concentration near the crack tip. Furthermore, a failure envelop for crack growth was
constructed based on the consideration of both oxygen concentration and accumulated
inelastic strain near the crack tip. Using the fatigue-oxidation failure envelop, crack growth
rates in a CT specimen were predicted for selected loading ranges, frequencies and dwell
periods. The predictions from the fatigue-oxidation failure envelop compared well with the
experimental results, with significant improvements achieved over those predicted from the
viscoplastic model alone [121].
It should be noted that, so far, simulation of stress-driven oxygen diffusion near a crack tip is
largely limited to the continuum level, where the material is treated as homogeneous
medium and no micro-structural details have been considered. In fact, for polycrystalline
nickel and its alloys, oxidation and the associated diffusion of oxygen are primarily along the
grain boundaries [47, 56, 80 and 118], therefore consideration of grain microstructure is
necessary in order to simulate the oxygen diffusion process properly.
2.7. REVIEW OF RESEARCH ON ALLOY RR1000
Alloy RR1000, also the material studied in this work, is one of the latest fine-grained nickel-
based superalloys, developed at Rolls-Royce plc of the UK through a powder metallurgy
process, to meet the increasing demand of mechanical performance for turbine discs in the
latest Trent 1000 Rolls-Royce aero engines. This section provides a review of the research
work carried out on this particular alloy.
52
2.7.1. CRACK GROWTH STUDY
Fatigue and creep behaviour of alloy RR1000 at elevated temperature have been
systematically studied through crack growth testing using compact tension (CT) and corner
notch (CN) specimens. Tong et al. [84] showed that mixed time and cycle dependent crack
growth seems to be the dominant crack growth mode for RR1000 at 650C. Knowles and
Hunt [85] investigated the influence of microstructure and environment on crack growth
behaviour of RR1000 at 725C. Both environment and exposure-induced sigma-phase
precipitation at grain boundaries were shown to increase the crack growth rate through
increased crack tip cavity nucleation, whilst rapid near-tip stress relaxation induced by
coarsening has a beneficial effect on the severity of crack tip cavity damage. For short crack
growth at elevated temperature, Pang and Reed [125] showed that RR1000 demonstrated
superior fatigue performance compared to Udimet 720 and was attributed to the benefit of
stress relaxation due to its alloy chemistry. Most recently, Dalby and Tong [30, 31, 99]
systematically studied the crack growth behaviour for RR1000 at elevated temperature by
considering a range of loading waveforms and frequencies. Accelerated crack growth rates
were observed in specimens tested at low frequencies, long dwell periods or under slow-fast
loading waveform, due to oxidation effects [99].
2.7.2. CYCLIC CONSTITUTIVE BEHAVIOUR
Comprehensive experimental and analytical studies [15-17] have also been carried out
towards a fundamental understanding of time-dependent deformation behaviour using
material constitutive models. Strain-controlled cyclic tests were carried out at 650C for
selected strain ratio, range and rate. A unified Chaboche viscoplastic model [126], with a
53
combined non-linear kinematic and isotropic hardening, was adopted and modified to
describe the time-dependent constitutive behaviour of RR1000 alloy. Model parameters
were determined from an optimisation of the basic monotonic, cyclic and creep experimental
data [15-17]. Model simulations compared very well with the experimental results for a
range of strain-controlled loading cases in terms of stress-strain loops, cyclic hardening
behaviour and stress relaxation behaviour. The cyclic behaviour of the material was also
linked with the evolution of dislocation patterns using TEM observations [127]. This model
was further enhanced by Cornet et al. [128-130] to simulate strain ratchetting and stress
softening by introducing a third kinematic hardening component and a damage variable,
respectively.
Very recently, crystal plasticity model has been developed by Lin et al. [131, 132] to simulate
the cyclic deformation of alloy RR1000 at elevated temperature using finite element and
representative volume element approach. The model simulated well the stress-strain loops,
cyclic hardening behaviour and stress relaxation behaviour, as shown in Figure 2.9. Stress and
strain distributions within the representative volume element were shown to be of
heterogeneous nature due to the orientation mismatch between neighbouring grains, as
shown in Figure 2.10 for the von Mises equivalent strain.
54
Figure 2. 9: Comparison of model simulations and experimental results: (a). Stress-strain
loop at saturation (dε/dt=0.05%/s and Δε=2%) (b). Stress relaxation behaviour during 100-
second dwells at maximum and minimum strain levels (dε/dt=0.5%/s and Δε=2%) *132+.
Figure 2. 10: Contour plot of the von Mises equivalent strain at the maximum strain level of
the 2nd cycle (strain rate dε/dt=0.005%/s, strain range Δε=20%, strain ratio =-1) [132].
(a) (b)
55
2.7.3. MODELLING OF CRACK TIP DEFORMATION AND GROWTH
All the constitutive models, as described above, have been implemented in the finite element
software ABAQUS via a user-defined material subroutine (UMAT), and applied for modelling
of crack tip deformation and crack growth.
Figure 2. 11: (a). The evolution of stress-strain loops over 30 cycles ahead of a stationary
crack tip. (b). The evolution of stress-strain loops registered at an integration point O as the
crack grew towards it [133].
Using the viscoplasticity model, finite element analyses for stationary cracks showed
distinctive strain ratchetting behaviour near the crack tip, leading to progressive
accumulation of tensile strain normal to the crack growth plane [130, 133] (Figure 2.11(a)).
This is also true for time-independent cyclic loading cases [134]. Low frequencies and
superimposed hold periods at peak loads significantly enhanced strain accumulation at crack
tip. Finite element simulation of crack growth was carried out under a constant K-controlled
loading condition, again ratchetting was observed ahead of the crack tip (Figure 2.11(b)),
similar to that for stationary cracks. Further on, a crack growth criterion based on strain-
56
accumulation is proposed where a crack is assumed to grow when the accumulated strain
ahead of the crack tip reaches a critical value over a characteristic distance. The criterion has
been utilised in the prediction of crack growth rates in a CT and CN specimens at selected
loading ranges, frequencies and dwell periods [130, 133], and the predictions were compared
with the experimental results.
In addition, the viscoplasticity-damage model was utilised to investigate the damage
evolution near a crack tip for a Single Edge Notch Tension specimen (SENT), where the
damage accumulation rate with respect to cycles was found to be linear once an inelastic
strain threshold was reached [128, 129]. The damage development seems to be localized to
the crack tip and becomes more significant at lower frequencies and longer dwell periods
[128, 129]. While a crystal plasticity model was applied to study the near-tip deformation of a
transgranular crack, in a compact tension specimen using a submodelling technique [132].
Grain microstructure is shown to have an influence on the von Mises stress distribution near
the crack tip and the grain texture heterogeneity disturbs the well-known butterfly shape
obtained from the viscoplasticity analysis at continuum level. The stress-strain response near
the crack tip, as well as the accumulated shear deformation along slip system, is influenced
by the orientation of the grain at the crack tip, which might dictate the subsequent crack
growth through grains. Individual slip systems near the crack tip tend to have different
amounts of accumulated shear deformation, which was utilised as a criterion to predict the
zigzag crack growth path [132]. Moreover, simulation of creep crack growth in RR1000
showed the creep-brittle nature due to its limited creep deformation [135, 136].
57
2.7.4. OXIDATION WORK
Very recently, oxidation damage of alloy RR1000 has been investigated by Encinas-Oropesa
et al. [48] at temperatures between 700C and 800C. Oxidation kinetics of RR1000 has been
characterized by mass change data from thermogravimetric analyses of small disc samples at
temperatures from 700C to 800C, where the oxide scale rich in Cr and Ti, confirmed from
EDX analyses, has been found to grow in a parabolic dependence with time. On the other
hand, FIB examinations of oxidised samples for RR1000 revealed the appearance of the oxide
scale on the surface and the formation of recrystallised grains and voids beneath the oxide
scales [48], as shown in Figure 2.12. The voids are considered to be the result of the loss of
base alloy elements (e.g., Cr, Ti Ni), which diffuse outwards to the specimen surface to form
the oxide scale. Also voids are mainly located along the grain boundaries which are the
preferential diffusion paths. The oxidation-linked microstructure damage becomes more
significant with the increase of testing temperature. The thickness of oxide scale and the
depth of internal damage, exemplified as voids along grain boundaries, are more than tripled
at 800C when compared to those at 700C. In addition, oxide particles can also be observed
on the surface of those internal micro-voids from the FIB images [48].
58
Figure 2. 12: Focused ion beam image of an oxidised RR1000 surface, exposed to 700°C for
200 hours [48].
Oxidation was also observed in fatigue crack growth by Tong et al. [99] and Figure 2.13 shows
the oxide layers formed on a fracture surface of RR1000 CT specimen during dwell fatigue
tests at 725C [99]. The oxide layer was considered to consist of outer nickel oxides and a
denser chromia layer underneath, and the chromia layer indicates a stage of oxidation
saturation on the alloy. Microscopy examination of crack tip regions in Knowles and Hunt
[85] also revealed oxidation near the crack tip, extending continuously along the grain
boundaries. No evidence of discrete damage was observed, indicating that the operating
environmental damage mechanism is highly localised at the crack-tip region.
59
Figure 2. 13: Oxide layer on the fracture surface during dwell fatigue tests at 725°C [98].
Modelling of oxidation for RR1000 has been first attempted by Zhao [137] using the oxygen
diffusion approach and finite element method, with and without mechanical loading. Grain
microstructures were considered explicitly in the finite element model, where the grain
boundary was taken as the primary path for oxygen diffusion. The diffusion of oxygen,
coupled with crystal plasticity deformation, was controlled by the parabolic oxidation rate,
oxygen diffusivity and pressure factor. It was found that heterogeneous deformation
presented at grain level imposes a great influence on oxygen diffusion at 750C and above,
leading to further penetration of oxygen into the bulk material. In the case of an existing
surface microcrack, oxygen tends to accumulate around the crack tip due to the high stress
level presented near the crack tip, which promotes localized material embrittlement and
rapid crack propagation. However, in Zhao's work [137], no attempt was made to model
60
chemical reaction kinetics, vacancy formation, the transport of other species and effect of
oxygen partial pressure, mechanisms which might also be operative in RR1000 [48].
2.8. CONCLUSIONS
The literature review shows that a large amount of work has been carried out to study the
oxidation damage and oxidation-assisted crack growth in nickel and its alloys. In particular,
the oxygen diffusion approach has been proposed and used to model crack growth under
fatigue-oxidation conditions, and the results are very encouraging. For nickel alloy RR1000,
although oxidation kinetics studies have also been carried out, the influence of fatigue was
not considered. Also, crack growth predictions, based on viscoplastic deformation alone,
showed a great discrepancy with experimental data for low frequencies and long dwell
periods, which was due to the neglect of oxidation effects at high temperature. Existing life
prediction models for fatigue-oxidation crack growth are still highly empirical. These research
gaps led to this PhD programme in which a combined experimental and computational study
has been carried out to investigate the effect of fatigue on oxidation damage and model the
oxidation-assisted crack growth in alloy RR1000.
61
2.9. REFERENCES
[1]. King, J.E. (1983). Fatigue crack propagation in nickel-base superalloys—effects of
microstructure, load ratio, and temperature. Materials Science and Technology, 3(9), 750-
764.
[2]. Abdul-Latif, A. (1996). Constitutive equations for cyclic plasticity of Waspaloy.
International Journal of Plasticity 12(8), 967–985.
[3]. Arrell, D.J., Bressers, J., Timm, J. (2000). Effect of cycle history on thermomechanical
fatigue response of nickel-base superalloys. ASME Journal of Engineering Materials and
Technology, 122(3), 300-305.
[4]. Zhao, L.G., Tong, J., Vermeulen, B., Byrne, J. (2001). On the uniaxial mechanical
behaviour of an advanced nickel base superalloy at high temperature. Mechanics of
Materials, 33(10), 593-600.
[5]. Manoukul, A., Dunne, F.P.E., Knowles, D. (2002). Physically based model for creep in
nickel-base superalloy C263 both above and below the gamma solvus. Acta Materialia,
37(11), 1067-1077.
[6]. Manoukul, A., Dunne, F.P.E., Knowles, D., Williams, S. (2005). Multiaxial creep and cyclic
plasticity in nickel-base superalloy C263. International Journal of Plasticity, 21(1), 1-20.
[7]. Chaboche, J.L., Gailetaud, G. (1986). On the calculation of structures in cyclic plasticity or
viscoplasticity. Computers & Structures, 23(1), 23-31.
[8]. Chaboche, J.L., Nouailhas, D., Pacou, D., Paulmier, P. (1991). Modeling of the cyclic
response and ratchetting effects on inconel-718 alloy. European Journal of Mechanics.
A.Solids, 10(1), 101-121.
[9]. James, G.H., Imbrie, P.K., Hill, P.S., Allen, D.H., Haisler, W.E. (1987). An experimental
comparison of several current viscoplastic constitutive models at elevated temperature. ASME
Journal of Engineering Materials and Technology, 109(2), 130-139.
[10]. Ramaswamy, V.G., Stouffer, D.C., Laflen, J.H. (1990). A unified constitutive model for
the inelastic uniaxial response of Rene’80 at temperatures between 538°C and 982°C. ASME
Journal of Engineering Materials and Technology, 112(3), 280-286.
[11]. Yaguchi, M., Yamamoto, M., Ogata, T. (2002). A viscoplastic constitutive model for
nickel-base superalloy. part 1: kinematic hardening rule of anisotropic dynamic recovery.
International Journal of Plasticity, 18(8), 1083-1109.
62
[12]. Aktaa, J., Schinke, B. (1996). Unified modelling of time dependent damage taking into
account an explicit dependency on back stress. International Journal of Fatigue, 19(3), 195-
200.
[13]. Tong, J., Vermeulen, B. (2003). The description of cyclic plasticity and viscoplasticity of
waspaloy using unified constitutive equations. International Journal of Fatigue, 25(5), 413-
420.
[14]. Tong, J., Zhan, Z.L., Vermeulen, B. (2004). Modelling of cyclic plasticity and
viscoplasticity of a nickel-based alloy using Chaboche constitutive equations. International
Journal of Fatigue, 26(8), 829-837.
[15]. Zhan, Z.L. (2004). A study of cyclic plasticity and viscoplasticity in a new nickel-based
superalloy using unified constitutive equations. PhD Thesis. University of Portsmouth.
Portsmouth, U.K.
[16]. Zhan, Z.L., Tong J. (2007). A study of cyclic plasticity and viscoplasticity in a new nickel-
based superalloy using unified constitutive equations. Part I: Evaluation and determination of
material parameters. Mechanics of Materials, 39(1), 64-72.
[17]. Zhan, Z.L., Tong J. (2007). A study of cyclic plasticity and viscoplasticity in a new nickel-
based superalloy using unified constitutive equations. Part II: simulation of cyclic stress
relaxation. Mechanics of Materials, 39(1), 73-80.
[18]. Clavel M, Pineau A. (1978). Frequency and wave-form effects on the fatigue crack
growth behavior of alloy 718 at 298 K and 823 K. Metallurgical and Materials Transactions A,
9(4), 471-480.
[19]. Andersson H, Persson C. (2001). Crack growth in IN718 at high temperature.
International Journal of Fatigue, 23(9), 817-827.
[20]. Osinkolu GA, Onofrio G. (2003). Fatigue crack growth in polycrystalline IN 718
superalloy. Materials Science and Engineering: A, 356(1-2), 425-433.
[21]. Smith, H.H., Michel, D.J. (1986). Effect of environment on fatigue crack propagation
behavior of alloy 718 at elevated temperatures. Metallurgical and Materials Transactions A,
17(2), 370-374.
[22]. James, L.A. (1981). Effect of product form upon fatigue-crack growth behavior in alloy
718. ASME Journal of Engineering Materials and Technology, 103(3), 234-240.
63
[23]. Connolley, T., Reed, P.A.S., Starink, M.J. (2003). Short crack initiation and growth at
600 C in notched specimens of Inconel718. Materials Science and Engineering: A, 340(1-2),
139-154.
[24]. Clavel, M., Pineau, A. (1982). Fatigue behaviour of two nickel-base alloys I:
Experimental results on low cycle fatigue, fatigue crack propagation and substructures.
Materials Science and Engineering, 55(2), 157-171.
[25]. Yuen, J.L., Roy, P. (1985). Effect of grain size on the near threshold fatigue crack
propagation of a nickel base precipitation hardened superalloy. Scripta Metallurgica, 19(1),
17-22.
[26]. Shahinian, P., Sadananda, K. (1979). Effects of stress ratio and hold-time on fatigue
crack growth in alloy 718. Journal of Engineering Materials and Technology, 101(3), 224-230.
[27]. Onofrio, G., Osinkolu, G.A., Marchionni, M. (2001). Fatigue crack growth of UDIMET
720 Li superalloy at elevated temperature. International Journal of Fatigue, 23(10), 887-895.
[28]. Tong, J., Byrne, J. (1999). Effects of frequency on fatigue crack growth at elevated
temperature. Fatigue & Fracture of Engineering Materials & Structures, 22(3), 185-193.
[29]. Fournier, D., Pineau, A. (1977). Low cycle fatigue behavior of inconel 718 at 298 K and
823 K. Metallurgical and Materials Transactions A, 8(7), 1095-1105.
[30]. Dalby, S. (2002). The effects of frequency and waveshape on the fatigue crack growth of
an advanced nickel based superalloy at elevated temperatures. PhD Thesis. University of
Portsmouth, Portsmouth, UK.
[31]. Dalby, S., Tong, J. (2005). Crack growth in a new nickel-based superalloy at elevated
temperature, Part I: Effects of loading waveform and frequency on crack growth. Journal of
Materials Science, 40(5), 1217-1228.
[32]. Andersson, H., Persson, C. (2004). In-situ SEM study of fatigue crack growth behaviour
in IN718. International Journal of Fatigue, 26(3), 211-219.
[33]. Stevens, R.N., Dutton, R. (1971). The propagation of Griffith cracks at high
temperatures by mass transport processes. Materials Science and Engineering, 8(4), 220-234.
[34]. Fu, L.S. (1980). Creep crack growth in technical alloys at elevated temperature—a
review. Engineering Fracture Mechanics, 13(2), 307-330.
[35]. Floreen, S. (1975). The creep fracture of wrought nickel-based alloys by a fracture
mechanics approach. Metallurgical and Materials Transactions A, 6(9), 1741-1749.
64
[36]. Wei, R.P. (2002). Influence of dwell time on fatigue crack growth in nickel-base
superalloys. Materials Science and Engineering A, 336(1-2), 209-214.
[37]. Nicholas, T., Weerasooriya, T. (1986). Hold-time effects in elevated temperature fatigue
crack propagation. In: Fracture mechanics, vol. 17, Am. Soc. Test. Mater., Spec. Tech. Publ.,
905, 155-168.
[38]. James, L.A. (1986). The effect of grain size upon the fatigue-crack propagation behavior
of alloy 718 under hold-time cycling at elevated temperature. Engineering Fracture
Mechanics, 25(3), 305-314.
[39]. Ghonem, H., Nicholas, T., Pineau, A. (1993). Elevated temperature fatigue crack growth
in alloy 718—Part I: Effects of mechanical variables. Fatigue & Fracture of Engineering
Materials & Structures, 16(5), 565-576.
[40]. Ghonem, H., Nicholas, T., Pineau, A. (1993). Elevated temperature fatigue crack growth
in alloy 718—Part II: Effects of environmental and material variables. Fatigue & Fracture of
Engineering Materials & Structures, 16(6), 577-590.
[41]. Leo Prakash, D.G., Walsh, M.J., Maclachlan, D., and Korsunsky, A.M. (2009). Crack
growth micro-mechanisms in the IN718 alloy under the combined influence of fatigue, creep
and oxidation. International Journal of Fatigue, 31(11-12), 1966-1977.
[42]. Smithells, C. J. and Ransley, C. E. (1936). Proc. Roy. Soc. London Ser. A, vol. 155, p.195.
[43]. Giggins, C.S. and Pettit, F.S. (1971). Oxidation of Ni-Cr-Al alloys between 1000 and
1200°C. Journal of the Electrochemical Society, 118(11), 1782-1790.
[44]. Chen, J.H., Rogers, P.M. and Little, J.A. (1997). Oxidation behavior of several chromia-
forming commercial nickel-base superalloys. Oxidation of Metals, 47(5-6), 381-410.
[45]. Geng, L., Young-Sang, Na and Nho-Kwang, P. (2007). Oxidation behaviour of Alloy 718
at high temperature. Materials & Design, 28(3), 978-981.
[46]. Miller, C.F., Simmons, G.W. and Wei, R.P. (2003). Evidence for internal oxidation during
oxygen enhanced crack growth in P/M Ni-based superalloys. Scripta Materalia, 48(1), 103-
108.
[47]. Garat, V., Brucelle, O., Cloue, J.-M., Rebeyrolle, V., Monceau, D., Viguier, B. and
Andrieu, E. (2004). Comparing different methods to determine the intergranular oxidation
damage on a nickel based superalloy. Materials Science Forum, 461-464, 537-544.
65
[48]. Encinas-Oropesa, A., Drew, G.L., Hardy, M.C., Leggett, A.J., Nicholls, J.R., Simms, N.J.
(2008). Effects of oxidation and hot corrosion in a nickel disc alloy. Champion, Pennsylvania,
USA : Wiley, 2008. Superalloys 2008. 609-618.
[49]. Andrieu, E., Molins, R., Ghonem, H. and Pineau, A. (1992). Intergranular crack tip
oxidation mechanism in a nickel-based superalloy. Materials Science and Engineering: A,
154(1), 21-28.
[50]. Douglass, D.L. (1969). The effect of oxidation on the mechanical behavior of nickel at
600°C. Materials Science Engineering, 3(5), p.255-263.
[51]. Chang, W.H. (1972). Superalloys-processing. MCIC-72-10, AIME, Littleton, CO.
[52]. Woodford, D.A. (1981). Environmental damage of a cast nickel base superalloy.
Metallurgical and Materials Transactions A, 12(2), 299-308.
[53]. Antolovich, S.D., Baur, R., and Liu S. (1980). Low cycle fatigue behavior of René 80 at
elevated temperature. Metallurgical and Materials Transactions A, 12(3), 473-481.
[54]. Bricknell, R.H., Mulford, R.A. and Woodford, D.A. (1982). The role of sulfur in the air
embrittlement of nickel and its alloys. Metallurgical and Materials Transactions A, 13(7),
1223-1232
[55]. Woodford, D.A. and Bricknell, R.H. (1981). Grain boundary embrittlement of the iron-
base superalloy IN903A. Metallurgical and Materials Transactions A, 12(9), 1673-1680.
[56]. Hancock, P. and Fletcher, R. (1966). The oxidaiton of nickel in the temperature range
700-1000C. Metallurgie, vol. 6, p.1
[57]. Hales, R. and Hill, A.C. (1972). The role of metal lattice vacancies in the high
temperature oxidation of nickel. Corrosion Science, 12, 843-853.
[58]. Bricknell, R. H. and Woodford, D. A. (1982). The mechanism of cavity formation during
high temperature oxidation of nickel. Acta Metallurgica, 30(1), 257-264.
[59]. Bricknell, R. H. and Woodford, D. A. (1981). The embrittlement of nickel following high
temperature air exposure. Metallurgical and Materials Transactions A, 12(3), 425-433.
[60]. Mrowec, S., Grzesik, Z. and Rajchel, B. (2004). Oxidation of Nickel and Ni-Cr and Ni-Na
alloys at high temperatures. High Temperature Materials and Processes, 23(1), 59-71.
[61]. Fournier, L., Delafosse, D. and Magnin, T. (2001). Oxidation induced intergranular
cracking and Portevin-Le Chatelier effect in nickel base superalloy 718. Materials Science and
Engineering: A, 316(1), 166-173.
66
[62]. Reed, P.A.S. (2003). Microstructural influences on notch fatigue crack initiation and
propagation in a range of nickel base superalloys. Materials Science Forum, 426-432, 731-
736.
[63]. Calvarin-Amiri, G., Gourgues, A.F., Huntz, A.M. and Molins, R. (2001). Effect of
environment on the mechanical behaviour of Ni-20Cr thin strip. Materials Science and
Engineering: A, 298(1-2), 200-208.
[64]. Osgerby, S. and Dyson, B.F. (1990). Effects of oxygen on creep performance:
mechanisms and predictive modelling. Materials Science and Technology, 6(1), 2-8.
[65]. Moulin, G., Arevalo, P. and Salleo, A. (1996). Influence of external mechanical loadings
(creep, fatigue) on oxygen diffusion during nickel oxidation. Chemistry and Materials Science,
45(1-2), 153-181.
[66]. Berger, P., Moulin, G. And Viennot, M. (1997). Nuclear microprobe study of stress-
oxidation of nickel. Nuclear Instruments and Methods in Physics Research, Section B; Beam
interactions with materials and Atoms, 130(1-4), 717-721.
[67]. Berger, P., Gaillet, L., El Tahhann, R., Moulin, G. and Viennot, M. (2001). Oxygen
diffusion studies in oxide scales thermally grown or deposited on mechanically loaded
metallic surfaces (MS-P2). Nuclear Instruments and Methods in Physics Research, Section B;
Beam interactions with materials and Atoms, 181(1-4), 382-388.
[68]. Zhou, C.H., Ma, H.T. and Wang, L. (2010). Comparative study of oxidation kinetics for
pure nickel oxidized under tensile and compressive stress. Corrosion Science, 52(1), 210-215.
[69]. Reger, M., Remy, L. (1988). Fatigue oxidation interaction in IN 100 superalloy.
Metallurgical Transactions. A, Physical Metallurgy and Materials Science, 19A(9), 2259-2268.
[70]. Calvarin-Amiri, G., Huntz, A.M. and Molins, R. (2001). Effect of an applied stress on the
growth kinetics of oxide scales formed on Ni-20Cr alloys. Materials at High Temperatures,
18(2), 91-99.
[71]. Krupp, U. (2010). Nitridation of alloys. Cottis, R.A., Graham, M.J., Lindsay, R., Lyon, S.B.,
Richardson, J.A., Scantlebury, J.D., Stott, F.H., Shreir's Corrosion, (304-315). UK: Academic
Press-Elsevier.
[72]. Shahinian, P., and Sadananda, K. (1976). Crack growth behavior under creep-fatigue
conditions in alloy 718, creep fatigue interaction. NY: ASME, 365-390.
67
[73]. Solomon, H.D. and Coffin, L.F. (1973). Effects of frequency and environment on fatigue
crack growth in A286 at 1100 F. Fatigue at Elevated Temperatures. (112-122). Philadelphia:
American Society for Testing and Materials.
[74]. Chang, K.-M., Henry, M.F. and Benz, M.G. (1990). Metallurgical control of fatigue crack
propagation in superalloys, JOM, 42, 29-35.
[75]. Manson, S.S. (1973). Fatigue at Elevated Temperatures. ASTM STP 520. Philadelphia,
744-782.
[76]. Webster, G.A. and Ainsworth, R.A. (1994). High Temperature Component Life
Assessment. London: Chapman and Hall.
[77]. Woodford, D.A. and Coffin, L.F. Jr. (1974). Grain boundaries in engineering materials.
Proceedings of the Fourth Bolton Landing Conference. The role of grain boundaries in high
temperature fatigue (p. 432). Lake George, N.Y. USA. 9-12 June 1974.
[78]. Coffin, L.F. Jr. (1972). The effect of high vacuum on the low cycle fatigue law.
Metallurgical and Materials Transactions B, 3(7), 1777-1788.
[79]. McMahon, C.J. and Coffin, L.F. Jr. (1970). Mechanisms of damage and fracture in high-
temperature, low-cycle fatigue of a cast nickel-based superalloy. Metallurgical and Materials
Transactions B, 1(12), 3443-3450.
[80]. Woodford, D.A. and Bricknell, R.H. (1983). Environmental embrittlement of high
temperature alloys by oxygen. Treatise on Materials Science and Technology, 25, 157-199.
[81]. Floreen, S., and Kane, R.H. (1979). Effects of environment on high-temperature fatigue
crack growth in superalloy. Metallurgical and Materials Transactions A, 10(11), 1745-1751.
[82]. Shahinian, P. and Achter, M.R. (1959). A comparison of the creep–rupture properties of
nickel in air and in vacuum. Trans Metall Soc AIME, 215, 37–41.
[83]. Sadananda, K. and Shahinian, P. (1982). Creep crack growth behavior of several
structural alloys. Metallurgical and Materials Transactions A , 14(7), 1467-1480.
[84]. Tong, J., Dalby, S., Byrne, J., Henderson, M.B. and Hardy, M.C. (2001). Creep, fatigue
and oxidation in crack growth in advanced nickel base superalloys. International Journal of
Fatigue, 23(10), 897-902.
[85]. Knowles, D. and Hunt, D. (2002). The influence of microstructure and environment on
the crack growth behaviour of powder metallurgy nickel superalloy RR1000. Metallurgical and
Materials Transactions A, 33(10), 3165-3172.
68
[86]. Pedron, J.P. and Pineau, A. (1982). Effect of microstructure and environment on crack
growth behaviour in Inconel 718 alloy at 650°C under fatigue, creep and combined loading.
Materials Science and Engineering, 56(2), 143-156.
[87] Ghonem, H. and Zheng, D. (1992). Depth of intergranular oxygen diffusion during
environment-dependent fatigue crack growth in alloy 718. Materials Science and Engineering
A, 150(2), 151-160.
[88]. Ghonem, H., Nicholas, T. and Pineau, A. (1993). Elevated temperature fatigue crack
growth in alloy 718—Part II: Effects of environmental and material variables. Fatigue &
Fracture of Engineering Materials & Structures, 16(6), 577-590.
[89]. Woodford, D.A. (2006). Gas phase embrittlement and time dependent cracking of nickel
based superalloys. Energy Materials: Materials Science and Engineering for Energy Systems,
1(1), 59-79.
[90]. Molins, R., Hochstetter, G., Chassaigne, J.C. and Andrieu, E. (1997). Oxidation effects
on the fatigue crack growth behaviour of alloy 718 at high temperature. Acta Materialia,
45(2), 663-674.
[91]. Andrieu, E., Hochstetter, G., Molins, R. and Pineau, A. (1994). Loria, E.A. (Ed.).
Oxidation Mechanisms in Relation to High Temperature Fatigue Crack Growth Properties of
Alloy 718. Superalloys 718, 625, 706 and Various Derivatives, TMS, 619-631.
[92]. Gao, M., Dwyer, D.J. and Wei, R.P. (1994). Loria, E.A. (Ed.). Chemical and
Microstructural Aspects of Creep Crack Growth in Inconel 718 Alloy. Superalloys 718, 625, 706
and Various Derivatives, TMS, 581-592.
[93]. Huang, Z., Iwashita, C., Chou, I. and Wei, R.P. (2002). Environmentally assisted,
sustained-load crack growth in powder metallurgy nickel-based superalloys. Metallurgical and
Materials Transactions A, 33(6), 1681-1687.
[94]. Miller, C.F., Simmons, G.W. and Wei, Robert P. (2000). Mechanism for oxygen
enhanced crack growth in Inconel 718. Scripta Materalia 44(10), 2405-2410.
[95]. Wei, R.P., Miller, C., Huang, Z., Simmons, G.W. and Harlow, D.G. (2009). Oxygen
enhanced crack growth in nickel-based superalloys and material damage prognosis.
Engineering Fracture Mechanics, 76(5), 715-727.
[96]. Sadananda, K. and Shahinian, P. (1985). Effects of environment on high temperature
crack growth behavior of several nickel base alloys. Corrosion of nickel-base alloys. Cincinnati
(OH): American Society for Metals, 101-115.
69
[97]. Balsone, S.J., Nicholas, T. and Khobaib. M. (1990). Effects of stress history on the
magnitude of the environmental attack in Rene 80.In: Environmentally assisted cracking:
science and engineering, ASoc Test Mater, Spec Tech Publ, 1049; 1990. p. 303-318.
[98]. Coffin, Jr. L.E. (1963). Cyclic-strain-induced oxidation of high temperature alloys. Trans
Am Soc Met 56, 339-344.
[99]. Tong, J., Dalby, S. and Byrne, J. (2005). Crack growth in a new nickel-based superalloy at
elevated temperature Part III: Characterisation. Journal of Materials Science, 40(5), 1237-
1243.
[100]. Achter, M.R. (1968). The adsorption model for environmental effects in fatigue crack
propagation. Scripta Metallurgica, 2, 525-528.
[101]. Nicholas, T. (1990). Elevated Temperature Crack Growth. Mall, S. and Nicholas, T.
(ed.). ASME, New York, NY, 107-112.
[102]. King, E. and Coterill, P.J. (1990). The role of oxides in fatigue crack propagation.
Materials Science and Technology, 6, 19-31.
[103]. Reuchet, J. and Remy, L. (1982). Fatigue oxidation interaction in a superalloy—
application to life prediction in high temperature low cycle fatigue. Metallurgical and
Materials Transactions A, 14(1), 141-149.
[104]. Gao, H., Brown, M.W. and Miller, K.J. (1982). Mixed mode fatigue threshold. Fatigue
of Engineering Materials and Structures, 5(1), 1-17.
[105]. Miller, M.P., McDowell, D.L. and Oehmke, R.L.T. (1992). A creep-fatigue-oxidation
microcrack propagation model for thermomechanical fatigue. Engineering Materials and
Technolody, 114(3), 282-288.
[106]. McGowan, J.J., and Liu, H.W. (1983). Fatigue-Environment and Temperature Effects, J.
B. John and V. Weis, Plenum Press, New York, 377-390.
[107]. Pfaendtner, J.A. and McMahon, Jr. C.J. (2001). Oxygen-induced intergranular cracking
of a Ni-base alloy at elevated temperatures—an example of dynamic embrittlement. Acta
Materialia, 49(16), 3369-3377.
[108]. Krupp, U., Kane, W.M., Laird, C. and McMahon, C.J. (2004). Brittle intergranular
fracture of a Ni-base superalloy at high temperatures by dynamic embrittlement. Materials
Science and Engineering: A, 387-389, 409-413.
70
[109]. Krupp, U. (2005). Dynamic Embrittlement — Time–dependent Quasi–brittle
Intergranular Fracture at High Temperatures. International Materials Review, 50(2), 83-97.
[110]. Bika, D., McMahon, Jr. C.J. (1995). A model for dynamic embrittlement. Acta
Metallurgica et Materialia, 43(5), 1909-1916.
[111]. Liu, C.T. and White, C.L. (1987). Dynamic embrittlement of boron-doped Ni3Al alloys at
600°C. Acta Metallurgica, 35(3), 643-649.
[112]. Bika, D. (1992). Dynamic embrittlement in metallic alloys. PhD thesis, University of
Pennsylvania, Philadelphia, PA, USA.
[113]. Birnbaum, H.K., Sofronis, P. (1994). Hydrogen-enhanced localized plasticity - a
mechanism for hydrogen-related fracture. Materials Science and Engineering: A, 176(1-2),
191-202
[114]. Misra, R.D.K., Prasad, V.S. (2000). On the dynamic embrittlement of copper-chromium
alloys by sulphur. Journal of Materials Science, 35(13), 3321-3325.
[115]. Misra, R.D.K., McMahon, Jr. C.J. and Guha, A. (1994). Brittle behavior of a dilute
copper-beryllium alloy at 200°C in air. Scripta Metallurgica et Materialia, 31(11), 1471-1474
[116]. Bika, D., Pfaendtner, J.A., Menyhard, M., McMahon, Jr. C.J. (1995). Sulfur-induced
dynamic embrittlement in a low-alloy steel. Acta Metallurgica et Materialia, 43(5), 1895-
1908.
[117]. Krupp, U., Kane, W.M., Liu, X., Pfaendtner, J.A., Laird, C. and McMahon, Jr C.J. (2004).
Oxygen-induced intergranular fracture of the nickel-base alloy IN718 during mechanical
loading at high temperatures. Materials Research, 7(1), 35-41.
[118]. Trindade, V.B., Krupp, U., Wagenhuber, Ph.E.-G., Virkar, Y.M., and Christ, H.-J.
(2005). Studying the role of the alloy-grain-boundary character during oxidation of Ni-base
alloys by means of the electron back-scattered diffraction technique. Materials at High
Temperatures. 22(3-4), 207-212.
[119]. Xu, Y. and Bassani, J.L. (1999). A steady-state model for diffusion-controlled fracture.
Materials Science and Engineering: A, 260(1), 48-54.
[120]. Carranza, F.L., and Haber, R.B. (1999). A Numerical Study of Intergranular Fracture and
Oxygen Embrittlement in an Elastic-Viscoplastic Solid. Journal of the Mechanics and Physics
of Solids, 47(1), 27-58.
71
[121]. Zhao, L.G., Tong, J., and Hardy, M.C. (2010). Prediction of crack growth for a nickel-
base superalloy under fatigue-oxidation condition. Engineering Fracture Mechanics. 77(6),
925-938.
[122]. Li, J.C.M., Oriani, R.A., and Darken, L.S. (1966). The Thermodynamics of Stressed
Solids. Zeitschrift für Physikalische Chemie, 49(3-5), 271-290.
[123]. Larché, F.C., and Cahn, J.W. (1982). The effect of self-stress on diffusion in solids. Acta
Metallurgica, 30(10), 1835-1845.
[124]. Stephenson, G.B. (1988). Deformation during interdiffusion. Acta Metallurgica, 36(10),
2663-2683.
[125]. Pang, H.T., Reed, P.A.S. (2003). Fatigue crack initiation and short crack growth in
nickel-base turbine disc alloys—the effects of microstructure and operating parameters.
International Journal of Fatigue, 25(9-11), 1089-1099.
[126]. Chaboche, J.L. (1989). Constitutive equations for cyclic plasticity and cyclic
viscoplasticity. International Journal of Plasticity 5(3), 247-302.
[127]. Stöcker, C., Zimmermann, M., Christ, H.-J., Zhan, Z.-L., Cornet, C., Zhao, L.G., Hardy,
M.C. and Tong, J. (2009). Microstructural characterisation and constitutive behaviour of alloy
RR1000 under fatigue and creep-fatigue loading conditions. Materials Science and
Engineering A, 518(1-2), 27-34.
[128]. Cornet, C. (2009). Modelling damage and deformation behaviour of a nickel-based
superalloy using constitutive models. PhD Thesis. University of Portsmouth. Portsmouth, UK.
[129]. Cornet, C., Zhao, L.G. and Tong, J. (2011). A study of cyclic behaviour of a nickel-based
superalloy at elevated temperature using a viscoplastic-damage model. International Journal
of Fatigue, 33(2), 241-249.
[130]. Cornet, C., Zhao, L.G. and Tong, J. (2009). Ratchetting strain as a damage parameter in
controlling crack growth at elevated temperature. Engineering Fracture Mechanics, 76(16),
1538-2553.
[131]. Lin, B., Zhao, L.G., Tong, J. and Christ, H.-J. (2010). Crystal plasticity modelling of cyclic
deformation for a polycrystalline nickel based superalloy at high temperature. Materials
Science and Engineering: A, 527(15), 3581-3587.
[132]. Lin, B., Zhao, L.G. and Tong, J. (2011). A crystal plasticity study of cyclic constitutive
behaviour, crack-tip deformation and crack-growth path for a polycrystalline nickel-based
superalloy. Engineering Fracture Mechanics, 78(10), 2174-2192.
72
[133]. Zhao, L.G., Tong, J. (2008). A viscoplastic study of crack-tip deformation and crack
growth in a nickel-based superalloy at elevated temperature. Journal of the Mechanics and
Physics of Solids, 56(12), 3363-3378.
[134]. Zhao, L.G., Tong, J. and Byrne, J. (2004). The evolution of the stress-strain fields near a
fatigue crack tip and plasticity-induced crack closure revisited. Fatigue & Fracture of
Engineering Materials & Structures, 27(1), 19-29.
[135]. Zhao, L.G., Tong, J. and Byrne, J. (2001). Finite element simulation of creep-crack
growth in a nickel base superalloy. Engineering Fracture Mechanics, 68(10), 1157-1170.
[136]. Zhao, L.G. and Tong, J. (2005). Crack growth in a new nickel-based superalloy at
elevated temperature Part II: Finite element analysis of crack growth. Journal of Materials
Science, 40(5), 1229-1235.
[137]. Zhao, L.G. (2011). Modelling of oxygen diffusion along grain boundaries in a nickel-
based superalloy. ASME Journal of Engineering Materials and Technology, 133(3), 031002(7
pages).
73
CHAPTER 3
EXPERIMENTAL WORK
74
CHAPTER 3
3.1. INTRODUCTION
Oxidation damage of alloy RR1000 has been investigated by Encinas-Oropesa et al. [1] at
temperatures between 700C and 800C. Oxidation kinetics of RR1000 has been
characterized by mass change data from thermo gravimetric analyses of small disc samples at
temperatures from 700C to 800C, where the oxide scale rich in Cr and Ti has been found to
grow in a parabolic dependence with time. On the other hand, Focused Ion Beam (FIB)
examinations of oxidized samples for RR1000 revealed the appearance of the oxide scale on
the surface and the formation of re-crystallized grains and voids beneath the oxide scales [1].
The oxidation-linked microstructure damage becomes more significant with the increase of
testing temperature. The thickness of oxide scale and the depth of internal damage,
exemplified as voids along grain boundaries, are more than tripled at 800C when compared
to those at 700C. In addition, oxides can also be observed on the surface of those internal
micro-voids from the FIB images [1] as also confirmed in our recent study [2].
However, the work of Encinas-Oropesa et al. [1] was carried out without considering the
influence of mechanical loading such as fatigue, which can have a significant effect on
oxidation and corrosion damage for nickel and its alloys. For instance, Moulin et al. [3]
studied the influence of external mechanical loading on oxygen diffusion during nickel
oxidation. It was shown that relative oxygen enrichment is observed in the oxide scale under
mechanical loading and the ingress of oxygen in the substrate beneath the oxide scale
becomes easier with a constant tensile load (creep). In the work of Berger et al. [4, 5], the
75
application of a constant load induced multiple cracks at the surface of nickel oxide scale as
well as an increase of oxygen diffusivity by two orders of magnitude, although the
enhancement of oxygen diffusivity decreases with the further increase of load level. The
oxidation kinetics curves gained by thermogravimetric analysis (TGA) indicate that the
oxidation rate of pure nickel was accelerated by both tensile and compressive external stress
[6]. A study of cast nickel superalloy IN 100 showed that matrix oxidation was strongly
enhanced by fatigue cycling [7]. Effects of applied tensile stress on the growth kinetics of
oxide scales were also studied for Ni-20Cr alloys [8], where the internal oxidation and the
oxide layers are shown to be much thicker than that formed without load. This was also the
case for internal nitridation of nickel alloy Inconel 718 under mechanical stress [9]. Stress-
enhanced oxidation was also found for other metallic materials such as Cr-Mo steel, where
both oxide layer and penetration of oxide along preferential paths (e.g., cracks, grain
boundaries) increased significantly under strain-controlled low cycle fatigue [10, 11].
The objective of this chapter is to investigate, experimentally, the effects of cyclic stress and
exposure temperature on oxidation damage of alloy RR1000. Waisted specimens were tested
under cyclic loads in laboratory air at three temperatures (700C, 750C and 800C). FIB
microscopic analyses were performed to examine the effects of cyclic stress level on
oxidation damage at three selected surface locations. Measurements were made for the
thickness of surface oxide scale, the depth of internal damage caused by the oxidation
process as well as the combination of the two. Energy dispersive x-ray (EDX) analyses were
also carried out to identify the chemical composition of the oxides and the element
concentration at varying depth from the specimen surface.
76
3.2. MATERIAL, SPECIMEN AND EXPERIMENTAL DETAILS
3.2.1. MATERIAL
The alloy under investigation is fine grain RR1000, produced through a powder metallurgy
route by Rolls-Royce for disc rotor applications in the latest aero-engines. The chemical
composition of RR1000 is 18.5Co-15Cr-5Mo-3.6Ti-3Al-2Ta-0.5Hf-0.03C-0.02B-0.06Zr and
balance Ni in weight percentage. The alloy has a two-phase microstructure consisting of γ
matrix and strengthening γ΄precipitates Ni3(Al, Ti, Ta) which are largely responsible for the
elevated temperature strength of the alloy. An average grain size of 4 to 5µm was
determined from an EBSD (Electron Backscatter Diffraction) analysis [12].
3.2.2. SPECIMEN GEOMETRY
Waisted specimens, as shown in Figure 3.1, were provided by Rolls-Royce for fatigue testing
in air at high temperatures. The specimen has a diameter of 4.4mm for the narrowest section
and 18mm for the section at the ends. All specimens have been carefully cleaned before the
fatigue tests by longitudinal and cross-hatch polishing before the fatigue tests. A surface
roughness analysis was performed on the specimen using a Mitutoyo Surface tracer SV-C524
and the arithmetic mean value, Ra, was found to be: 0.788μm.
The waisted specimen was chosen for its gradual variation of the cross sectional area, from a
diameter of 4mm at the middle section to a diameter of 18mm at the ends (Figure 3.1).
Under fatigue loading, this will produce a variable stress level across the specimen length.
This specific geometry allows the investigation of the effect of the cyclic stress on oxidation
damage by carrying out the fatigue test on a single specimen for a given temperature.
77
Figure 3. 1: Waisted specimen geometry and the three positions “A”, “B” and “C” selected
for FIB examinations. Drawing provided by Rolls-Royce Plc. Internal communication.
3.2.3. FATIGUE TESTS
The tests were carried out on a computer controlled Instron 8500 servo-hydraulic machine
with a load cell of ±25kN. The machine has a SF868E furnace mounted which can be heated
up to 1100°C. The temperature was monitored by three K-type thermocouples placed on the
side of the specimen. The temperature has a variation of ±2°C between the thermocouple
calibrator and the digital controller, and a variation of less than ±5°C along the specimen. The
tests were conducted in laboratory air condition with vapour content less than 5ppm.
78
Fatigue tests were carried out at three temperatures: 700°C, 750°C and 800°C, under a
triangular loading waveform. For 700°C and 750°C, a load of 10kN was applied, with a load
ratio of R = 0.1 and a loading frequency f = 0.25Hz. For 800°C, the load was reduced to 8kN,
considering the reduced mechanical properties at such a high temperature, while the load
ratio and loading frequency remain unchanged (R = 0.1 and f = 0.25Hz).
3.2.4. FIB ANALYSES
FIB system was used to cut trenches on the specimen surface to reveal and examine the
nature of the oxidised surface after the fatigue tests. The equipment is an FEI Strata 200xP
FIB with a fine gallium (Ga+) beam. The specimen was mounted onto a special grip and then
placed inside the FIB chamber where vacuum conditions apply (1.6×10-5 mbar for the
platinum strip deposition, 5.9×10-6 mbar for the coarse milling and 3.3×10-6 mbar for the fine
cross sectional milling). An area of interest, with a dimension of 25μm2μm1μm, was
selected and a layer of platinum (Pt) deposition was applied on the selected surface to avoid
both contamination and stray sputter damage during the FIB milling operations (Figure 3.2).
The cutting operation using the Ga+ beam was initially in coarse steps (at a current I =
6600pA), followed by finer steps (at a current I = 1000pA) to produce one flat side to the
trench [1, 2]. FIB images were produced via detected back scattered electrons that resulted
from bombardment of the sample using low power Ga+ ion beam, which scans the flat side of
the trench.
79
Figure 3. 2: Waisted specimen surface, the Pt strip applied to protect the oxide scale and
the first stage of rough milling.
In the present work, two different modes have been used for the FIB imaging, i.e., secondary
ions and secondary electrons. FIB secondary ion images (Figure 3.3(b)) are particularly
sensitive to the presence of oxides and carbides in metallic systems due to the effect of these
elements on the secondary ion yield of the metal, which is ideal to reveal oxides and carbides
as the oxygen or carbon “enhanced yield” increases the brightness of the area. FIB secondary
electron imaging (Figure 3.3(a)) is very effective in revealing the grain orientation and
contrast without having to use chemical or electrolytic etching on the material [13].
80
Figure 3. 3: (a). Secondary electron image (b). Secondary ion image.
FIB has been carried out for all specimens to observe the oxide scale and the associated
damage inside the material under fatigue-oxidation conditions. Three locations on each
specimen, as illustrated by "A", "B", "C" in Figure 3.4, were chosen for the FIB analysis to see
the influence of cyclic stress levels on oxidation damage. Point A is located at the narrowest
section, which has the highest stress level, while points B and C are 15mm and 30mm,
respectively, away from the point A and have reduced stress levels due to the increased
cross-sectional areas.
3.2.5. EDX ANALYSES OF THE OXIDES
Energy dispersive X-ray analyses (EDX) of both surface and internal oxides were carried out in
a Field Emission Gun Scanning Electron Microscope (FEG-SEM). Locations at selected
distances from the oxide scale surface were examined to identify the chemical composition
of the oxides at different depths into the specimen. Concentrations of the detected elements
81
were measured in weight percentage and compared with each other to reveal the nature of
oxidation ingress into the substrate beneath the surface oxide scale.
3.3. RESULTS
3.3.1. STRESS ANALYSIS
Finite element analyses were first carried out to compute the stress concentration factor for
the specimen. A quarter of the specimen was considered due to the geometrical symmetry
and meshed into 8-node axisymmetric elements with full integration [14]. The right and
bottom edges are constrained in the x and y directions, respectively, to remove the rigid
body motion. A distributed load of 39MPa, equivalent to 10kN for concentrated load, was
applied to the top surface of the specimen. Contour plot of the normal stress in the y
direction is shown in Figure 3.4(a), where the highest stress concentration is located in the
narrowest section. The stress concentration factor, defined as the ratio of the normal stress
in the y-direction (yy) to the applied stress (app = 39MPa), was calculated for the specimen
and shown in Figure 3.4(b), against the vertical distance from the narrowest section. The
middle section has a stress concentration factor as high as 17. The stress concentration is
gradually reduced towards the ends of the specimen, which allows the study of the influence
of cyclic stress level on the oxidation damage. The stress concentration factors are 17.1, 8.9
and 2.5 for point A, B and C, respectively. Specifically, under the load of 10kN, the stress level
is 670MPa, 350MPa and 100MPa, i.e., 65%, 34% and 10% of 0.2% proof stress of the material
at 650C, at point A, B and C, respectively (see Table 3.1).
82
Table 3. 1: Stress level at the three positions selected on the waisted specimen under a
10kN load.
Position Stress level (MPa) % of proof stress
A 650 65
B 350 34
C 100 10
83
Figure 3. 4: (a). Contour plot of the normal stress in the y direction for the waisted
specimen under an applied load of 10kN, where d is the distance from the narrowest
section of the specimen (b). Stress concentration factor, defined as the ratio of the normal
stress in the y-direction (σyy) to the applied stress (σapp = 39.0MPa), against the distance
from the narrowest section of the specimen.
A
B
C
84
3.3.2. FATIGUE EXPERIMENTS
At 700°C, the specimen was loaded for 176,763 cycles, approximately 196 hours, and
survived the test duration without failure; while fatigue failure occurred in specimens tested
at 750°C and 800°C, with complete fracture at the narrowest section. The fatigue life is
174,392 cycles (194h) at 750°C and 33,473 cycles (38h) at 800°C, indicating a dramatic
reduction of fatigue resistance at temperatures up to 800°C.
Microscopic examination of the fracture surface is shown in Figure 3.5Error! Reference source
not found.(a), for the specimen tested at 750°C, where fatigue crack seems to initiate from the
surface and propagate through the cross section till the final fracture. The side surface was
also examined and shown in Figure 3.5(b) for the specimen tested at 800°C, which shows
numerous surface cracks near the fracture surface. Observed surface cracks for 800°C
specimen represents the fracture of the surface oxide layer, which was essentially caused by
localized severe deformation near the waisted section as a result of the reduced mechanical
properties at 800°C. Surface cracks, i.e., cracking of oxide layers, were not observed for
specimens tested at 700°C and 750°C due to the less extent of oxidation damage and
mechanical deformation near the waisted section.
85
Figure 3. 5: (a). The fracture surface of the specimen fatigue tested at 750°C. (b). Surface
cracks on the side surface of the specimen fatigue tested at 800°C.
86
3.3.3. FIB EXAMINATIONS
FIB analyses were performed to examine the oxidation damage for the tested specimens and
the secondary electron images taken for position "A" are shown in Figure 3.6(a), Figure 3.7(a)
and Figure 3.8(a) for the three temperatures, all of which reveal the surface oxide scales and
internal micro-voids beneath the oxide scale. At 700°C, the oxidation is limited to the surface
area of the specimen. The oxide nodules on the surface do not appear to be uniform, with
patches of oxidised and un-oxidised areas. The micro-voids are almost invisible. The oxide
scale above the surface has a thickness of up to 0.8μm and the depth of internal damage
(below the surface and into the material) is measured up to 1.0μm. The thickness of oxide
scale was measured as the distance between the identified specimen surface and the surface
of the oxide scale; while the depth of internal damage was measured as the distance
between the identified specimen surface and the very tips of internal damage. The
measurements were based on the FIB projections at the selected positions (see Appendix A),
which will vary if a different FIB position is selected.
At 750°C and 800°C, as shown in Figure 3.7(a) and Figure 3.8(a), the specimens showed more
severe oxidation damage when compared to that at 700°C, with a thicker oxide scale and a
deeper damage into the specimen. At 750°C, the thickness of oxide scale is measured up to
1.6μm from Figure 3.6 (a), almost twice as much as that for the 700°C specimen. The depth
of internal damage measured from the obtained FIB image is up to 2.5μm, which is more
than doubled when compared to that for the 700°C specimen. At 800°C, the thickness of
oxide scale and the depth of internal damage have values up to 1.7μm and 7.0μm,
respectively (Figure 3.6Error! Reference source not found.(a)). There is an increased propensity for
87
voids at higher temperatures of 750°C and 800°C, as shown in Figure 3.7(a) and Figure 3.8(a).
The voids are considered to be the result of the loss of base alloy elements (e.g., Cr, Ti Ni),
which diffuse outwards to the specimen surface to form the oxide scale. Also voids are
mainly located along the grain boundaries which are the preferential diffusion paths.
In addition, internal oxide particles can be clearly seen from the secondary ion images in
Figure 3.6(b), Figure 3.7(b) and Figure 3.8(b), where the bright areas below the specimen
surface indicate the oxides within the material. A comparison with the electron images
(Figure 3.6(a), Figure 3.7(a), and Figure 3.8(a)) suggests that internal oxides tend to follow
the traces of the micro-voids formed beneath the oxide scale. Therefore, in addition to the
outwards diffusion of alloy elements (e.g., Cr, Ti, Ni) to form surface oxides, oxygen also
transports into the material through the porous oxide scale and internal voids along the grain
boundaries, and chemically reacts with the alloy elements to form internal oxide particles,
especially, on the surface of the voids.
88
Figure 3. 6: FIB images for position "A" at 700°C (a) secondary electron and (b) secondary
ion, where the dashed line in red indicates the original specimen surface.
89
Figure 3. 7: FIB images for position "A" at 750°C (a) secondary electron and (b) secondary
ion, where the dashed line in red indicates the original specimen surface.
90
Figure 3. 8: FIB images for position "A" at 800°C (a) secondary electron and (b) secondary
ion, where the dashed line in red indicates the original specimen surface.
91
3.3.4. EFFECT OF STRESS LEVEL ON OXIDATION DAMAGE
To examine the effects of cyclic stress level on oxidation damage, FIB images, taken at the
three locations A, B and C, were compared. The same types of oxidation damage were
observed at all three locations, i.e., the oxide scales built on the surface and the internal
micro-voids along the grain boundaries. This can be seen from the FIB images of oxidation
damage presented in Error! Reference source not found. for position B and C at 800°C.
Furthermore, the thickness of surface oxide scale and the depth of internal damage were
measured for the three locations and presented in Error! Reference source not found.(a), (b) and
(c) for 700°C, 750°C and 800°C, respectively. It can be seen that, at 750°C and 800°C, the
cyclic stress has a notable influence on the severity of the oxidation damage. Both the oxide
scale and the depth of internal damage increased with the increase of the stress level. At
800°C, the scale of overall damage at the highest stress area (Point A) nearly doubled that at
the lowest stress area (Point C), with a measured difference of 3.8μm. At 750°C, the scale of
total oxidation damage at Point A is 0.5μm more than that at Point C. However, at 700°C, the
effect of cyclic stress on oxidation damage seems to be less clear, as shown in Error! Reference
source not found.(a). This might be due to the relatively low level of oxidation kinetics at this
temperature [1].
92
Figure 3. 9: (a). Comparison of the thickness of the oxide scale and depth of internal
damage, for positions A, B and C at 700°C. (b). Comparison of thickness of oxide scale and
depth of internal damage for positions A, B and C at 750°C. (c). Comparison of thickness of
oxide scale and depth of internal damage for positions A, B and C at 800°C.
93
3.3.5. COMPOSITION OF OXIDES
Figure 3. 10: Five spectrums selected for EDX analyses, over a depth of 2.2μm from the
oxide scale surface and with equal spacing (0.54μm) between each other.
As shown in Error! Reference source not found.(b), Figure 3.7(b) and Figure 3.8(b), secondary ion
imaging confirms the formation of oxides both on the specimen surface and in the specimen.
For further information, EDX analyses of the trench at position "A" were performed for
specimens tested at 750°C and 800°C to analyse the chemical composition of those oxides at
five different depths from the surface. As shown in Figure 3.10, the five spectrums selected
for EDX analyses spread over a depth of 2.2μm from the oxide scale surface, with equal
spacing (0.54μm) between each other. Specifically, the first spectrum is located at the oxide
scale surface while the fifth spectrum has a depth of 2.2μm into the material. The obtained
element profiles at the five spectrums are shown in Figure 3.11 and Figure 3.12 for 750°C and
800°C, respectively.
94
Figure 3. 11: (a). EDX element profiles showing weight% of Cr, Ni, Ti and Co against the
distance from the oxide scale surface (750°C). (b). EDX element profiles showing weight%
of O, Mo, Al and Ta against the distance from the oxide scale surface (750°C).
Figure 3. 12: (a). EDX element profiles showing weight% of Cr, Ni, Ti and Co against the
distance from the oxide scale surface (800°C). (b). EDX element profiles showing weight%
of O, Mo, Al and Ta against the distance from the oxide scale surface (800°C).
As seen from Figure 3.11(a) and Figure 3.12(a), the surface oxide scale mainly consists of Ni,
Cr, Ti and Co oxides, indicating the formation of Cr2O3, TiO2, NiO and Co3O4 oxides on the
specimen surface, which is consistent with the work of Encinas-Oropesa et al. [1]. The
formation of the Cr2O3 layer indicates a stage of saturation during the oxidation process of
95
the alloy, which may act as a protection from further oxidation. Cr and Ti levels decreased
progressively with the increase of depth, accompanied by the progressive increase of Ni and
Co levels, a result of approaching the base material. The oxides also contain a small
percentage of Mo, Al and Ta, as shown in Figure 3.11(b) and Figure 3.12(b). The level of O
decreases with the increase of depth at 750°C (Figure 3.11(b)), while it seems to remain
steady at 800°C (Figure 3.11(b)) probably due to the presence of multiple surface cracks
which facilitated the penetration of oxygen into the specimen. This suggests the ingress of
oxygen into the material and formation of internal oxide particles, as also evidenced from the
secondary ion images (Figure 3.6(b), Figure 3.7(b) and Figure 3.8(b)). At 800°C, a reduced
percentage of O element was obtained when compared to that at 750°C (see Figure 3.11(b)
and Figure 3.11(b)), which may be due to the lower density of oxides at 800°C caused by the
increased propensity and depth of micro-voids (Figure 3.7(a) and Figure 3.8(a)), as well as the
multiple surface cracks (Figure 3.5(b)) generated under fatigue loading. The patchy presence
of oxide scales for the specimen tested at 700°C made the EDX analyses rather difficult and
inaccurate. Also, the internal damage for the 700°C specimen is too limited to obtain a
meaningful concentration-depth profile and therefore EDX analyses were not carried out for
700°C specimen.
3.4. DISCUSSION
Oxidation damage of alloy RR1000 has been investigated by Encinas-Oropesa et al. [1] for
small disc samples at temperatures between 700C and 800C, in the absence of mechanical
loading. Oxidation kinetics has been characterized by mass change data from
thermogravimetric analyses, where the mass gain has been found to grow in a parabolic
96
function with time. In addition, FIB examinations of oxidised samples revealed the
appearance of the oxide scale, rich in Cr and Ti, on the surface and the formation of
recrystallised grains and micro-voids beneath the oxide scales [1]. Internal oxide particles can
also be observed on the surface of those micro-voids from the FIB images. The thickness of
oxide scale and the depth of internal damage at 800C are more than tripled when compared
to those at 700C. Our results suggest the same types of oxidation damage for alloy RR1000
in the presence of fatigue loading. Fatigue tends to increase the thickness of oxide scale and
depth of internal damage for temperatures at 750°C and above. For instance, at 750°C, the
added measurements of oxide scale and internal damage are 4.1µm, 3.3µm and 3.2µm for
Position A, B and C, respectively, as compared to 2.0µm reported in Encinas-Oropesa et al.
[1]. At 800°C, the oxidation damage becomes much more severe under fatigue loading. The
added measurements of oxide scale and internal damage are 8.7µm, 5.7µm and 4.8µm for
Position A, B and C, respectively, as compared to 3.9µm reported in Encinas-Oropesa et al.
[1]. The considerably increased oxidation damage for position A at 800°C may be due to the
numerous surface cracks formed near the fractured section (Figure 3.5(b)) which act as short
circuits for oxygen penetration. At 700°C, fatigue influence seems to be negligible on the
oxidation damage, at least for the load level considered in the present work. The measured
oxidation damage for the three positions A, B and C is in the range of 1.7µm to 1.9µm, which
is comparable to the measurement of 1.5µm in Encinas-Oropesa et al. [1]. It should be
pointed out that the specimens used in Encinas-Oropesa et al. [1] has a surface finish of
0.25µm, which is different from the 0.79µm surface finish of the waisted specimen tested in
the present work. This might have an influence on the measured oxidation damage.
97
EDX analyses in Encinas-Oropesa et al. [1] were reported for the surface oxides, which show
the enrichment of Cr and Ti as well as low Ni and Co level. This is consistent with the results
in the present work. Similar oxides were also found for other types of nickel superalloys. For
instance, Chen et al. [35] studied the oxidation behaviour for nickel alloys Waspaloy, Astroloy
and Udimet 720 after isothermal exposures at temperature between 750°C and 1,000°C. It
was reported that X-ray diffraction detected Cr2O3, TiO2 and NiCr2O4 oxides on the surface of
the materials [15]. In addition, NiO was also found in Astroloy after a 20 hour exposure at
750°C [15]. Surface oxides for Inconel 718 were reported to consist of Ni or Nb rich Cr2O3 [16,
17]. It was well recognized that the formation of Cr2O3 is largely responsible for the superior
oxidation resistance of nickel-based superalloys.
During oxidation, outwards diffusion of alloy elements (Cr, Ni and Co) leads to the growth of
surface oxides, as well as resulting in the formation of internal voids due to the loss of alloy
elements. The voids seem to locate along the grain boundaries, indicating the primary path
of alloying elements diffusion. Interestingly, internal oxidation was indentified from the
secondary ion imaging (Figure 3.6(b), Figure 3.7(b), Figure 3.8(b)), suggesting the
simultaneous inwards diffusion of oxygen to form internal oxide particles. Penetration of
oxygen was also confirmed by the presence of O element from the EDX analyses at the
selected depths into the material (Figure 3.9 and Figure 3.12). Internal oxide particles are
considered to be a mixture of TiO2 and Al2O3 [2, 16, and 17]. Internal oxidation, especially
along grain boundaries, may lead to critical embrittlement of the material and results in
significantly shortened fatigue life via intergranular fracture process. As shown in Zhao [18],
simulation of internal oxidation can be done from oxygen diffusion analyses based on Fick’s
98
first and second laws. The depth of oxygen penetration and concentration-depth profile can
be predicted from the oxygen diffusion analyses and further utilized to quantify the material
embrittlement caused by internal oxidation process.
Investigation of the effect of mechanical stress on oxidation has significant implications for
the oxidation-accelerated crack growth and failure behaviour. For polycrystalline nickel
superalloys, environmental effects were frequently observed to modify the fracture
morphology from transgranular to intergranular during crack growth, as well as to increase
the crack growth rates significantly for a given stress intensity factor [19-22]. Recently, it has
been acknowledged and evidenced that the strong environmental effect on crack growth of
nickel based superalloy is due to the penetration of oxygen into a crack tip. Under the
influence of a tensile stress, oxygen may move into the grain boundaries by diffusion,
attacking the grain boundaries by themselves or through chemical reactions with alloy
elements. The oxygen diffusion process at a crack tip is a dynamic process, a combined effect
of time, temperature, local deformation and microstructure. Oxygen diffusion may be
facilitated by high tensile stresses near the crack tip [23-26], especially for superimposed
dwell periods at the maximum load where the crack tip is fully open and experiences
sustained high tensile stresses. The work of Zhao et al. [27] showed that crack growth rates in
air at high-temperature can be well predicted from a dynamic embrittlement hypothesis.
For modelling of stress-assisted oxygen diffusion near the crack tip, a coupling between the
oxygen diffusion and the mechanical deformation is generally required to include the effect
of crack tip deformation (stress fields) on oxygen diffusion process [27, 28]. In this case,
diffusion of oxygen within the material is controlled by two essential parameters, namely,
99
the oxygen diffusivity and the pressure factor [27, 28]. So far, no direct measurements are
available for oxygen diffusion parameters in nickel alloys. In fact, the present work measured
the scale of oxidation damage as well as the concentration of O element under varying cyclic
stress levels from the fatigue-oxidation tests on waisted specimens, which are useful in the
determination of the two essential parameters in oxygen diffusion analyses. Some pilot work
has been carried out in Zhao [18], where modelling of oxygen diffusion along grain
boundaries was attempted under creep loading condition based on the coupling of crystal
plasticity with oxygen diffusion. In Chapter 4, finite element simulations are presented for
calibration of the oxygen diffusion parameters based on the modelling of oxygen diffusion
and penetration in waisted specimen under fatigue.
3.5. CONCLUSIONS
Formation of surface oxide scales (Cr, Ti, Ni and Co oxides) and micro-voids beneath the
oxide scales seem to be the major mechanisms for oxidation damage of alloy RR1000 under
fatigue loading conditions. Both surface oxides and internal voids could lead to crack
initiation and subsequent propagation into the substrate under mechanical loading
conditions, resulting in material failure as occurred for the cyclic tests at 750C and 800C.
Cyclic stress influenced oxidation damage at 750C and 800C, and the thickness of oxide
scale and the depth of internal damage were enhanced by the increase of stress level. Both
EDX analyses and secondary ion imaging show the penetration of oxygen into the material to
form internal oxide particles, which leads to further embrittlement of the material.
Measurements of oxidation damage and oxygen concentration have been carried out, and
100
could be utilised to estimate the oxygen diffusivity and the pressure factor which are two
essential parameters for modelling of oxidation damage and prediction of crack growth
under fatigue-oxidation conditions.
101
3.6. REFERENCES
[1]. Encinas-Oropesa, A., Drew, G.L., Hardy, M.C., Leggett, A.J., Nicholls, J.R., Simms, N.J.
(2008). Effects of oxidation and hot corrosion in a nickel disc alloy. Champion, Pennsylvania,
USA : Wiley, 2008. Superalloys 2008. pp. 609-618.
[2]. Karabela, A., Zhao, L.G., Tong, J., Nicholls, J. R., Simms, N. J., Hardy, M. C. (2011). Effects
of cyclic stress and temperature on oxidation damage of a nickel-based superalloy. Materials
Science and Engineering: A, 528(19-20), 6194-6202 .
[3]. Moulin, G., Arevalo, P., Salleo, A. (1996). Influence of external mechanical loadings
(creep, fatigue) on oxygen diffusion during nickel oxidation. Chemistry and Materials Science,
45(1-2), 153-181.
[4]. Berger, P., Moulin, G., Viennot, M. (1997). Nuclear microprobe study of stress-oxidation
of nickel. Nuclear Instruments and Methods in Physics Research, Section B; Beam interactions
with materials and Atoms, 130(1-4), 717-721.
[5]. Berger, P., Gaillet, L., El Tahhann, R., Moulin, G., Viennot, M. (2001). Oxygen diffusion
studies in oxide scales thermally grown or deposited on mechanically loaded metallic surfaces
(MS-P2). Nuclear Instruments and Methods in Physics Research, Section B; Beam interactions
with materials and Atoms, 181(1-4), 382-388.
[6]. Zhou, C.H., Ma, H.T., Wang, L. (2010). Comparative study of oxidation kinetics for pure
nickel oxidized under tensile and compressive stress. Corrosion Science, 52(1), 210-215.
[7]. Reger, M., Remy, L. (1988). Fatigue oxidation interaction in IN 100 superalloy.
Metallurgical Transactions. A, Physical Metallurgy and Materials Science, 19A(9), 2259-2268.
[8]. Calvarin-Amiri, G., Huntz, A.M., Molins, R. (2001). Effects of an applied stress on the
growth kinetics of oxide scales formed on Ni-20Cr alloys. Materials at High Temperatures,
18(2), 91-99.
[9]. Krupp, U. (2010). Nitridation of alloys. Shreir's Corrosion, 1, 304-315.
[10]. Azari, Z., Abbadi, M., Moustabchir, H., Lebienvenu, M. (2008). The influence of fatigue
cycling on the oxidation kinetics and crack initiation of a Cr-Mo steel. International Journal of
Fatigue, 30(3), 517-527.
[11]. Fournier, B., Sauzay, M., Caës, C., Noblecourt, M., Mottot, M., Bougault, A., Rabeau,
V., Pineau, A. (2008). Creep-fatigue-oxidation interactions in a 9Cr-1Mo martensitic steel.
102
Part I: Effect of tensile holding period on fatigue lifetime. International Journal of Fatigue,
30(4), 649-662.
[12]. Stöcker, C., Zimmermann, M., Christ, H.-J., Zhan, Z. L., Cornet, C., Zhao, L. G., Hardy,
M. C., Tong, J. (2009). Microstructural Characterisation and Constitutive Behaviour of Alloy
RR1000 under Fatigue and Creep-Fatigue Loading Conditions. Materials Science and
Engineering A, 518, 27-34.
[13]. Fibics. Using FIB secondary ion images to observe the presence of corrosion. Retrieved
from: http//www.fibics.com/fib/application/using-fib-secondary-ion-images-to-observe-the-
presence-of-corrosion-/22/, 2009.
[14]. ABAQUS. Version 6.8. Dassault Systems Simulia Corp., Providence, USA, 2009.
[15]. Chen, J.H., Rogers, P.M., Little, J.A. (1997). Oxidation behavior of several chromia-
forming commercial nickel-base superalloys. Oxidation of Metals, 47(5-6), 381-410.
[16]. Trindade, V.B., Krupp, U., Wagenhuber, Ph.E.-G. , Virkar, Y.M., Christ, H.-J. (2005).
Studying the role of the alloy-grain-boundary character during oxidation of Ni-base alloys by
means of the electron back-scattered diffraction technique. Materials at High Temperatures,
22(3-4), 207-212.
[17]. Geng, L., Na, Y.S., Park, N.K. (2007). Oxidation behavior of Alloy 718 at a high
temperature. Matrials & Design, 28(3), 978–981.
[18]. Zhao, L.G. (2011). Modelling of oxygen diffusion along grain boundaries in a nickel-
based superalloy. ASME Journal of Engineering Materials and Technology, 133(3), 031002.
[19]. Andrieu, E., Molins, R. (1992). Intergranular crack tip oxidation mechanism in a nickel-
based superalloy. Materials Science and Engineering A, 154(1), 21-28.
[20]. Ghonem, H., Zheng, D. (1992). Depth of intergranular oxygen diffusion during
environment-dependent fatigue crack growth in alloy 718. Materials Science and Engineering:
A, 150(2), 151-160.
[21]. Molins, R., Hochstetter, G., Chassaigne, J.C. and Andrieu, E. (1997). Oxidation effects
on the fatigue crack growth behaviour of alloy 718 at high temperature. Acta Materialia,
45(2), 663-674.
[22]. Andrieu, E., Hoshstetter, G., Molins, R., Pineau, R. (1997). Oxidation and intergranular
cracking behaviour of two high strength Ni-based superalloys. Corrosion-Deformation
Interactions (EFC 21), 461-475.
103
[23]. Pfaendtner, J.A., McMahon Jr., C.J. (2001). Oxygen-induced intergranular cracking of a
Ni-base alloy at elevated temperatures—an example of dynamic embrittlement. Acta
Materialia, 49(16), 3369-3377.
[24]. Krupp, U., McMahon Jr., C.J. (2004). Dynamic embrittlement—time dependent-brittle
fracture. Journal of Alloys and Compounds, 378(1-2), 79-84.
[25]. Krupp, U., Kane, W.M., Laird, C., McMahon Jr., C.J. (2004). Brittle intergranular fracture
of a Ni-base superalloy at high temperatures by dynamic embrittlement. Materials Science
and Engineering A, 387-389, 409-413.
[26]. Krupp, U. (2005). Dynamic embrittlement: time-dependent quasi-brittle intergranular
fracture at high temperatures. International materials reviews, 50(2), 83-97.
[27]. Zhao, L.G., Tong, J. and Hardy, M.C. (2010). Prediction of crack growth in a nickel-based
superalloy under fatigue-oxidation conditions. Engineering Fracture Mechanics, 77(6), 925-
938.
[28]. Carranza, F.L., Haber, R.B. (1998). A numerical study of intergranular fracture and
oxygen embrittlement in an elastic-viscoplastic solid. Journal of the Mechanics and Physics of
Solids, 47(1), 27-58.
104
CHAPTER 4
MICROSCOPY ANALYSES OF FRACTURE
SURFACES IN DWELL CRACK GROWTH
105
CHAPTER 4
4.1. INTRODUCTION
Damage tolerance approaches have been used to evaluate the service life for critical
structural components such as gas turbine discs made of nickel alloys [1]. Such
methodologies require data and predictions for crack growth rates under service loading
conditions. For alloy RR1000, crack growth tests on CT specimens have been carried out at
elevated temperature to investigate the effects of loading frequency, waveform and dwell
time on crack growth [2, 3]. In the frequency domain, the majority of the crack growth rate
data seem to have three distinct regimes, cycle-dependent, time-dependent and the mixed
cycle/time-dependent crack growth [4]. Similar clarifications can also be found in literatures
[5-9]. Time-dependent crack growth is usually observed at low frequencies with
predominantly intergranular cracking while cycle-dependent crack growth occurs at high
frequency with transgranular cracking [4]. Transitional frequencies indicate mixed
cycle/time-dependent crack growth behaviour [4]. Predominantly time-dependent crack
growth may be because of creep or environmental embrittlement due to oxidation [10]. For
alloy RR1000, as creep deformation was found extremely limited within the temperature
range of interest [11, 12], oxidation embrittlement was thought to be the dominant
mechanism for time-dependent crack growth [4].
Crack growth tests, with a dwell period (ranging from 1s to 1000s) imposed at the maximum
load, showed an increased crack growth rate with the increase of the dwell period [2, 3]. For
instance, at 650C, crack growth rate for 1000s dwell is about 20 times that for 1s dwell
106
under K=30MPam and load ratio R = 0.1 (see Figure 6.5 in Chapter 6). Recent tests on
coarse grain RR1000 showed that crack growth rate for 1 hour dwell can be 60 times that
without dwell at 700C [13]. The accelerated crack growth under long dwell has been
attributed to the oxidation embrittlement, i.e., time-dependent crack growth. In this chapter,
scanning electron microscopy (SEM) has been used to examine the fracture surface of CT and
CN specimens tested under selected dwell times, in order to further establish the
mechanisms for crack growth under such conditions. Focussed Ion Beam (FIB) analysis of
fracture surfaces was also performed to examine the oxidation damage during crack growth
under dwell loading conditions. Table 4.1 gives a list of the specimens examined using the
SEM and FIB.
Table 4. 1: Specimens used for SEM and FIB examinations (tested under selected
temperatures and dwell times).
Specimen type Temperature (°C) Dwell time (s)
CT 650 & 725 20, 300, 1000
CN 650, 700 & 750 3600
4.2. MICROSCOPY ANALYSES OF CT SPECIMENS
Standard CT specimens [14], cut from an unused turbine disk of fine grain RR1000
(FGRR1000) supplied by Rolls-Royce plc, were used for dwell fatigue tests at 650C and 725C
by Dalby and Tong [2, 3]. The tests were performed on a servo-hydraulic Instron machine and
crack growth was monitored using a Direct Current Potential Drop (DCPD) method [2, 3]. A
constant K was applied with a load ratio of 0.1 and a base-line frequency of 0.25Hz. During
107
the testing, a dwell period between 20s and 1000s was introduced at the maximum load. The
microscopy analyses of CT specimens [2, 3] were carried out for comparison purposes with
the CN specimens of coarse grain RR1000 (Section 4.3). Figure 4.1 shows a view of the
fracture surface, with indication of crack growth direction and imposed variable dwell time.
As can be seen, the surface within the crack growth area shows varying darkness and
brightness due to the variation of dwell time introduced in the testing. The measured crack
growth rates were shown to increase with the increase of dwell time for both temperatures
[2, 3]. Specifically, at 650C, crack growth rate for 1000s dwell is about 20 times that for 1s
dwell under K=30MPam and load ratio R = 0.1 (see Figure 6.5 in Chapter 6). While at
725C, crack growth rate for 1000s dwell is about 30 times that for 1s dwell under
K=30MPam and load ratio R = 0.1 [2, 3].
Figure 4. 1: Fracture surface of a CT specimen tested at T=650°C for variable dwell times.
108
4.2.1. SEM FRACTOGRAPHY
SEM has been employed to study the fracture modes with respect to the dwell time at both
temperatures and the results are shown in Figures 4.2 and Figure 4.3. The influence of load
level was removed by the K-controlled testing methods such that all micrographs taken
were at a similar stress intensity level. As seen from Figures 4.2 and 4.3, a mixture of
transgranular and intergranular crack growth was observed for all cases, although the
proportion of the transgranular crack growth decreased as the temperature and dwell time
increased. No cavities in the vicinity of the crack tip or grain boundaries were observed in all
cases studied, indicating a lack of creep deformation, consistent with previous experimental
and numerical studies [2, 3, 11, and 12]. Figures 4.2 shows the micrographs of fracture
surfaces of the specimen tested under 20s, 300s and 1000s dwell at 650C. Typical features
of transgranular cracking can be observed at 20s dwell with clearly visible striations,
characteristic of cycle-dependent fatigue crack growth. At 300s dwell, some fatigue striations
are still visible but with a reduced proportion, and a fair amount of intergranular cracking can
be observed. At 1000s dwell, the fracture surface exhibits a predominant intergranular
cracking, although transgranular cracking is still present. The general morphology of the
fracture surface seems to be consistent with the SEM micrographs, with a smooth and flat
surface at 20s dwell as opposed to a much rougher surface at 1000s dwell; while the
roughness of fracture surface at 300s dwell lies in between. It seems that as the dwell time
increased, the proportion of intergranular cracking also increased and became predominant
at 300s dwell and above.
109
Figures 4.3 shows the micrograph of fracture surfaces of the specimen tested under 20s,
300s and 1000s dwell at K = 30MPam and 725C. The fracture mode, as indicated in Figure
4.3, is mainly intergranular for all three dwell times, a direct result of temperature increase.
Some transgranular cracking is still present, but the proportion is reduced with the increase
of dwell time. The surface is generally rougher when compared to those shown in Figure 4.2
for 650C. A closer look also indicates the presence of oxides on fracture surfaces at this
temperature, which is believed to be the main mechanism for intergranular cracking due to
grain boundary, decohesion by oxidation.
Figure 4. 2: SEM micrographs of the fractured surface of a CT specimen tested at 650°C
under (a) 20s, (b) 300s and (c) 1000s dwell.
110
Figure 4. 3: SEM micrographs of the fractured surface of a CT specimen tested at 725°C
under (a) 20s, (b) 300s and (c) 1000s dwell.
4.2.2. FIB EXAMINATION
It is well known that nickel-based alloys are particularly susceptible to oxygen exposure at
intermediate temperatures. Oxide formation may reduce boundary mobility and prompts
crack initiation and growth. The oxide deposits have been observed on fracture surfaces
using SEM examination for all specimens tested at 650C and 725C [2, 4]. To further
examine the oxide morphology, FIB analyses have been carried out here following the same
procedures as described in Chapter 3. The FIB Position at the final crack front, as indicated in
Figure 4.1, was chosen for the FIB analysis. A micrograph of FIB position is shown in Figure
111
4.4, which reveals the severe roughness of the fracture surface. This roughness tended to
spatter the Pt ions on the surface and made FIB analysis very difficult.
Figure 4. 4: Micrograph of the selected FIB location, which also shows the severe surface
roughness.
Figure 4.5 shows the secondary electron image taken for FIB position, which reveals a very
thin and light oxide film built on the surface. Grain microstructure can also be clearly seen
from the picture. For a clearer view of the surface oxides, secondary ion images were also
taken and shown in Figure 4.6. In addition to the surface oxide scale, occasional penetration
of oxides into the material, about 3.6μm, can be seen, and seems to follow the path of a
secondary crack. Overall, the oxide scale is very thin for 650C and generally below 1μm, as
shown by the measurements in Figure 4.6.
Figure 4.7 shows the secondary ion FIB image at 725C, which again reveals the oxide scale
built on the fracture surface. The oxide scale has an increased thickness when compared to
112
that at 650C, and measured to be between 0.7 and 1.6μm, which is consistent with the SEM
examination by Dalby and Tong [2, 4].
Figure 4. 5: The secondary electron image for FIB analysis which reveals a very thin and
light oxide film built on the surface.
113
Figure 4. 6: The secondary ion image for FIB analysis, showing the surface oxide scale and
penetration of oxides into the material along the path of a secondary crack.
Figure 4. 7: The secondary ion image for FIB analysis, which reveals a thicker oxide scale,
built on the fracture surface (Testing temperature at 725°C).
114
4.3. MICROSCOPY ANALYSES OF CN SPECIMENS
Recently, dwell tests as long as 1 hour have been carried out for coarse grain RR1000 alloy in
corner notch (CN) specimens at Portsmouth, as part of the DISPLACE programme [13]. The
specimens were provided by Rolls-Royce plc. Tests were carried out at 650C, 700C and
750C on Single Screw Instron 1362 machine with crack length monitored by a DCPD system
[13]. The tests were carried out under constant load. For 650°C, a load of 31.5kN was applied
and for 700 and 750°C, the applied load was 21.6kN. The fracture surfaces of three tested
specimens are shown in Figures 4.8, 4.9, and 4.10, respectively, with indications of the
testing conditions as well as the three positions chosen for the FIB analyses. Variation of
surface morphologies can be observed, reflecting the variation of testing conditions. Imposed
one-hour dwell leads to severe roughness of the fracture surfaces when compared to those
for pure fatigue tests at both room and elevated temperature.
115
Figure 4. 8: Fractured surface of the CN specimen tested at 650°C, with indications of the
three FIB positions.
Figure 4. 9: Fractured surface of the CN specimen tested at 700°C, with indications of the
three FIB positions.
116
Figure 4. 10: Fractured surface of the CN specimen tested at 750°C, with indications of the
three FIB positions.
4.3.1. SEM FRACTOGRAPHY
Figure 4.11 shows the SEM micrographs of fracture surfaces of the specimen tested at 650C,
dominantly transgranular cracking can be observed for room temperature fatigue, with the
presence of fatigue striations which indicate the crack growth path. For one-hour dwell, the
fracture surface clearly exhibits intergranular cracking growth. This is also the case for the
specimen tested at 700°C, as shown in Figure 4.12, i.e., intergranular crack propagation for
one-hour dwell condition and transgranular mode for crack growth at room temperature.
This is consistent with the general morphology of the fracture surface, with a smooth and flat
surface at room temperature fatigue and a much rougher surface at 3600s dwell (Figures 4.8
and 4.9).
117
At 750C, SEM examinations (Figure 4.13) also confirm intergranular fracture for one-hour
dwell and transgranular cracking for room temperature fatigue. While a mixture of
transgranular and intergranular cracking can be seen for 0.25Hz fatigue tests, without dwell,
at 750C, indicating a combined nature of cycle and time dependent crack growth. The
yellow line in Figure 4.13(a) indicates the boundaries of crack growth between 0.25Hz fatigue
and the one-hour dwell at 750C.
Figure 4. 11: SEM pictures of the fractured surface of the CN specimen R134 (a) room
temperature fatigue and (b) 1 hour dwell at 650°C.
Figure 4. 12: SEM pictures of the fractured surface of the CN specimen R134 (a) room
temperature fatigue and (b) 1 hour dwell at 700°C.
118
Figure 4. 13: SEM pictures of the fractured surface of the CN specimen tested at 750°C: (a)
0.25Hz followed by 1 hour dwell and (b) 1 hour dwell.
4.3.2. FIB EXAMINATION
A closer look at SEM pictures, particularly those at 750C (Figure 4.13), shows bright-colour
oxides built on the fracture surface. For further examination, FIB analyses of fracture surfaces
have been carried out for all three specimens. The three selected FIB locations for each
specimen are indicated in Figures 4.8, 4.9 and 4.10, respectively. Position 1 is close to the
free surface while positions 2 and 3 are located in the middle section. More specifically,
Position 2 is located close to the final crack front while Position 3 is at the starting crack front
for the one-hour dwell test. Secondary Galium ion scan was used in the FIB analyses. Figure
4.14 gives a comparison of the FIB images at position 1 for the three temperatures, all
showing the oxide scales built on the fracture surface. The oxide scales have been traced
using colour lines which clearly show the increase of oxide scale thickness with the increase
of temperature. Penetration of oxides into the material is evidenced at 750C; similar results
were also obtained for Positions 2 and 3, as shown in Figures 4.15 and 4.16. There seems no
119
discernible difference in the apparent oxide thickness for the three different positions at a
given temperature.
Figure 4. 14: Secondary Ion images at FIB position 1 for the specimens tested under one-
hour dwell at (a) 650°C, (b) 700°C and (c) 750°C. The colour lines indicate the surface oxide
boundaries.
120
Figure 4. 15: Secondary Ion images at FIB position 2 for the specimens tested under one-
hour dwell at (a) 650°C, (b) 700°C and (c) 750°C. The colour lines indicate the surface oxide
boundaries.
121
Figure 4. 16: Secondary Ion images at FIB position 3 for the specimens tested under one-
hour dwell at (a) 650°C, (b) 700°C and (c) 750°C. The coloured lines indicate the surface
oxide boundaries.
For nickel alloys, an appreciable increase in crack propagation rate in air is generally
observed when static loading periods (dwells) are included in the fatigue cycle [16, 17]. This
increase may be due to oxidation-caused embrittlement in oxygen environment, i.e.,
environmentally enhanced crack growth [18]. This is also the case for crack growth in alloy
RR1000 [2-4]. Nickel alloys are particularly sensitive to oxidation-assisted crack growth
mechanisms [18-20], and the loading frequency as well as micro structural effects appear
122
only in air, and disappear under vacuum [21]. The process of oxidation embrittlement occurs
at the crack tip during fatigue crack propagation. The abrupt change in fracture mode (from
transgranular to intergranular cracking) confirms [22] that the role of oxygen is to reduce the
grain boundary cohesion in such a way that the crack growth path becomes intergranular.
This has also been evidenced from our SEM studies of fracture surfaces, especially for the
long dwell tests.
At the crack front, intergranular diffusion of oxygen takes place, which also depends on the
stress and strain states along the affected grain boundaries [23–25]. Miller [26] showed that
environmentally enhanced crack growth in IN718 occurs by the stress-enhanced oxygen
diffusion ahead of the crack tip, followed by the oxidation of Ti, Al, Cr, as well as Nb if
present. The oxidation of all these elements was considered to affect (either favourably or
otherwise) the environmentally enhanced crack growth. So far, there have been no direct
measurements of oxygen ahead of the crack tip, probably due to very limited level of
oxidation into the crack tip as well as the insufficient resolution of current facilities. For
instance, when oxidation was studied by photo-electron spectroscopy [27], it was found that
the oxide thickness was in the order of the escape depth of photoelectrons. Similarly, in our
work FIB analyses only reveal oxides on the fracture surfaces while oxides ahead of the crack
front could not be seen. However, the rate of oxide formation in the steady state oxygen-
affected area was found to be compatible with the oxygen atom flux at the crack tip [27], an
indirect support for oxygen flux into the crack tip.
As discussed by Ghonem [28], oxygen tends to penetrate the crack-tip material along rapid
diffusion paths, for example, slip planes and grain boundaries, which may be characterised by
123
internal oxide sites, cavity formations and/or solute segregation. Oxygen could also take part
in chemical reactions, thus releasing embrittlement agents onto a grain boundary [29]. These
processes prevent the sliding and migration of boundaries, and reduce the ability of the
material to relieve the build-up of local stresses during deformation. Consequently, grain
boundary cracking, which leads to intergranular cracking, tends to occur. This has been
evidenced from the SEM examinations in the current work, which shows pre-dominant
intergranular cracking with the increased dwell time. Oxygen penetration into the crack tip is
both time and deformation dependent. High-level stress and strain fields [30-33], as well as
strain rates, ahead of the crack tip could play a significant role in the mechanism of oxygen-
induced intergranular fracture. This is especially true for superimposed dwell periods at the
maximum load, where the crack tip is fully open and experiences sustained high tensile
stresses.
Oxygen partial pressure is another important factor which influences oxidation and oxygen
diffusion at the crack tip. The pressure of oxygen controls the preferential formation of
particular species of oxides which may either shield or contribute to the crack-tip damage
process through the passivation or the enhancement of the oxidation process, respectively
[28]. For instance, Andrieu et al. [33] carried out tests and analyses of oxide formation on
deformed and non-deformed free surfaces of large grain sized alloy IN718 as a function of
oxygen partial pressure. The oxidation process starts with the formation of selective oxides
of FeO and NiO and their spinels once the atmospheric oxygen partial pressure is available at
the crack tip. These oxides are porous and thus permit the oxidation process to continue at
the base grain boundary material, hence contributes to the crack-tip damage process. Over a
124
time equal to or less than the oxide passivation time, the oxygen partial pressure tends to
decrease at the interface thereby, providing the required reaction conditions for the
formation of a less porous, denser sub-layer of Cr2O3, which may shield the crack-tip damage
process. These aspects have not been studied in this work.
In addition, the well-known tunnelling phenomenon in crack growth can be clearly observed
for the CN specimens tested under 1 hour dwell (Figures 4.8 to 4.10), i.e., faster crack growth
in the middle section than near the free surface. This has been attributed to the stronger
crack-tip constraint in the middle section, exemplified by higher T-stress level, which
facilitates crack growth by limiting the near-tip plastic deformation [34]. On the other hand,
an oxidation-based rationale has also been proposed to explain the observed tunnelling
phenomenon in crack growth in air, i.e., due to more severe oxidation damage in the middle
section than near the free surface. However, such rationale could not be supported from the
observed oxide accumulation on the fracture surface, as oxides accumulation does not
appear to have a link with the position on the surface, according to our FIB examinations
(Figures 4.14 to 4.16). Further and in-depth investigation is required in order to understand
the true mechanisms for tunnelling crack growth.
4.4. CONCLUSIONS
SEM studies of fracture surfaces of samples tested under dwell conditions have been carried
out, and the results confirmed the transition from transgranular to predominantly
intergranular cracking in alloy RR1000 for increased dwell time. This change in fracture mode
supports the oxidation-assisted crack growth mechanism via grain boundary embrittlement,
125
as creep deformation is very limited for alloy RR1000 [2, 3, 11, 12]. FIB analyses of fracture
surfaces also confirmed the presence of oxidation for alloy RR1000 at the temperatures
studied.
Oxygen penetration into the crack tip, which is both time and deformation dependent, is
believed to be the source for grain boundary embrittlement ahead of the crack tip and
subsequent intergranular crack growth; although experimental evidence of crack tip oxygen
diffusion process and measurements of oxygen concentration near the crack tip are yet to be
available.
126
4.5. REFERENCES
[1]. King, J.E. (1983). Fatigue crack propagation in nickel-base superalloys—effects of
microstructure, load ratio, and temperature. Materials Science and Technology, 3(9), 750-
764.
[2]. Dalby, S. (2005). The effects of frequency and waveshape on the fatigue crack growth of
an advanced nickel based superalloy at elevated temperatures. Ph.D. Thesis. University of
Portsmouth, Portsmouth, UK.
[3]. Dalby, S. Tong, J. (2005). Crack growth in a new nickel-based superalloy at elevated
temperature, Part I: Effects of loading waveform and frequency on crack growth. Journal of
Materials Science, 40(5), 1217-1228.
[4]. Tong, J., Dalby, S. and Byrne, J. (2005). Crack growth in a new nickel-based superalloy at
elevated temperature Part III Characterisation. Journal of Materials Science, 40(5), 1237-
1243.
[5]. Branco, C.M. and Byrne, J. (1996). Elevated temperature fatigue of IN718: Effects of
stress ratio and frequency. In: Proceedings of the 81st Meeting of the AGARD/SMP, Thermal
Mechanical Fatigue of Aircraft Engine Materials, Oct 2–4 1995; Banff, Canada (1996) [AGARD
Rep. 669, Paris, France. p. 6-1–6-12].
[6]. Pineau, A. (1984). High Temperature Fatigue: Creep--Fatigue--Oxidation Interactions in
Relation to Microstructure. Subcritical Crack Growth Due to Fatigue, Corrosion and Creep;
Ispra; Italy; 19-23 Oct. 1981, 483-530.
[7]. Nicholas, T., Weerasooriya, T., and Ashbaugh, N.E. (1985). A model for creep/fatigue
interaction in alloy 718. Fracture mechanics, 16th Symposium. American Society for Testing
and Materials, Special Technical Publications, 868; 167-80.
[8]. Weerasooriya, T., and Venkataraman, S. (1987). Frequency and environment effect on
crack growth in Inconel 718. Effects of load and thermal histories. Warrendale (PA):
Metallurgical Society of AIME; 101-8.
[9]. Weerasooriya, T. (1988). Effect of frequency on fatigue crack growth rate of Inconel 718
at high temperature. Fracture mechanics, 19th Symposium American Society for Testing and
Materials, Special Technical Publications, 969; 907-23.
[10]. Tong, J., Dalby, S., Byrne, J., Henderson, M.B. and Hardy, M.C. (2001). Creep, fatigue
and oxidation in crack growth in advanced nickel base superalloys. International Journal of
Fatigue, 23(10), 897-902.
127
[11]. Zhao, L.G. and Tong, J. (2005). Crack growth in a new nickel-based superalloy at
elevated temperature Part II Finite element analysis of crack growth. Journal of Materials
Science, 40(5), 1229-1235.
[12]. Zhao, L.G., Tong, J. and Byrne, J. (2001). Finite element simulation of creep-crack
growth in a nickel base superalloy. Engineering Fracture Mechanics, 68(10), 1157-1170.
[13]. Tong J. (2009-2011) Developing improved service propagation lives in arduous cyclic
environments (DISPLACE). TSB funded project.
[14]. ASTM Standards. (1995). Standard test method for measurement of fatigue crack
growth rates. ASTM E647-95a, 557-593.
[15]. Tong, J., Cornet, C., Lupton, C. and Karabela, A. (2010). Developing improved service
propagation lives in arduous cyclic environment (DISPLACE) project. TSB Project. University of
Portsmouth, Portsmouth, UK.
[16]. Shahinian, P., and Sadananda, K. (1976). Crack growth behavior under creep-fatigue
conditions in alloy 718. Proceedings ASME-MPC Symposium on Creep-Fatigue Interaction,
New York, 365-390.
[17]. Wei, R.P. and Huang, Z. (2002). Influence of dwell time on fatigue crack growth in
nickel-base superalloys. Materials Science and Engineering A, 336(1–2), 209-214.
[18]. Floreen, S., and Kane, R.H. (1979). Effects of environment on high-temperature fatigue
crack growth in a superalloy. Metallurgical and Materials Transactions A, 10(11), 1745-51.
[19]. Shahinian, P., and Achter, M.R. (1957). A comparison of the creep–rupture properties of
nickel in air and in vacuum. Trans. Metall. Soc. AIME, 215, 37-41.
[20]. Sadananda, K. and Shahinian, P. (1983). Creep crack growth behavior of several
structural alloys. Metallurgical and Materials Transactions A, 14(7), 1467-1480.
[21]. Pedron, J.P. and Pineau, A. (1982). Effect of microstructure and environment on crack
growth behaviour in Inconel 718 alloy at 650°C under fatigue, creep, and combined loading.
Materials Science and Engineering, 56, 143-156.
[22]. Osinkolu, G.A., Onofrio, G. and Marchionni, M. (2003). Fatigue crack growth in
polycrystalline IN 718 superalloy. Materials Science and Engineering A, 356(1–2), 425-433.
[23]. Sadananda, K. and Shahinian, P. (1985). Effects of environment on high temperature
crack growth behavior of several nickelbase alloys. Corrosion of nickel-base alloys. Cincinnati
(OH), American Society for Metals, 101-115.
128
[24]. Balsone, S.J., Nicholas, T. and Khobaib, M. (1990). Effects of stress history on the
magnitude of the environmental attack in Rene 80. Environmentally assisted cracking:
Science and Engineering, ASTM STP 1049, Special Technical Publications, 303-318.
[25]. Coffin, Jr. L.E. (1963). Cyclic-strain-induced oxidation of high temperature alloys. Trans.
Am. Soc. Met. , 56, 339-344.
[26]. Miller, C.F., Simmons, G.W. and Wei, R.P. (2003). Evidence for internal oxidation during
oxygen enhanced crack growth in P/M Ni-based superalloys. Scripta Materialia, 48(1), 103-
108.
[27]. Gao, M., Dwyer, D.J. and Wei, R.P. (1995). Niobium enrichment and environmental
enhancement of creep crack growth in nickel-base superalloys. Scripta Metallurgica et
Materialia, 32(8), 1169-1174.
[28]. Ghonem, H., Nicholas, T. and Pineau, A. (1993). Elevated temperature fatigue crack
growth in alloy 718—Part II: Effects of environmental and material variables. Fatigue &
Fracture of Engineering Materials & Structures, 16(6), 577-590.
[29]. Woodford, D.A. and Bricknell, R.H. (1983). Environmental embrittlement of high
temperature alloys by oxygen. Treatise on Materials Science and Technology, 25, 157-199.
[30]. Clavel M. and Pineau, A. (1978). Frequency and wave-form effects on the fatigue crack
growth behaviour of alloy 718 at 298 K and 823 K. Metallurgical and Materials Transactions
A, 9(4), 471-480.
[31]. Andersson, H. and Persson, C. (2001). Crack growth in IN718 at high temperature.
International Journal of Fatigue, 23(9), 817-827.
[32]. Fournier, L., Delafosse, D. and Magnin, T. (2001). Oxidation induced intergranular
cracking and Portevin-Le Chatelier effect in nickel base superalloy 718. Materials Science and
Engineering: A, 316(1), 166-173.
[33]. Andrieu, E., Molins, R., Ghonem, H. And Pineau, A. (1992). Intergranular crack tip
oxidation mechanism in a nickel-based superalloy. Materials Science and Engineering: A,
154(1), 21-28.
[34]. Zhao, L.G., Tong, J. and Byrne, J. (2001). Stress intensity factor K and the elastic T-stress
for corner cracks. International Journal of Fracture, 109(2), 209-225.
129
CHAPTER 5
MODELLING OF FATIGUE-ASSISTED
OXYGEN DIFFUSION AT GRAIN LEVEL
130
CHAPTER 5
5.1. INTRODUCTION
Stress-assisted oxidation has been confirmed for RR1000 in the fatigue-oxidation study in
Chapter 3, where penetration of oxygen and associated internal oxidation are facilitated by
the influence of fatigue from a temperature of 750°C and above. A diffusion approach can be
used to model oxygen penetration and associated internal oxidation damage of the material
at high temperatures. Generally, computational simulation of oxygen diffusion [1-4] are
based on Fick’s first and second laws, where the effect of stress on oxygen diffusion is
assumed to be controlled by the gradient of hydrostatic stress [5-8]. In this case, diffusion of
oxygen within the material is controlled by two essential parameters, namely, the oxygen
diffusivity and the pressure factor [1-4]. This approach has been applied to study the
hydrogen and oxygen embrittlement of engineering materials under mechanical loading
conditions, especially for prediction of crack growth under fatigue-oxidation [3, 4].
Very recently, simulation of stress-driven oxygen diffusion for RR1000 has been carried out at
grain level, where the material micro-structure was explicitly considered [9]. The
heterogeneous deformation at grain level, caused by the orientation mismatch between
neighbouring grains, introduces a large local stress gradient and was shown to have
significant influence on the oxygen diffusion process along the grain boundaries [9].
However, the diffusion parameters used in [9] were estimated from the surface oxide scale
growth, rather than the depth of internal oxidation.
131
The objective of this chapter is to model the oxygen diffusion process for alloy RR1000 at
grain level, aiming to quantify the oxidation damage and, more importantly, to calibrate the
diffusion parameters based on the internal oxygen penetration. Finite element method has
been used to simulate the diffusion process for a waisted specimen, where the grain
boundaries were taken as the primary path for oxygen diffusion and subsequent internal
oxidation. Both the natural and the fatigue-influenced diffusion processes were modelled. A
sequentially coupled mechanical-diffusion analysis was adopted to account for the effects of
deformation on diffusion during fatigue loading with the assistance of a finite element
submodelling technique. Diffusion parameters have been obtained by matching the
simulated results with FIB measurements of internal oxidation depth in Chapter 3.
5.2. MODELLING OF OXYGEN DIFFUSION
5.2.1. GOVERNING EQUATIONS
Under applied stress, the oxygen transport inside the material can be modelled as [5-7]:
where C is the concentration of the solute, t the time, the gradient, D the oxygen
diffusivity, P the hydrostatic stress (pressure) and M the pressure factor. The constitutive
law shown in Equation (5.1) has been used to account for the influence of mechanical stress
on mass transport [1-8, 10, 11]. The gradient of hydrostatic stress has been regarded as the
driving force for stress assisted diffusion. In the absence of M ( M = 0 ) the constitutive law
will be reduced to natural diffusion. As shown in Suo et al. [12] and El Kadiri et al. [13],
132
oxidation, by itself, can generate growth stresses, which have been neglected in this work for
simplification.
5.2.2. OXYGEN FLUX BOUNDARY CONDITIONS
The boundary condition for oxygen diffusion from the surface into the material follows the
parabolic law, which can be determined from the oxidation kinetics study. The oxidation
kinetics of RR1000 have been studied in [14] for temperatures between 700C to 800C. The
mass gain due to oxidation is a parabolic function of time and can be expressed as:
Δ
where Δ is the mass gain, the parabolic oxidation rate and the time.
The flux of oxygen into the material at the surface can be derived by performing
differentiation with respect to time for equation (5.2), which gives:
Equation (5.3) has been applied as flux boundary condition in the diffusion analyses, where
the value of is given in Encinas-Oropesa et al. [14] for RR1000.
5.2.3. CRYSTAL PLASTICITY
The crystal plasticity theory was used to describe the material deformation at grain level. The
flow rule is expressed in terms of, slip resistance and back stress [15].
133
where, k is the Boltzmann constant, the resolved shear stress on the slipped system ,
the absolute temperature, and are the shear moduli at and 0 Kelvin, respectively, and
, , , and are material constants.
The two internal variables, the slip resistance in each slip system and back stress , are
introduced at the slip system level, which represent the state associated with the current
dislocation network and can be expressed as [16]:
and
where and are hardening constants and and describe the dynamic recovery
behaviour.
Equation (5.5) describes the grain interior deformation without taking into account the grain
boundary deformation, as our simulation is focussed on the effects of heterogeneous stress
(caused by the mismatched mechanical behaviour of the individual grains) on oxygen
diffusion. This is believed to be a major source of stress gradient generated across the grain
boundaries and also the driving force for stress assisted diffusion.
5.2.4. MODEL PARAMETERS AND UMAT
Crystal plasticity formulation was implemented numerically into the finite element code
ABAQUS [17] via a user-defined material subroutine (UMAT), where the fully implicit (Euler
backward) integration algorithm was adopted [18, 19]. Values of the model parameters were
fitted from the uniaxial experimental data of alloy RR1000 at 650°C [20, 21], which give a
very good simulation of the material’s constitutive behaviour tested for a wide range of
strain-controlled loading conditions [22, 23].
134
5.2.5. FINITE ELEMENT MODEL
For natural diffusion analyses, an FE model consists of 150 grains (Voronoi tessellation, with
randomly assigned orientation), as shown in Figure 5.1(a) and (c), was created and meshed
into 1,450 four-node first-order elements with full integration using ABAQUS [17]. The
element size is about 1μm, quarter of the average grain size (4 to 5μm) of the material.
Specifically, each grain has about 20 to 30 finite elements. A local stress-strain study has
been carried out in the recent work [22, 23] and the results verified the convergence of the
mesh density. Grain boundaries have been meshed into one-dimensional diffusion elements
[17], which share the same nodes with the bulk elements. On the right edge, the oxygen flux
boundary condition, as given by Equation (5.3), was applied to each node to represent the
surface oxidation process.
To study the effects of mechanical stress on oxygen diffusion, a quarter of the waisted
specimen geometry was considered under fatigue loading condition and the finite element
mesh for the global model, as shown in Figure 5.1(b), consists of first-order elements with full
integration [17]. The sub model is the same as that used for natural diffusion in Figure 5.1(a).
To solve the fatigue-oxidation problem, a sequentially coupled mechanical-diffusion analysis
was performed to include the dependence of oxygen diffusion on the stress state. The
mechanical analyses for the global model and sub-model were carried out first, with the
history of hydrostatic stress for each node stored in the ABAQUS [17] output files. Crystal
plasticity model has been used for the sub-model FE analyses, while elasticity has been used
for the global model analyses as the applied load is below the yield level as described in
135
Chapter 3. During the diffusion analysis, the output files were then used as inputs to include
the effect of hydrostatic stress gradient on oxygen diffusion.
Oxygen diffusion was assumed to occur along the grain boundaries only, as for polycrystalline
nickel and its alloys; oxidation and the associated diffusion of oxygen are primarily along the
grain boundaries. Oxygen diffusion in bulk material was excluded in the diffusion analyses by
assigning negligible diffusivity to the bulk elements.
Figure 5. 1: (a) The 150-grain submodel with random orientation (b) Finite element mesh
for the global model and (c) Inverse pole figure for the 150 grains.
136
5.3. RESULTS AND DISCUSSION
5.3.1. NATURAL DIFFUSION
Modelling of natural diffusion of oxygen has been carried out first, where the oxygen
diffusivity was calculated using the depth of oxygen penetration measured in Chapter 3. The
diffusivity (or diffusion coefficient) was calculated based on [24]:
where is the diffusivity, is the depth of internal oxidation measured from the FIB analysis
in Chapter 3 (position C for the waisted specimen) and is the time. The obtained values for
700C, 750C and 800C are given in Table 5.1, where the value for 650C has been fitted
using an Arrhenius equation (
, with D0 = 205.504mm2/s, Q = 270 kJ/mol and R
= 8.314J/(Kmol)).
Table 5. 1: Oxygen diffusivity at four different temperatures.
T (°C) D(mm2/s)
650 1.085E-13
700 1.397E-12
750 5.859E-12
800 1.423E-11
137
Figure 5. 2: For 650°C the diffusion value has been determined using the Arrhenius
equation and the experimental values obtained for 700, 750 and 800°C, as shown on the
graph.
The contour plot of oxygen concentration for temperatures 650C and 800C are shown in
Figure 5.3 after 200 hours of oxidation, where grain boundaries in red colour show the
penetration of oxygen into the material. It is seen that, for 800C, the depth (10.135µm) of
oxygen penetration along grain boundary is about four times that (2.753µm) for 650C due
to the significantly increased oxygen diffusivity at 800°C (Table 5.1).
0
0.5x10-11
1.0x10-11
1.5x10-11
960 1020
eqn: y=a*exp(-270000/(8.314472*x)), R:1.88e-012,a=205.504
Temperature (K)
Diffu
sio
n (
mm
2/s
)
138
Figure 5. 3: Contour plots of oxygen concentration along the grain boundaries for natural
diffusion: (a) 650°C and (b) 800°C.
139
Results in Figure 5.4 show the predicted oxygen concentration against the distance from the
surface after 200 hours of oxidation at four different temperatures. For 650C and 700C,
oxygen penetration is very limited, about 1 to 3μm, with very low values of oxygen
concentration as well. Severe oxygen penetration can be seen for 750C and 800C,
accompanied by high values of oxygen concentration. The depth of oxygen penetration along
grain boundaries is about 6.0μm for 750C and 10.0μm for 800C, indicating a strong
dependency of oxidation on temperature.
In the absence of mechanical stress, oxygen diffusion purely depends on the temperature,
which is consistent with the experimental study of oxidation of alloy RR1000 [14]. The
oxidation damage of alloy RR1000 has been investigated by Encinas-Oropesa et al. [14] at
temperatures between 700C and 800C. Oxidation kinetics of RR1000 has been
characterized by mass change data from thermogravimetric analyses of small disc samples at
temperatures from 700C to 800C, where the oxide scale rich in Cr and Ti have been found
to grow in a parabolic dependence with time. Mass gain increased significantly with the
increase of temperature. On the other hand, Focused Ion Beam (FIB) examinations of
oxidized samples for RR1000 revealed the appearance of the oxide scale on the surface and
the formation of recrystallised small grain and large voids beneath the oxide scales [14]. The
oxidation-linked microstructure damage becomes more significant with the increase of the
testing temperature. The thickness of oxide scale and the depth of internal damage at 800C
are more than tripled when compared to those at 700C. Internal oxides can also be
observed on the surface of those voids from the FIB images [14]. However, in the present
140
work, no attempt was made to model chemical reaction kinetics, vacancy formation or the
transport of other species [14]. Further work is required to address these issues.
Figure 5. 4: Oxygen concentration against the distance from the specimen surface for four
different temperatures.
5.3.2. EFFECT OF MECHANICAL STRESS ON OXYGEN DIFFUSION
To study the effects of mechanical stress on oxygen penetration, finite element analyses of
the global model and sub-model were first carried out for 800C under fatigue loading
condition using elasticity and crystal plasticity material model, respectively. Figure 5.5 shows
the contour plot of the hydrostatic stress for the sub-model loaded for 33473 cycles at an
applied load of 8kN (f = 0.25Hz and R = 0.1). Heterogeneous stress distributions can be clearly
observed and the maximum value of the local hydrostatic stress (~328MPa) is about twice
0
0.2
0.4
0.6
0.8
1
1.2
0 0.002 0.004 0.006 0.008 0.01
Co
nce
ntr
atio
n (
mg/
mm
3)
Distance from specimen surface (mm)
T=650°C
T=700°C
T=750°C
T=800°C
141
the global hydrostatic stress (~181MPa). A closer inspection shows that most stress
concentrations are located at the triple or multiple points along grain boundaries due to the
mismatch of the mechanical properties of multiple grains.
Figure 5. 5: Contour plot of the hydrostatic stress (peak load level) for the submodel after
33473 cycles at 800°C for an applied load of 8kN.
Following the mechanical deformation analyses, diffusion analyses was carried out for a
chosen pressure factor of 0.3MPa-1, where the stress-driven oxygen diffusion was accounted
for through Equation (5.1). A contour plot of the oxygen concentration is shown in Figure
5.6(b) for 800C, where fatigue loading tends to induce a non-uniform distribution of oxygen
concentration along the grain boundaries when compared to the case for natural diffusion
(Figure 5.6(a)). As illustrated in Figure 5.6(b), the maximum depth of oxygen penetration into
the material is found to be = 7.313μm, which is twice more than that for natural diffusion
(0 = 3.207μm).
Grain
142
Figure 5. 6: Contour plot of the oxygen concentration in the submodel at 800°C after 33473
cycles: (a) natural diffusion and (b) with a pressure factor of 0.3MPa-1.
5.3.3. DETERMINATION OF PRESSURE FACTOR
According to Equation (5.1), two parameters (diffusivity and pressure factor) are essentially
required to model fatigue-assisted oxygen diffusion. Diffusivity has been calculated from the
143
FIB measurements of depth of internal oxidation and given in Table 5.1. To work out the
pressure factor, FE analyses have been carried out to obtain the depth of oxygen penetration
for different pressure factor values, and the results are shown Figures 5.7 and 5.8 for 750°C
and 800°C, respectively. For both temperatures, the oxygen penetration depth () is
normalised against that (0) for natural diffusion.
Figure 5. 7: Depth of oxygen penetration against the pressure factor for T=750°C, where the
penetration depth (δ) under fatigue is normalized against that (δ0) for natural diffusion.
At 750°C, the penetration depth of oxygen follows a linear trend with the increase of the
pressure factor (Figure 5.7). While, at 800°C, the penetration depth of oxygen has a linear
trend for pressure factor up to a value of 0.17MPa-1, then followed by a sudden increase
(Figure 5.8). According to the FIB measurements in Chapter 3, oxygen penetration depth has
a normalised value (δ/δ0) of 1.209 for 750C and 2.012 for 800C, which corresponds to a
0.95
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11
δ/δ
0 (
μm
)
Pressure factor (MPa-1)
Pressure factor = 0.068MPa-1
δ/δ0=1.209 (FIB measurement)
144
pressure factor value of 0.068MPa-1 and 0.188MPa-1 for 750C and 800C respectively, as
indicated in Figure 5.7 and Figure 5.8.
Figure 5. 8: Depth of oxygen penetration against the pressure factor for T=800°C, where the
penetration depth (δ) under fatigue is normalized against that (δ0) for natural diffusion.
The obtained values for 750C and 800C are given in Table 5.2, where the values for 650C
and 700C have been fitted using an Arrhenius equation (
, with
, Q = 270 kJ/mol and R = 8.314J/(Kmol)).
Table 5. 2: Oxygen pressure factor at four different temperatures.
T (°C) Pressure Factor (MPa-1)
650 0.001558
700 0.0098
750 0.068
800 0.188
0.9
1.15
1.4
1.65
1.9
2.15
2.4
0 0.05 0.1 0.15 0.2 0.25 0.3
δ/δ
0 (
μm
)
Pressure factor (MPa-1)
Pressure factor = 0.188MPa-1
δ/δ0=2.012 (FIB measurement)
145
5.4. CONCLUSION
Finite element analyses of oxygen diffusion along grain boundaries have been carried out for
a waisted specimen using a submodelling technique. Oxygen diffusivity was calculated from
the FIB measurements of depth of internal oxidation in Chapter 3. For natural diffusion,
oxygen penetration and concentration along grain boundaries are controlled by the
diffusivity, which is temperature dependent. The higher is the temperature, the deeper is the
oxygen penetration. Penetration of oxygen is facilitated by the presence of fatigue for both
750°C and 800°C, and the simulation results have been used to obtain the pressure factor
which has been further used to model the oxidation-accelerated fatigue crack growth in
Chapter 6.
146
5.5. REFERENCES
[1]. Krupp, U., Kane, W.M., Laird, C., McMahon Jr., C.J. (2004). Brittle intergranular fracture
of a Ni-base superalloy at high temperatures by dynamic embrittlement. Materials Science
and Engineering A, 387-389, 409-413.
[2]. Krupp, U. (2005). Dynamic embrittlement: time-dependent quasi-brittle intergranular
fracture at high temperatures. International materials reviews, 50(2), 83-97.
[3]. Carranza, F.L., Haber, R.B. (1998). A numerical study of intergranular fracture and oxygen
embrittlement in an elastic-viscoplastic solid. Journal of the Mechanics and Physics of Solids,
47(1), 27-58.
[4]. Zhao, L.G., Tong, J. and Hardy, M.C. (2010). Prediction of crack growth in a nickel-based
superalloy under fatigue-oxidation conditions. Engineering Fracture Mechanics, 77(6), 925-
938.
[5]. Li, J.C.M., Oriani, R.A. and Darken, L.S. (1966). The thermodynamics of stressed solids. Z.
Phys. Chem. Neue Folge, 49, 271-290.
[6]. Larchè, F.C., and Cahn, J.W. (1982). The effect of self-stress on diffusion in solids. Acta
Metallurgica, 30(10), 1835-1845.
[7]. Stephenson, G.Brian. (1988). Deformation during interdiffusion. Acta Metallurgica,
36(10), 2663-2683.
[8]. Bika, D. and McMahon Jr, C.J. (1995). A model for dynamic embrittlement. Acta
Metallurgica et Materialia, 43(5), 1909-1916.
[9]. Zhao, L.G. (2011). Modelling of oxygen diffusion along grain boundaries in a nickel-based
superalloy. ASME Journal of Engineering Materials and Technology, 133(3), 031002.
[10]. Sofronis, P. and Mc Meeking, R.M. (1989). Numerical analysis of hydrogen transport
near a blunting crack tip. Journal of the Mechanics and Physics of Solids, 37(3), 317-350.
[11]. Yokobori Jr., A.T., Chinda, Y., Nemoto, T., Satoh, K. and Yamada, T. (2002). The
characteristics of hydrogen diffusion and concentration around a crack tip concerned with
hydrogen embrittlementnext term. Corrosion Science, 44(3), 407-424.
[12]. Suo, Z., Kubair, D.V., Evans, A.G., Clarke, D.R. and Tolpygo, V.K. (2003). Stresses
induced in alloys by selective oxidation. Acta Materialia, 51, 959-974.
147
[13]. El Kadiri, H., Horstemeyer, M.F. and Bammann, D.J. (2008). A theory of stress-driven
interfacial damage upon cationic-selective oxidation of alloys. Journal of the Mechanics and
Physics of Solids, 56, 3392-3415.
[14]. Encinas-Oropesa, A., Drew, G.L., Hardy, M.C., Leggett, A.J., Nicholls, J.R., Simms, N.J.
(2008). Effects of oxidation and hot corrosion in a nickel disc alloy. Champion, Pennsylvania,
USA : Wiley, 2008. Superalloys 2008. pp. 609-618.
[15]. Busso, E.P. (1990). Cyclic deformation of monocrystalline nickel alluminide and high
temperature coatings. (Ph.D. Thesis )Department of Mechanical Engineering, Massachusetts
Institute of Technology. Cambridge, MA, USA.
[16]. Denis, R.J. (2000). Mechanistic modelling of deformation and void growth behaviour in
superalloy single crystals. (PhD Thesis). Imperial College London. London, UK.
[17]. ABAQUS, 2009, Version 6.8, Dassault Systemes Simulia Corp., Providence, USA.
[18]. Busso, E.P., Meissonnier, F.T. and O'Dowd, N.P. (2000). Gradient-dependent
deformation of two-phase single crystals. Journal of the Mechanics and Physics of Solids,
48(11), 2333-2361.
[19]. Meissonnier, F.Y., Busso, E.P. and O'Dowd, N.P. (2001). Finite element implementation
of a generalised non-local rate-dependent crystallographic formulation for finite strains.
International Journal of Plasticity, 17(4), 601-640.
[20]. Zhan, Z.L. (2004). A study of cyclic plasticity and viscoplasticity in a new nickel-based
superalloy using unified constitutive equations. PhD Thesis. University of Portsmouth.
Portsmouth, U.K.
[21]. Zhan, Z.L., Tong J. (2007a). A study of cyclic plasticity and viscoplasticity in a new nickel-
based superalloy using unified constitutive equations. Part I: evaluation and determination of
material parameters. Mechanics of Materials, 39, 64-72.
[22]. Lin, B., Zhao, L.G., Tong, J. and Christ, H.-J. (2010). Crystal plasticity modeling of cyclic
deformation for a polycrystalline nickel based superalloy at high temperature. Materials
Science and Engineering: A, 527(15), 3581-3587.
[23]. Lin, B., Zhao, L.G. and Tong, J. (2011). A crystal plasticity study of cyclic constitutive
behaviour, crack-tip deformation and crack-growth path for a polycrystalline nickel-based
superalloy. Engineering Fracture Mechanics, 78(10), 2174-2192.
148
[24]. Azari, Z., Abbadi, M., Moustabchir, H., Lebienvenu, M. (2008). The influence of fatigue
cycling on the oxidation kinetics and crack initiation of a Cr-Mo steel. International Journal of
Fatigue, 30(3), 517-527.
149
CHAPTER 6
MODELLING OF CRACK GROWTH UNDER
FATIGUE-OXIDATION CONDITION
150
CHAPTER 6
6.1. INTRODUCTION
For nickel alloys, it was experimentally observed that oxidation modifies the fracture
morphology from transgranular to intergranular during crack growth, as well as increases the
growth rates significantly at a given stress intensity factor [1-4]. The effect of oxidation on
crack growth becomes more pronounced at low frequencies and long dwell periods.
Recently, it has been acknowledged [5-7] that oxidation-assisted crack propagation in nickel
alloys is a result of oxygen diffusion into a local area with increasing tensile stress, such as
crack tips. The oxygen element attacks the grain boundaries by lowering boundary cohesion
and prompts accelerated intergranular cracking [5-7]. The diffusion process is facilitated by
high tensile stresses near the crack tip [8, 9] especially for superimposed dwell periods at the
maximum load where the crack tip is fully open and experiences sustained high tensile
stresses.
Modelling of oxygen diffusion coupled with viscoplasticity has been developed to study
oxidation-assisted crack growth for alloy RR1000 [10]. A failure envelope, expressed in terms
of oxygen concentration and accumulated inelastic strain, has been constructed and utilised
to predict crack growth rates in a CT specimen under fatigue-oxidation conditions. The
predictions were evaluated against the experimental results for triangular and dwell loading
waveforms, with marked improvements achieved over those predicted from the mechanical
deformation alone, i.e., neglecting the influence of oxidation [10]. However, one major
limitation of the work is the use of assumed values of the two essential diffusion parameters,
151
namely the oxygen diffusivity and pressure factor, required in the modelling of oxygen
diffusion near the crack tip.
The objective of this work is to further investigate the process of oxygen diffusion into a
crack tip and its subsequent effects on crack growth, using the newly obtained diffusion
parameters in Chapter 5. Finite element analyses of a compact tension (CT) specimen were
carried out to model crack-tip oxygen diffusion, and more importantly, construct a failure
curve for crack growth based on the consideration of both oxygen concentration and
accumulated inelastic strain near the crack tip. The failure curve was then utilized to predict
crack growth rates under fatigue-oxidation conditions for selected loading frequencies and
dwell periods, with comparison against the experimental results [11, 12] and those predicted
from the viscoplastic model.
6.2. VISCOPLASTICITY DEFORMATION ANALYSIS
Mechanical deformation, mainly fatigue and creep, must be considered when designing and
assessing the service life of high temperature turbines. Viscoplastic or time-dependent
inelastic behaviour plays a very important role in crack tip deformation and crack
propagation for nickel alloys subjected to high temperature cyclic loading. Reduced loading
frequency and imposed hold time at the maximum load often lead to increased deformation
near the crack tip and plastic zone size. Here, finite element analyses have been carried out
first to investigate the role of viscoplasticity in crack growth.
152
6.2.1. THE MATERIAL MODEL
Unified constitutive equations, developed by Chaboche [13], were adopted to describe the
viscoplastic deformation for RR1000. The isotropic and kinematic hardening
variables are considered during the transient and saturated stages of cyclic response [13]. A
small strain hypothesis is used where the strain rate tensor is divided into two parts an
elastic part and a plastic part [13]:
It is assumed that the elastic strain obeys Hooke’s law:
Where and are the Young’s modulus and Poisson’s ratio
respectively, σ and I are the stress tensor and unit tensor of rank two respectively and tr is
the trace.
The inelastic strain represents both plastic and creep strains. A power relationship
adopted for the viscoplastic strain rate is expressed as [13]:
where is the yield function, the back stress or kinematic hardening variable, and are
material constants and the bracket "<>" is defined by:
153
According to the von Mises yield criterion, the yield function is defined as:
where is the isotropic hardening variable and the initial value of the radius of the yield
surface and denotes the von Mises distance in the deviatory stress space:
where and are the deviators of and and : represents the inner product of two
tensors. Plastic flow occurs under the condition: and
For this model, the motion of yield surface continues to hold but the stress in excess of the
yielding stress is now admissible and often termed as overstress.
The evolution of the kinematic stress tensor and the isotropic stress may be described
through the following rules [13]:
154
here , , , , and are six material and temperature-dependent constants, which
determine the shape and amplitude of the stress-strain loops during the transient and
saturated stage of cyclic response, and is the accumulated inelastic strain rate defined by:
The constitutive equations contain eleven material parameters: , , , , , , , , ,
and . The kinematic hardening behaviour is described by , , and , where and
are the saturated values of the kinematic hardening variables and and indicate the
speed with which the saturation is reached. The isotropic hardening is depicted by and ,
where is the asymptotic value of the isotropic variable at saturation and indicates the
speed towards the saturation. The initial size of the yield surface is represented by , and
and are the viscous parameters. The model has already been established [14-16], with
parameters optimised from cyclic test data of RR1000 at 650°C. The model has also been
programmed into a UMAT [17], using a fully implicit integration and the Euler backward
iteration algorithm, and implemented in the finite element software ABAQUS [18].
6.2.2. FINITE ELEMENT ANALYSIS
A standard compact tension (CT) specimen is considered for the finite element study of
fatigue crack-tip deformation and crack growth, and has the same geometry as that used for
the crack growth testing by Dalby and Tong [11, 12]. The specimen (width W = 26mm and
height H = 15.6mm), was meshed into 2D plane strain elements. As shown in Figure 6.1,
refined mesh (4-node, full integration) has been used for the crack tip area, while slightly
155
coarse mesh (3-node, full integration) was used outside the crack tip area in order to save
the computation time. At the crack tip, the element has a size of 6.35μm, which is about the
average grain size (5~6μm) of the material [11, 12]. The crack length was chosen to be a =
13mm, half of the specimen’s width (a/W = 0.5). All FE analyses have been carried out with a
constant stress intensity factor range of and ratio of R = 0.1. The load was
applied to the specimen through a rigid pin, modelled as a pair of contact surfaces in
ABAQUS. For pure fatigue loading, a triangular waveform was used and frequencies varied
from 0.001 to 2.5Hz. For dwell loading, a trapezoidal waveform was used with a 1-X-1-1
pattern, where X is the dwell time and varies from 1 to 1000s [11, 12].
156
Figure 6. 1:(a) Finite element mesh for a CT specimen and (b) refined mesh for the crack tip
area (element size is 6.35μm).
6.2.3. NEAR-TIP INELASTIC DEFORMATION
Figure 6.2 shows the contour plot of the accumulated inelastic strain obtained from the finite
element analyses after ten loading cycles at an applied frequency of . The
accumulated inelastic deformation is highly localized in the vicinity of the crack tip. Figure 6.3
shows the contour plot of the accumulated inelastic strain after ten loading cycles for an
imposed dwell time of . It can be seen that imposed 1000s dwell period introduces
more severe inelastic deformation (a maximum value of 0.892) near the crack tip when
compared to that (a maximum value of 0.732) for pure fatigue (Figure 6.2). For a given
157
number of cycles, the lower the frequency or the longer the dwell period, the more severe
the accumulated inelastic deformation near the crack tip.
Figure 6. 2: Contour plot of the accumulated inelastic strain near the crack tip after ten
loading cycles for a stress intensity factor range ΔK=30MPa√m and a loading frequency
f=0.25Hz2.
2 In Figures 6.2 and 6.3, different contour colours in a single finite element represent different values obtained
at four different integration points using ABAQUS [18].
158
Figure 6. 3: Contour plot of the accumulated inelastic strain near the crack tip after ten
loading cycles for a stress intensity factor range ΔK=30MPa√m and a dwell period of 1000s.
6.2.4. PREDICTION OF CRACK GROWTH
Similar to the work of Zhao et al. [10], the accumulated inelastic strain has been used here as
a criterion to predict crack growth rate in CT specimens at 650C for selected loading
conditions. The choice of accumulated inelastic strain seems justified in that, under fatigue
loadings, the reversed deformation during unloading is accounted for as well as that during
loading. In the FE analyses, the number of cycles was recorded when the inelastic strain
accumulation over the characteristic distance d* = 6.35m ahead of the crack tip reached a
critical value of 0.35. This critical value 0.35 was back calculated using finite element analysis,
159
which predicts the same crack growth rate (2.72107 m/cycle) as the experimental result in
Dalby and Tong [11, 12] for a base-line fatigue test ( f = 2.5 Hz and R = 0.1)
at 650C. The average crack growth rate was then calculated by dividing the characteristic
distance with the recorded number of cycles.
The predicted effects of loading frequency on crack growth rates are shown in Figure 6.4 for
a triangular waveform with and load ratio R = 0.1, including the test data
for CT specimens at 650C [11, 12]. The predictions showed the same trend as the
experimental data, i.e., the increase of crack growth rate with the decrease of loading
frequency, due to the enhanced crack tip deformation for lower frequencies. The predictions
agree well with the test data for higher frequency region (f > 0.1Hz), while marked difference
is noted in the lower frequency region (f < 0.01Hz). This difference may be due to the
oxidation effect at high temperature, which was not taken into account in the model
prediction.
The effects of dwell period on crack growth rates are predicted by considering a trapezoidal
loading waveform (1-X-1-1 pattern) with , baseline frequency f = 0.25 and
load ratio R = 0.1. The predictions are presented in Figure 6.5, including the experimental
data [11, 12] for CT specimens at 650C under the same loading conditions. Again the
predictions show the same trend as the experimental data, i.e., the increase of crack growth
rate with the increase of hold period. On the other hand, the predictions show an increasing
difference between the prediction and the experimental data as the dwell period is
increased, especially when the dwell period is over 60 seconds. Again, this difference may be
160
due to the oxidation effect at high temperature, which is not accounted for in the current
model prediction.
Figure 6. 4: Effects of loading frequency on crack growth rate for a triangular loading
waveform with ΔK = 30MPa√m and load ratio R = 0.1 at 650°C, comparison of model
prediction and experimental results [11, 12].
0.00E+00
5.00E-07
1.00E-06
1.50E-06
2.00E-06
2.50E-06
0.001 0.01 0.1 1 10
da/
dN
(m
/cyc
le)
Frequency (Hz)
Prediction - accumulated inelastic strain
Experimental data
161
Figure 6. 5: Effects of dwell periods on crack growth rate for a trapezoidal loading
waveform with ΔK = 30MPa√m and load ratio R = 0.1 at 650°C, comparison of model
prediction and experimental results [11, 12].
The detrimental effects of environmental factors, in particular oxygen, on high-temperature
crack growth behaviour in nickel alloys have been well demonstrated by experimental results
[19, 20]. Oxidation-accelerated crack growth is a direct consequence of oxygen diffusion into
the material, which attacks the grain boundary through chemical reaction with alloy
elements or oxygen segregation at grain boundaries. The diffusion process becomes more
significant with the assistance of high tensile stress near crack tip [21], especially for
superimposed dwell periods at the maximum load where the crack tip is fully open and
experiences the highest tensile stress level. As a consequence, a coupled mechanical-
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1 10 100 1000
da/
dN
(m
/cyc
le)
Dwell time [s]
Prediction - accumulated inelastic strain
Experimental data
162
diffusion analysis [22] is required to account for the effects of oxygen diffusion as well as that
of viscoplasticity on crack growth.
6.3. COUPLED VISCOPLASTICITY-OXYGEN DIFFUSION ANALYSIS
Here we are concerned with the diffusive transport of oxygen into the crack tip under the
influence of a mechanical loading field. The analyses follows exactly the same method as
described in Chapter 5, where the gradient of hydrostatic stress (or pressure) is considered to
be the driving force for the stress-assisted diffusion process.
To solve the overall fatigue-oxidation problem, a sequentially coupled mechanical-diffusion
analysis was performed to include the dependence of oxygen diffusion on the stress state.
The mechanical analyses were carried out first, with the history of hydrostatic stress for each
integration point stored in the ABAQUS output files. During the diffusion analysis, the output
files were then used as inputs, to include the effect of hydrostatic stress gradient on oxygen
diffusion. Although the availability of oxygen may vary in the vicinity of the crack tip, as does
the oxygen partial pressure, the oxygen intake on the crack surface was assumed to be
uniform and follows the Equation (3) in Chapter 5.
The diffusion parameters were already obtained in Chapter 5 from the FE analyses
performed on the waisted specimen and, at 650°C, they are found to be
= 1.67×10-13mg2/mm4/s for the parabolic rate constant, D = 1.085×10-13mm2/s for the
diffusivity and M = 1.585×10-3mm2/N for the pressure factor.
163
6.3.1. NEAR-TIP OXYGEN CONCENTRATION
Figure 6.6(a) shows the contour plots of oxygen concentration near a fatigue crack tip after
10 loading cycles, where the stress intensity factor range , load ratio R = 0.1
and loading frequency f = 0.001Hz. Figure 6.6(b) shows the contour plot of oxygen
concentration after 10 cycles for the 1000s dwell loading situation, where the dwell is
superimposed at peak loads of a 0.25Hz base line fatigue waveform. Both results show the
penetration of oxygen into the crack tip, as well as the crack flanks, due to oxygen diffusion
process. It is also noted that the oxygen concentration level ahead of the crack tip is less
severe compared to the one on the fracture surface, which may be due to the limited oxygen
availability at the crack tip. This is consistent with the FIB experimental examinations in
Chapter 4, where oxides are mainly observed on fracture surfaces. Also, for a given number
of cycles, the lower the frequency or the longer the dwell period, the more severe the near-
tip oxygen concentration becomes.
164
Figure 6. 6: (a) Contour plot of the oxygen concentration near the crack tip after 10 cycles
for a frequency of 0.001Hz.
(a)
Crack tip
165
Figure 6. 6: (b) Contour plot of the oxygen concentration near the crack tip after 10 cycles
for a dwell period of 1000s.
6.3.2. FAILURE CURVE
In order to predict the crack growth rates under fatigue-oxidation conditions, it is necessary
to formulate an appropriate criterion to account for the contributions from both mechanical
deformation and oxidation. Two parameters, the accumulated inelastic strain and oxygen
concentration, have been used to represent the viscoplastic deformation and the oxidation
effect, respectively. The idea is to use the two parameters to construct a failure curve from
which crack growth prediction under fatigue-oxidation may be made. Experimental crack
growth data for selected loading frequencies at [11, 12] were utilised to
determine the number of cycles required to grow the crack in the FE model. Specifically, the
(b)
Crack tip
166
number of cycles was calculated by dividing the near-tip element size (6.35µm) with the
measured crack growth rate. FE analyses were then carried out for the calculated number of
cycles, i.e., just before the crack grows into the next element, and the values of the two
parameters were extracted. The extracted values of the two parameters were then plotted
against each other to obtain a failure curve. The results (give in Table 6.1) are presented in
Figure 6.7, where all the data seem to follow the fitted failure curve. As shown in Figure 6.7,
the failure curve indicates a "safe zone" outside which the crack will grow.
Table 6. 1: The accumulated inelastic strain and oxygen concentration obtained from the FE
analyses for different frequencies (ΔΚ=30ΜPa√m, R=0.1, T=650°C).
Frequency (Hz) Number of cycles Acc. inelastic strain Oxygen con. (mg/mm3)
0.001 3.05E+00 0.095539 9.23565E-04
0.0033 3.30E+00 0.097097 5.33232E-04
0.01 5.21E+00 0.130146 3.78586E-04
0.1 9.38E+00 0.189051 1.57638E-04
0.25 1.17E+01 0.217812 1.10645E-04
1 1.60E+01 0.265895 6.4548E-05
2.5 2.33E+01 0.35038 4.95818E-05
167
Figure 6. 7: Failure curve in terms of the accumulated inelastic strain and the oxygen
concentration near the crack tip (ΔΚ=30ΜPa√m, R=0.1, T=650°C).
6.3.3. PREDICTION OF CRACK GROWTH UNDER FATIGUE-OXIDATION
CONDITION
Using the failure curve constructed above, crack growth rates in a CT specimen at 650C
were predicted from the finite element analyses for selected loading frequencies and dwell
time at . During the analyses, the accumulated inelastic strain and the
oxygen concentration over the characteristic distance d* = 6.35m ahead of the crack tip
were recorded against the number of cycles. The results were then used to obtain the
number of cycles for the crack to grow, i.e., when the combination of the accumulated
inelastic strain and the oxygen concentration reached the failure curve. The average crack
growth rate was then calculated by dividing the characteristic distance with the number of
168
cycles. The procedures have been illustrated in Figures 6.8 and 6.9, and the obtained
numbers of cycles for the crack to grow are given in Tables 6.2 and 6.3.
Table 6. 2: The obtained number of cycles for crack to grow for different frequencies.
Frequency (Hz) Critical number of cycles
0.001 3
0.0033 4
0.01 5
0.1 9
0.25 11
1 16
2.5 17.5
169
Figure 6. 8: Illustration of determining the number of cycles for the crack to grow, i.e.,
when the combination of the accumulated inelastic strain and the oxygen concentration
reached the failure curve, for different frequencies.
170
Figure 6. 9: Illustration of determining the number of cycles for the crack to grow, i.e.,
when the combination of the accumulated inelastic strain and the oxygen concentration
reached the failure curve, for different dwell time.
Table 6. 3: The obtained number of cycles for crack to grow for different dwell time.
Dwell time (s) Critical number of cycles
1 10
20 6.5
60 5.25
120 4.5
300 3.5
1,000 2.75
171
The predicted effects of loading frequency on crack growth rates are also shown in Figure
6.10 for a Κ , in a direct comparison with test data for CT specimens at 650C
[11, 12]. For the predictions based on fatigue-oxidation failure curve, excellent agreements
were achieved for all loading frequencies.
The predicted effects of dwell time on crack growth rates is also shown in Figure 6.11 for
, in a direct comparison with test data for CT specimens at 650C [11, 12]. The
predictions from accumulated inelastic strain show increased deviations from the
experimental data as the dwell period is increased, especially when a dwell period is over 60
seconds. Again, this difference is much reduced when the fatigue-oxidation failure curve was
used for the prediction.
172
Figure 6. 10: The effects of loading frequency on the crack growth rate for a triangular
waveform with ΔK=30MPa√m and load ratio R = 0.1 at 650°C; comparison of the model
prediction and the experimental results [11, 12].
1.00E-07
1.00E-06
1.00E-05
0.001 0.01 0.1 1 10
da/
dN
(m
/cyc
le)
Frequency (Hz)
Experimental data
Prediction - Accumulated inelastic strain
Prediction - Fatigue-oxidation
173
Figure 6. 11: The effects of dwell period on the crack growth rate for a trapezoidal loading
waveform with ΔK=30MPa√m and load ratio R = 0.1 at 650°C; comparison of the model
prediction and the experimental results [11, 12].
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1 10 100 1000
da/
dN
(m
/cyc
le)
Dwell (s)
Experimental data
Prediction - Accumulated inelastic strain
Prediction - Fatigue-oxidation
174
6.4. DISCUSSION
For polycrystalline nickel alloys under low cycle fatigue at elevated temperature above
500C, a transition from cycle-dependent transgranular crack propagation to time-dependent
intergranular crack propagation may occur, when the frequency is sufficiently low and/or the
hold period at peak loads is sufficiently long, resulting in significant increase in crack growth
rates. As frequency or cycle duration does not seem to influence the crack propagation rate
at low temperatures or under vacuum conditions, oxygen from the environment may be the
key factor for the time-dependent crack propagation behaviour, which is attributed to the
oxidation reaction at grain boundaries in most studies. For instance, Miller et al. [19, 20]
proposed that the formation of thin intergranular Nb2O5 films, which fracture in a brittle
manner, leads to high crack propagation rates. This was also supported by studies on Ni-
18Cr-18Fe and powder-metallurgical superalloys with varying Nb concentrations, where the
material’s sensitivity to environmental attack depends solely on the Nb content. On the other
hand, strong environmental effects on fatigue crack propagation were also observed by
Molins et al. [2, 4] in Ni-Cr model alloys without any Nb. Molins et al. [4] proposed that
epitactic NiO-formation ahead of the crack tip introduced the vacancies into near-tip
materials, which act as nucleation sites for pores and microcracks and thereby give rise to
high crack propagation rates. High Cr contents were suggested to bring beneficiary effects
due to the much slower kinetics of Cr2O3 formation than that of NiO formation, and, in
addition, Cr could scavenge O2 near the crack tip and reduces the level of elemental O
required for oxidation reaction [4].
175
The oxidation-reaction mechanism has been challenged by the dynamic embrittlement
hypothesis in recent studies of crack growth in nickel alloys under fatigue-oxidation
conditions [5, 7]. Dynamic embrittlement was introduced to define a time-dependent
intergranular brittle cracking process in which the supply of embrittling elements is due to
grain boundary diffusion into an area of increasing tensile stress ahead of a crack tip [21, 22].
The embrittling elements lower the cohesion of the grain boundary and facilitate the crack
advance. The case of hydrogen embrittlement in steels is one of the most important damage
mechanisms that fit in the dynamic embrittlement scheme [23]. Other established dynamic
embrittlement cases include: sulphur-induced stress relief cracking in steels [22], S-induced
cracking in Cu Cr alloys [24], O-induced cracking in Cu Be alloys [25] and Ni3Al [21], and Sn-
induced cracking in Cu-Sn model alloys [22, 26].
Dynamic embrittlement for nickel superalloys has been supported by experimental
observations in Inconel 718 for four-point bending tests [5-7]. First, the individual grain
boundary facets were found to be smooth on the sub-micrometer level and oxidation of NbC
precipitates was only observed on the fractured surface where the crack front had passed.
Second, the cracking process was found to be extremely sensitive to rapid changes in oxygen
pressure. Crack growth could be turned off and on in less than 10s by alternately pumping
oxygen from the chamber and re-admitting it. Given the fact that oxidation kinetics for nickel
alloys at 650°C is extremely slow [27], there is insufficient time for oxide formation to happen
ahead of the crack tip within the order of seconds. Direct observations of internal oxidation
and associated damage were only available on smooth specimens, which have been exposed
to oxidizing environment for a considerably long time, mostly greater than 20 hours [8, 28-
176
30]. The current work also showed that, the crack growth rates in air at high-temperature,
can be predicted well from the dynamic embrittlement hypothesis, though collecting
experimental evidence of crack tip oxygen diffusion and measurements of oxygen
concentration near the crack tip, are challenging tasks and yet to be accomplished.
The construction of a failure curve and associated crack growth predictions, are based on the
values of the accumulated inelastic strain and the oxygen concentration averaged over the
element just ahead the crack tip, which corresponds to a characteristic distance of 6.35µm
ahead of the crack tip. The need to specify a critical distance in association with a fracture
criterion is well established [31-40], particularly for sharp cracks, where, due to the stress
and strain singularities, numerical results are non-unique unless a critical distance is
specified. McClintock [32, 33] developed a mechanistic analysis of crack extension by fatigue,
employing a failure criterion based on plastic strain accumulation over a “structural size” of
the material. Quantitatively, a structural size of 5m seems to fit well with the experimental
results for three types of materials [33]. Krafft [34] proposed that ductile fracture would
occur when the instability strain for a uniaxial tensile specimen is achieved at the crack tip
over a distance equal to the spacing of void nucleating particles. Ritchie et al. [35]
successfully predicted fracture when the critical stress was achieved over two grain sizes.
Rawal and Gurland [36] reported a critical stress was achieved over a critical distance about
1.3 grain-sizes. Chen et al. [38] reported a characteristic distance of 13m for crack tip
fracture in C-Mn steel, over which the critical strain, the critical stress and the critical stress
triaxiality criteria are satisfied simultaneously. Said and Tasgetiren [39] used the
characteristic distance of 0.8-1.37 mean grain size to determine the static and dynamic
177
fracture toughness for bcc metals and alloys using the critical stress criterion. In the present
work, the chosen characteristic distance 6.35µm is on the similar order to the others [32-36,
38, 39] and it is about the average grain size, where crack growth is assumed to be associated
with the cracking of a grain. Our mesh sensitivity study showed that the strain distribution
ahead of the crack tip has reached a good convergence at d* μ [17].
6.5. CONCLUSIONS
In this chapter, the diffusion parameters obtained in Chapter 5 from the FE analyses of
waisted specimen were used to model near-tip oxygen diffusion process and predict the
oxidation-assisted crack growth in CT specimens for different frequencies and dwell times.
The FE analyses were carried out for selected loading frequencies to construct a failure curve
based on the accumulated inelastic strain versus the oxygen concentration near the crack tip.
The failure curve was then used to predict the critical number of cycles for crack to grow at
selected loading frequencies and dwell time.
Simulations showed that the crack growth rates predicted from the fatigue-oxidation failure
curve compare well with the experimental results, with marked improvements achieved over
those predicted from the viscoplastic model alone.
178
6.6. REFERENCES
[1]. Andrieu, E., Molins, R. (1992). Intergranular crack tip oxidation mechanism in a nickel-
based superalloy. Materials Science and Engineering: A, 154, 21-28.
[2]. Ghonem, H., Zheng, D. (1992). Depth of intergranular oxygen diffusion during
environment-dependent fatigue crack growth in alloy 718. Materials Science and Engineering:
A, 150(2), 151-160.
[3]. Andrieu, E., Hoshstetter, G., Molins, R., Pineau, A. (1997). Oxidation and intergranular
cracking behaviour of two high strength Ni-base superalloys. Magnin T, editor. Corrosion-
Deformation Interactions (EFC 21), p. 461-475.
[4]. Molins, R., Hochstetter, G., Chassaigne, J.C. and Andrieu, E. (1997). Oxidation effects on
the fatigue crack growth behaviour of alloy 718 at high temperature. Acta Materialia, 45(2),
663-674.
[5]. Pfaendtner, J.A., McMahon Jr., C.J. (2001). Oxygen-induced intergranular cracking of a
Ni-base alloy at elevated temperatures—an example of dynamic embrittlement. Acta
Materialia, 49(16), 3369-3377.
[6]. Krupp, U., Kane, W.M., Laird, C., McMahon Jr., C.J. (2004). Brittle intergranular fracture
of a Ni-base superalloy at high temperatures by dynamic embrittlement. Materials Science
and Engineering A, 387-389, 409-413.
[7]. Krupp, U. (2005). Dynamic embrittlement: time-dependent quasi-brittle intergranular
fracture at high temperatures. International Materials Reviews, 50(2), 83-97.
[8]. Calvarin-Amiri, G., Huntz, A.M., Molins, R. (2001). Effects of an applied stress on the
growth kinetics of oxide scales formed on Ni-20Cr alloys. Materials at High Temperatures,
18(2), 91-99.
[9]. Bika, D., McMahon Jr, C.J. (1995). A model for dynamic embrittlement. Acta Metallurgic
Materialia, 43, 1909-1916.
[10]. Zhao, L.G., Tong, J. and Hardy, M.C. (2010).Prediction of crack growth in a nickel-based
superalloy under fatigue-oxidation conditions. Engineering Fracture Mechanics, 77(6), 925-
938.
[11]. Dalby, Sara. (2002). The effects of frequency and waveshape on the fatigue crack
growth of an advanced nickel based superalloy at elevated temperatures. PhD Thesis.
University of Portsmouth, Portsmouth, UK.
179
[12]. Dalby, S. Tong, J. (2005). Crack growth in a new nickel-based superalloy at elevated
temperature, Part I: Effects of loading waveform and frequency on crack growth. Journal of
Materials Science, 40(5), 1217-1228.
[13]. Chaboche, J.L. (1989). Constitutive equations for cyclic plasticity and cyclic
viscoplasticity. International Journal of Plasticity, 5(3), 247-302.
[14]. Zhan, Z.-L. (2004). A study of cyclic plasticity and viscoplasticity in a new nickel-based
superalloy using unified constitutive equations. PhD Thesis. University of Portsmouth.
Portsmouth, U.K.
[15]. Zhan, Z.-L., Tong J. (2007). A study of cyclic plasticity and viscoplasticity in a new nickel-
based superalloy using unified constitutive equations. Part I: evaluation and determination of
material parameters. Mechanics of Materials, 39, 64-72.
[16]. Zhan, Z.-L., Tong J. (2007). A study of cyclic plasticity and viscoplasticity in a new nickel-
based superalloy using unified constitutive equations. Part II: simulation of cyclic stress
relaxation. Mechanics of Materials, 39, 73-80.
[17]. Zhao, L.G. and Tong, J. (2008). A viscoplastic study of crack tip deformation and crack
growth in a nickel based superalloy at elevated temperatures. Journal of the Mechanics and
Physics of Solids, 56(12) , 3363-3378 .
[18]. ABAQUS. Version 6.8. (2009). Providence, USA. Dassault Systemes Simulia Corp.
[19]. Miller, C.F., Simmons, G.W. and Wei, R.P. (2001). Mechanism for oxygen enhanced
crack growth in inconel 718. Scripta Materialia, 44(10), 2405-2410.
[20]. Miller, C.F., Simmons, G.W. and Wei, R.P. (2003). Evidence for internal oxidation during
oxygen enhanced crack growth in P/M Ni-based superalloys. Scripta Materialia, 48(1), 103-
108.
[21]. Liu, C.T., White, C.L. (1987). Dynamic embrittlement of boron-doped Ni3Al alloys at
600°C. Acta Metallurgica, 35(3), 643-649.
[22]. Bika, D. (1992). Dynamic embrittlement in metallic alloys. PhD Thesis. University of
Pensylvania, Philadelphia, PA, U.S.A.
[23]. Birnbaum, H.K., Sofronis, P. (1994). Hydrogen-enhanced localized plasticity—a
mechanism for hydrogen related fracture. Materials Science and Engineering: A, 176(1-2),
191-202.
180
[24]. Misra, R.D.K., Prasad, V.S. (2000). On the dynamic embrittlement of copper-chromium
alloys by sulphur. Journal of Materials Science, 35(13), 3321-3325.
[25]. Misra, R.D.K., McMahon Jr., C.., Guha, A. (1994). Brittle behavior of a dilute copper-
beryllium alloy at 200°C in air. Scripta Metallurgica et Materialia, 31(11), 1471-1474.
[26]. Bika, D., Pfaendtner, Menyhard, M., J.A., McMahon Jr., C.J. (1995). Sulfur-induced
dynamic embrittlement in a low-alloy steel. Acta Metallurgica e Materialia, 43(5), 1895-1908.
[27]. Krupp, U., Kane, W., Pfaendtner, J.A., liu, X., laird, C., McMahon Jr., C.J. (2004).
Oxygen-induced intergranular fracture of nickel-base alloy IN718 during mechanical loading
at high temperature. Materials Research, 7(1), 35-41.
[28]. Bricknell, R.H., Woodford, D.A. (1982). The mechanism of cavity formation during high
temperature oxidation of nickel. Acta Metallurgica, 30(1), 257-264.
[29]. Garat, V., Brucelle, O., Clouè, J.M., Rebeyrolle, V., Monceau, D., Viguier, B., Andrieu, E.
(2004). Comparing different methods to dtermine the intergranular oxidation damage on a
nickel based superalloy. Materials Science Forum, 461-464, 537-544.
[30]. Encinas-Oropesa, A., Drew, G.L., Hardy, M.C., Leggett, A.J., Nicholls, J.R., Simms, N.J.
(2008). Effects of oxidation and hot corrosion in a nickel disc alloy. Champion, Pennsylvania,
USA : Wiley, 2008. Superalloys 2008. pp. 609-618.
[31]. Riedel H. and Rice J.R. (1980). Tensile cracks in Creeping Solids. Fracture Mechanics:
Twelfth Conference, ASTM STP 700, 112–130.
[32]. McClintock F.A. (1958). Ductile instability in shear. ASME J Appl. Mech. 25, 582-588.
[33]. McClintock F.A, Irwin G.R. (1965). Plasticity aspects of fracture mechanics. Fracture
Toughness Testing and its Applications, ASTM STP 381, 84-113.
[34]. Krafft J.M. (1964). Crack toughness and strain hardening of steels. Applied Materials
Research, 3, 88-101.
[35]. Ritchie, R.O., Knott, J.F., Rice, J.R. (1973). On the relationship between critical tensile
stress and fracture toughness in mild steel. Journal of the Mechanics and Physics of Solids,
21(6), 395-410.
[36]. Rawal, S.P., Gurland, J. (1977). Observations on the effect of cementite particles on the
fracture toughness of spheroodized carbon steels. Metallurgical and Materials Transactions A,
8(5), 691-698.
181
[37]. Neville, D.J. (1988). On the distance criterion for failure at the tips of cracks, minimum
fracture toughness, and non-dimensional toughness parameters. Journal of the Mechanics
and Physics of Solids, 36(40), 443–457.
[38]. Chen, J.H., Wang, Q., Wang, G.Z., Li, Z. (2003). Fracture behavior at crack tip—A new
framework for cleavage mechanism of steel. Acta Materialia, 51(3), 1841-1855.
[39]. Said, G., Tasgetiren, S. (2004). An express technique for the determination of static and
dynamic fracture toughness (KIC, Kid) of bcc metals and alloys. Mechanics of Materials, 36(11),
1129-1142.
[40]. Neimitz, A., Graba, M., Galkiewicz, J. (2007). An alternative formulation of the Ritchie-
Knott-Rice local fracture criterion. Engineering Fracture Mechanics, 74(8), 1308-1322.
182
7. CONCLUSIONS & FURTHER WORK
7.1. CONCLUSIONS
Formations of surface oxide scales (Cr, Ti, Ni and Co oxides) and micro-voids beneath the
oxide scales are the major mechanisms for oxidation damage of alloy RR1000 with or
without fatigue loading. Oxidation damage becomes more severe with the increase of
temperature. The internal voids, created by the outwards diffusion of alloy elements to
form the oxide scale on the specimen surface, could lead to crack initiation and
subsequent crack propagation under mechanical loading condition.
Cyclic stress was shown to have an influence on oxidation damage for temperatures at
750C and above, and the thickness of oxide scale and the depth of internal damage
were enhanced by the increase of cyclic stress level. This has been confirmed from the
FIB measurements of the thickness of oxide scale and the depth of internal damage. In
addition, both EDX analyses and FIB secondary ion imaging showed the presence of
internal oxides, indicating the oxygen penetration into the material.
The SEM studies of the fracture surfaces in dwell crack growth confirmed the transition
from transgranular to predominant intergranular cracking in alloy RR1000 for increased
dwell time. This change in fracture mode supports the oxidation-assisted crack growth
mechanism via grain boundary embrittlement, considering that creep deformation is
extremely limited for alloy RR1000. Oxidation has also been evidenced by the FIB
analyses of both CT and CN specimens which revealed the oxides accumulated on the
fracture surface.
183
Finite element analyses of oxygen diffusion along the grain boundaries have been carried
out for a waisted specimen using a submodelling technique. For natural diffusion,
oxygen penetration and concentration along grain boundaries are controlled by the
diffusivity, which is temperature dependent. The higher the temperature, the deeper the
oxygen penetration. Fatigue-assisted penetration of oxygen was modelled for both
750°C and 800°C, and the simulation results have been used to obtain the pressure
factor for modelling the oxidation-assisted fatigue crack growth.
Oxygen diffusion process, coupled with viscoplasticity, near a crack tip has been
modelled using the diffusion parameters obtained from the FE analyses of the waisted
specimen. A failure curve for crack growth under fatigue-oxidation condition has been
constructed based on the consideration of both the accumulated inelastic strain and the
oxygen concentration near the crack tip. Crack growth rates, predicted from the failure
curve for selected frequencies and dwell periods, compare well with the experimental
results, with marked improvements achieved over those predicted from the viscoplastic
model alone.
7.2. FURTHER WORK
It is suggested to carry out fatigue-oxidation experiments on a notched specimen and
examine the effects of fatigue on oxidation damage near the notch area, as the
stress/strain field near the notch area resembles that near a crack tip. Waisted specimen
belongs to the smooth specimen family, and could not properly represent the high stress
triaxiality presented near the crack tip, which might influence the oxidation process
significantly.
184
FIB examinations of fracture surface oxides could not support the oxidation-based
rationale to explain the observed tunnelling phenomenon in dwell crack growth in CN
specimens, as there seems no difference in oxides accumulation between the middle
section and the free surface for a given temperature. Further and in-depth investigation
is required in order to understand the complex mechanisms for crack growth tunnelling.
For simplicity, oxygen diffusion approach has been used in the preset work to quantify
the oxidation damage of alloy RR1000 under fatigue. No attempts have been made to
model the chemical reaction kinetics, vacancy formation, transport of other species and
recrystallization, which are the mechanisms physically operative during the oxidation
process of RR1000. Further work is undergoing to address these issues.
The crystal plasticity formulation describes grain interior deformation without
consideration of grain boundary deformation. Grain boundary deformation is controlled
mainly by sliding and diffusion, which can be described by constitutive laws stating the
evolution of the normal and shear tractions in terms of both normal and tangential
displacements along the grain boundaries. Cohesive element approach has been
generally used to model grain boundary deformation, where the tensile and shear
tractions of the cohesive elements embedded along the grain boundaries can be
calculated. Further work is required to address the influence of grain boundary
deformation on oxygen diffusion.
The current work showed that, the crack growth rates in air at high-temperature, can be
predicted well from the dynamic embrittlement hypothesis (stress-assisted oxygen
diffusion into the crack tip). However, so far, there have been no direct evidence and
185
measurements of oxygen ahead of the crack tip, probably due to very limited level of
oxidation into the crack tip as well as the insufficient resolution of current facilities.
These challenging tasks are yet to be accomplished in further work.
186
APPENDIX A
MEASUREMENTS FROM THE FIB IMAGES
187
APPENDIX B
THE J INTEGRAL
In modelling material failure various energy methods are used which provide an insight into
fracture. For 2-D domains containing cracks, Rice in 1968 presented the J-integral.
Considering a 2-D linear body of linear or nonlinear elastic material free of body forces and
subjected to a 2-D deformation field (plane strain, plane stress) so that all stresses σij depend
only on two Cartesian coordinates (x, y). Suppose the body contains an edge crack as shown
in Figure B1.
Define the strain-energy density W as:
where ε=(εij) is the infinitesimal strain tensor. Now, define the J-integral as
As shown in Figure B1, Γ is the curve surrounding the notch tip; the integral being evaluated
counter clockwise starting from the lower surface and continuing along the path Γ to the
upper surface. In the J-integral equation, T is the traction vector defined according to the
outward normal along Γ, u is the displacement vector and ds is an element arc length along Γ.
Rice (1968) proved the path independent concepts and found that for small-scale yielding the
stress energy release rate G is equal to the J-integral. Therefore, the stress intensity factor
can be evaluated as follows:
188
where
for plane strain and for plane stress.
Figure B1: Crack tip coordinate system and typical line integral contour.
References:
Riveros, G.A. (2006). Numerical Evaluation of Stress Intensity Factors (KI) J-Integral Approach.
ERDC/CHL CHETN-IX-16. US Army Corps of Engineers.
Rice, J.R. (1968). A path independent integral and the approximate analysis of strain
concentrations by notches and cracks. Journal of Applied Mechanics. 35. 379-386.