Teresa Palacios Garcia 1 PDF

Universidad Politécnica de Madrid
Escuela Técnica Superior de Ingenieros de Caminos,

Canales y Puertos
Departamento de Ciencia de Materiales
Microstructural and Mechanical

Characterization of Tungsten based
Materials for Fusion Reactors
Tesis Doctoral
Teresa Palacios Garcı́a
Ingeniera de Materiales
Director de la tesis:
José Ygnacio Pastor Caño
Doctor en Ciencias Fı́sicas
Catedrático de Universidad
2015
Tribunal nombrado por el Magfco. y Excmo. Sr. Rector de la Universidad
Politécnica de Madrid, el dı́a ...... de .............. de 2015.
Presidente: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Vocal: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Vocal: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Vocal: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Secretario: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Suplente: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Suplente: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Realizado el acto de defensa y lectura de la Tesis el dı́a ..... de ............. de 2015 en la

E.T.S. de Ingenieros de Caminos, Canales y Puertos de la Universidad Politécnica
de Madrid.
Calificación: .................................................
EL PRESIDENTE LOS VOCALES
EL SECRETARIO
A mis padres
Abstract
Tungsten (W) and tungsten-based alloys are considered to be the best candidate
materials for fabricating the divertor in the next-generation nuclear fusion reactors.
This component will experience the highest thermal loads during the operation
of a reactor since it directly faces the plasma. In recent years, after thorough
analysis that followed a strategy of cost reduction, the ITER Organization decided
to built a full-tunsgten divertor before the first nuclear campaigns. Therefore,
tungsten will be used not only as a plasma-facing material (PFM) but also
in structural applications. Tungsten, due to its the excellent thermo-physical
properties fulfils the requirements of a PFM, however, its use in structural
applications is compromised due to its inherent brittleness. One of the objectives of
this phD thesis is therefore, to find a material with improved brittleness behaviour.
The microstructural and mechanical characterisation of different

tunsgten-based materials was performed. However, this is a challenging task
because of the reduced laboratory-scale size of the specimens provided, their fine
microstructure and their brittleness. Consequently, many techniques are required
to ensure an accurate measurement of all the mechanical and physical properties.
Some of the applied methods have been widely used such as nanoindentation
or three-point bending (TPB) tests. However, other methods were specifically
developed and implemented during this work such as the measurement of the
real fracture toughness of bulk-tunsgten alloys or the in situ fracture toughness
measurements of very thin tungsten foils.
I
Bulk-tunsgten materials with different compositions (W-1% Y2 O3 , W-2% V-
0.5% Y2 O3 , W-4% V-0.5% Y2 O3 , W-2% Ti-1% La2 O3 and W-4% Ti-1% La2 O3 )
were studied and compared with pure tungsten processed under the same
conditions. These alloys, produced by a powder metallurgical route of mechanical
alloying (MA) and hot isostatic pressing (HIP), were microstructural and
mechanically characterised from 77 to 1473 K in air and under high vacuum
conditions. Hardness, elastic modulus, flexural strength and fracture toughness
for all of the alloys were measured in addition to other physical and mechanical
properties. Finally, the fracture surfaces after the TPB tests were analysed to
correlate the micromechanisms of failure with the macroscopic behaviour. The
results reveal brittle mechanical behaviour in almost the entire temperature
range for the alloys and micromechanisms of failure with no improvement in the
ductile-brittle transition temperature (DBTT).
To continue the search of a tungsten material with lowered DBTT, a

preliminary study of pure tunsgten and 0.005 wt.% potassium (K)-doped tungsten
foils was also performed. These foils were industrially produced by sintering and
hot and cold rolling. After that, they were annealed from 1073 to 2673 K to analyse
the evolution of the microstructural and mechanical properties with increasing
annealing temperature. The results revealed the stabilisation of the tungsten
grains with increasing annealing temperature in the potassium-doped tungsten foil.
However, additional studies need to be performed to gain a better understanding
of the microstructure and mechanical properties of these materials such as fracture
toughness.
II
Resumen
El wolframio (W) y sus aleaciones se consideran los mejores candidatos para la

construcción del divertor en la nueva generación de reactores de fusión nuclear.
Este componente va a recibir las cargas térmicas más elevadas durante el
funcionamiento del reactor ya que estará en contacto directo con el plasma. En
los últimos años, después de un profundo análisis y siguiendo una estrategia de
reducción de costes, la Organización de ITER tomó la decisión de construir el
divertor ı́ntegramente de wolframio desde el principio. Por ello, el wolframio no sólo
actuará como material en contacto con el plasma (PFM), sino que también tendrá
aplicaciones estructurales. El wolframio, debido a sus excelentes propiedades
termo-fı́sicas, cumple todos los requerimientos para ser utilizado como PFM, sin
embargo, su inherente fragilidad pone en peligro su uso estructural. Por tanto, uno
de los principales objetivos de esta tesis es encontrar una aleación de wolframio
con menor fragilidad.
Durante éste trabajo, se realizó la caracterización microstructural y mecánica

de diferentes materiales basados en wolframio. Sin embargo, ésta tarea es un reto
debido a la pequeña cantidad de material suministrado, su reducido tamaño de
grano y fragilidad. Por ello, para una correcta medida de todas las propiedades
fı́sicas y mecánicas se utilizaron diversas técnicas experimentales. Algunas de ellas
se emplean habitualmente como la nanoindentación o los ensayos de flexion en
tres puntos (TPB). Sin embargo, otras fueron especificamente desarrolladas e
implementadas durante el desarrollo de esta tesis como es el caso de la medida
real de la tenacidad de fractura en los materiales masivos, o de las medidas in situ
de la tenacidad de fractura en las láminas delgadas de wolframio.
III
Diversas composiciones de aleaciones de wolframio masivas (W-1% Y2 O3 ,
W-2% V-0.5% Y2 O3 , W-4% V-0.5% Y2 O3 , W-2% Ti-1% La2 O3 y
W-4% Ti-1% La2 O3 ) se han estudiado y comparado con un wolframio puro
producido en las mismas condiciones. Estas aleaciones, producidas por ruta
pulvimetalúrgica de aleado mecánico (MA) y compactación isostática en caliente
(HIP), fueron microstructural y mecánicamente caracterizadas desde 77 hasta
1473 K en aire y en alto vacı́o. Entre otras propiedades fı́sicas y mecánicas se
midieron la dureza, el módulo elástico, la resistencia a flexión y la tenacidad de
fractura para todas las aleaciones. Finalmente se analizaron las superficies de
fractura después de los ensayos de TPB para relacionar los micromecanismos
de fallo con el comportamiento macroscópico a rotura. Los resultados obtenidos
mostraron un comportamiento mecánico frágil en casi todo el intervalo de
temperaturas y para casi todas las aleaciones sin mejorı́a de la temperatura de
transición dúctil-frágil (DBTT).
Con el fin de encontrar un material base wolframio con una DBTT más baja se
realizó también un estudio, aún preliminar, de láminas delgadas de wolframio puro
y wolframio dopado con 0.005wt.% potasio (K). Éstas láminas fueron fabricadas
industrialmente mediante sinterizado y laminación en caliente y en frio y se
sometieron posteriormente a un tratamiento térmico de recocido desde 1073 hasta
2673 K. Se ha analizado la evolución de su microestructura y las propiedades
mecánicas al aumentar la temperatura de recocido. Los resultados mostraron la
estabilización de los granos de wolframio con el incremento de la temperatura de
recocido en las láminas delgadas de wolframio dopado con potasio. Sin embargo,
es necesario realizar estudios adicionales para entender mejor la microstructura y
algunas propiedades mecánicas de estos materiales, como la tenacidad de fractura.
IV
Motivation of the Work
The rise in worldwide energy consumption due to the growth of some developing
countries predicts an increase in energy demands. The use of coal reserves can
satisfy such a demand for some years, however, burning coal also involves an
increase in CO2 emissions, which contribute to global warming. Therefore, the
development of new energy sources that can satisfy such a demand are necessary.
Nuclear fusion power plants will be able to cover the worldwide demand with
limited environmental effects, however, there are still some problems that need to
be solved before fusion power plants can become operational.
This work is driven by the necessity of finding tungsten materials with

decreased brittleness for their use in the divertor, one of the most challenging
components of future nuclear fusion reactors. Since tungsten-based materials have
not been considered for structural components until only a few years ago, there
are insufficient data about the mechanical properties of tungsten that make it very
difficult to select appropriate compositions or manufacturing conditions. This PhD
thesis focuses on investigating the microstructural and mechanical properties of
several bulk and foil tungsten-based materials to find one with a reduced DBTT
and also widen the database for tungsten materials.
During operation materials will work under vacuum and therefore, the
mechanical properties of the tungsten-based materials at high temperatures have
been determined under high vacuum. However, there is a safety concern in case
of an accidental scenario occurs such as a loss of coolant or a loss of vacuum.
Therefore, to have a knowledge about the reaction between materials and air For
that reason tests were also performed in oxidising atmosphere.
V
Contents
Abstract I
Resumen III
Motivation of the Work V
Contents VII
List of Figures XVI
List of Tables XVII
1 Introduction 1
1.1 Nuclear fusion process and devices . . . . . . . . . . . . . . . . . . 2
1.2 ITER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Materials for divertor applications . . . . . . . . . . . . . . . . . . 9
1.3.1 Tungsten materials . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.2 Tungsten-based alloys . . . . . . . . . . . . . . . . . . . . . 13
1.3.2.1 Tungsten-based alloys with rhenium . . . . . . . . 13
1.3.2.2 Tungsten-based alloys with tantalum . . . . . . . 14
1.3.2.3 Tungsten-based alloys with vanadium . . . . . . . 15
1.3.2.4 Tungsten-based alloys with titanium . . . . . . . . 16
1.3.3 Dispersion-strengthened tungsten alloys . . . . . . . . . . . 17
1.3.3.1 ODS tungsten with yttria . . . . . . . . . . . . . . 18
VII
Contents
2 Experimental Methods 21
2.1 Microstructural characterisation . . . . . . . . . . . . . . . . . . . . 21
2.1.1 Energy dispersive X-ray spectroscopy . . . . . . . . . . . . 25
2.1.2 Electron backscatter diffraction . . . . . . . . . . . . . . . . 25
2.2 Nanomechanical characterisation . . . . . . . . . . . . . . . . . . . 27
2.2.1 Nanoindentation tests . . . . . . . . . . . . . . . . . . . . . 27
2.3 Micromechanical characterisation . . . . . . . . . . . . . . . . . . . 30
2.3.1 Microindentation tests . . . . . . . . . . . . . . . . . . . . . 30
2.3.2 Microtensile tests . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4 Macromechanical characterisation . . . . . . . . . . . . . . . . . . . 32
2.4.1 Impulse excitation technique . . . . . . . . . . . . . . . . . 32
2.4.2 Three-point bending tests . . . . . . . . . . . . . . . . . . . 33
2.4.2.1 Flexural strength tests . . . . . . . . . . . . . . . 35
2.4.2.2 Fracture toughness tests . . . . . . . . . . . . . . . 36
2.5 Other physical characterisations . . . . . . . . . . . . . . . . . . . . 37
2.5.1 Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.5.2 Oxidation tests . . . . . . . . . . . . . . . . . . . . . . . . . 37
3 Advanced Experimental Methods: Fracture Toughness

Measurements 39
3.1 Bulk-tungsten samples . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.1.1 Techniques to introduce the notch . . . . . . . . . . . . . . 41
3.1.1.1 Single edge notched beam specimens using a
diamond disc . . . . . . . . . . . . . . . . . . . . . 42
3.1.1.2 Single edge notched beam specimens using a
diamond wire . . . . . . . . . . . . . . . . . . . . . 42
3.1.1.3 Single edge V-notched beam specimens . . . . . . 43
3.1.1.4 Single edge laser-notched beam specimens . . . . . 47
3.1.2 Influence of the notch root radius . . . . . . . . . . . . . . . 48
3.2 Tungsten foils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.2.1 Geometry selection . . . . . . . . . . . . . . . . . . . . . . . 51
3.2.1.1 The double-edge notched specimen . . . . . . . . . 51
3.2.1.2 The single-edge notched specimen . . . . . . . . . 52
VIII
4 Bulk-Tungsten Materials 55
4.1 Materials and samples . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Microstructure at room temperature . . . . . . . . . . . . . . . . . 57
4.2.1 Microstructure of pure tungsten . . . . . . . . . . . . . . . 57
4.2.2 Microstructure of W1Y . . . . . . . . . . . . . . . . . . . . 59
4.2.3 Microstructure of W2V0.5Y . . . . . . . . . . . . . . . . . . 61
4.2.4 Microstructure of W4V0.5Y . . . . . . . . . . . . . . . . . . 64
4.2.5 Microstructure of W2Ti1La . . . . . . . . . . . . . . . . . . 67
4.2.6 Microstructure of W4Ti1La . . . . . . . . . . . . . . . . . . 69
4.3 Micromechanical and physical characterisation . . . . . . . . . . . 72
4.3.1 Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3.2 Elastic modulus . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.3.3 Hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.3.4 High temperature stability in air . . . . . . . . . . . . . . . 77
4.4 Fracture behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4.1 Fracture behaviour of pure tungsten . . . . . . . . . . . . . 82
4.4.2 Fracture behaviour of W1Y . . . . . . . . . . . . . . . . . . 88
4.4.3 Fracture behaviour of W2V0.5Y . . . . . . . . . . . . . . . 92
4.4.4 Fracture behaviour of W4V0.5Y . . . . . . . . . . . . . . . 97
4.4.5 Fracture behaviour of W2Ti1La . . . . . . . . . . . . . . . . 102
4.4.6 Fracture behaviour of W4Ti1La . . . . . . . . . . . . . . . . 106
4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.5.1 W-1%Y2 O3 alloy - Effect of the addition of yttria . . . . . 110
4.5.2 W-V-Y2 O3 alloys . . . . . . . . . . . . . . . . . . . . . . . . 115
4.5.2.1 W-2%V-0.5%Y2 O3 alloy . . . . . . . . . . . . . . 115
4.5.2.2 W-4%V-0.5%Y2 O3 alloy . . . . . . . . . . . . . . 119
4.5.2.3 Effect of the addition of vanadium and yttria . . . 123
4.5.3 W-Ti-La2 O3 alloys . . . . . . . . . . . . . . . . . . . . . . . 127
4.5.3.1 W-2%Ti-1%La2 O3 alloy . . . . . . . . . . . . . . . 127
4.5.3.2 W-4%Ti-1%La2 O3 alloy . . . . . . . . . . . . . . . 131
4.5.3.3 Effect of the addition of titanium and lanthana . . 135
4.5.4 Effect of the alloying elements on the bulk-tungsten . . . . 139
IX
Contents
5 Tungsten Foils 143

5.1 Materials and samples . . . . . . . . . . . . . . . . . . . . . . . . . 144
5.2 Microstructural characterisation . . . . . . . . . . . . . . . . . . . . 144
5.3 Mechanical characterisation . . . . . . . . . . . . . . . . . . . . . . 148
5.3.1 Indentation tests . . . . . . . . . . . . . . . . . . . . . . . . 148
5.3.2 Microtensile tests . . . . . . . . . . . . . . . . . . . . . . . . 150
5.3.3 Three-point bending tests . . . . . . . . . . . . . . . . . . . 153
5.3.4 Fracture surfaces after the TPB tests . . . . . . . . . . . . . 153
6 Conclusions and Perspectives 157

6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
6.1.1 Experimental methods . . . . . . . . . . . . . . . . . . . . . 158
6.1.2 Bulk-tungsten alloys . . . . . . . . . . . . . . . . . . . . . . 159
6.1.2.1 Influence of the addition of yttria . . . . . . . . . 159
6.1.2.2 Influence of the addition of vanadium and yttria . 160
6.1.2.3 Influence of the addition of titanium and lanthana 161
6.1.3 Tungsten foils . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.2 Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
References i
List of Publications xv
Nomenclature xvii
Agradecimientos xix
X
List of Figures
1.1 Types of magnetic confinement configurations . . . . . . . . . . . . 3

1.2 Artistic view of JET and ITER . . . . . . . . . . . . . . . . . . . . 4
1.3 ITER cutaway model . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Cutaway of the ITER vacuum vessel . . . . . . . . . . . . . . . . . 7
1.5 Three-dimensional view of a divertor cassette . . . . . . . . . . . . 8
1.6 Operation of ITER . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.7 Some elements used for alloying tungsten . . . . . . . . . . . . . . 13
1.8 Binary phase diagram of W-Re . . . . . . . . . . . . . . . . . . . . 14
1.9 Binary phase diagram of W-Ta . . . . . . . . . . . . . . . . . . . . 15
1.10 Binary phase diagram of W-V . . . . . . . . . . . . . . . . . . . . . 16
1.11 Binary phase diagram of W-Ti . . . . . . . . . . . . . . . . . . . . 17
1.12 Binary phase diagram of Y-O . . . . . . . . . . . . . . . . . . . . . 18
1.13 Binary phase diagram of W-Y . . . . . . . . . . . . . . . . . . . . . 19
2.1 Interaction volume of high-energy electrons with solid materials . . 22

2.2 Effect of topography on SE2 detector signal . . . . . . . . . . . . . 23
2.3 Schematic of the GEMINI lens from Zeiss with the InLens detector 24
2.4 Schematic set-up of the EBSD detector used inside the FE-SEM . 26
2.5 High-resolution Kikuchi pattern of a single crystal . . . . . . . . . 26
2.6 Berkovich geometry . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7 Typical load-displacement curve obtained by nanoindentation . . . 29
2.8 Vickers geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.9 Miniaturised tensile testing machine for in situ tensile tests . . . . 31
XI
List of Figures
2.10 Experimental set-up for TPB tests at 77 K . . . . . . . . . . . . . 34

2.11 Experimental set-up for TPB tests at high temperature . . . . . . 34
2.12 Schematic of the smooth TPB test specimen . . . . . . . . . . . . . 35
2.13 Schematic of the SELNB TPB test specimen . . . . . . . . . . . . 36
3.1 Microstructure of the W4Ti1La nanostructured alloy . . . . . . . . 40

3.2 Vickers microindentation . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Micrographs of a SENB-D specimen . . . . . . . . . . . . . . . . . 42
3.4 Micrographs of a SENB-W specimen . . . . . . . . . . . . . . . . . 43
3.5 Razor blade machine to create the SEVNB specimens . . . . . . . 45
3.6 Micrographs of a SEVNB specimen . . . . . . . . . . . . . . . . . . 46
3.7 Micrograph of a SELNB notch . . . . . . . . . . . . . . . . . . . . 47
3.8 SELNB notch root . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.9 Influence of the notch root radius on the fracture toughness . . . . 50
3.10 Geometry of the double-edge notched specimen . . . . . . . . . . . 51
3.11 Geometry of the single-edge notched specimen . . . . . . . . . . . . 53
4.1 Microstructure of pure tungsten . . . . . . . . . . . . . . . . . . . . 58

4.2 Histogram of the equivalent grain size distribution for pure tungsten 58
4.3 EBSD measured tungsten grains for pure tungsten . . . . . . . . . 59
4.4 Microstructure of W1Y . . . . . . . . . . . . . . . . . . . . . . . . 60
4.5 High magnification BSE-SEM for W1Y with the EDX spectral data 60
4.6 Micrograph of W1Y with EDX spectral data . . . . . . . . . . . . 61
4.7 Microstructure of W2V0.5Y under low magnification . . . . . . . . 62
4.8 Microstructure of W2V0.5Y . . . . . . . . . . . . . . . . . . . . . . 62
4.9 Histogram of the equivalent grain size distribution for W2V0.5Y . 63
4.10 EBSD measured tungsten grains for W2V0.5Y . . . . . . . . . . . 63
4.11 Micrograph of W2V0.5Y with EDX elemental mapping . . . . . . . 64
4.12 Microstructure of W4V0.5Y under low magnification . . . . . . . . 65
4.13 High resolution micrograph of W4V0.5Y . . . . . . . . . . . . . . . 65
4.14 Histogram of the equivalent grain size distribution for W4V0.5Y . 66
4.15 EBSD measured tungsten grains for W4V0.5Y . . . . . . . . . . . 66
4.16 EDX linescan of a vanadium pool of W4V0.5Y . . . . . . . . . . . 67
XII
4.17 Microstructure of W2Ti1La under low magnification . . . . . . . . 68

4.18 Etched micrograph of W2Ti1La under low magnification . . . . . . 68
4.19 High resolution micrograph of W-1%La . . . . . . . . . . . . . . . 69
4.20 Micrograph of W2Ti1La with EDX elemental mapping . . . . . . . 70
4.21 Microstructure of W4Ti1La under low magnification . . . . . . . . 71
4.22 Microstructure of W4Ti1La . . . . . . . . . . . . . . . . . . . . . . 71
4.23 Density of the bulk materials . . . . . . . . . . . . . . . . . . . . . 72
4.24 Microstructure of the bulk materials at low magnification . . . . . 73
4.25 Elastic modulus for the bulk materials . . . . . . . . . . . . . . . . 75
4.26 Hardness for the bulk materials . . . . . . . . . . . . . . . . . . . . 76
4.27 Mass gain versus temperature for samples oxidised in air . . . . . . 78
4.28 Oxide scale developed for the bulk materials in air at 1273 K . . . 79
4.29 Evolution of the oxide scale in W2V0.5Y with increasing temperature 80
4.30 Flexural σ- curves for pure tungsten . . . . . . . . . . . . . . . . . 82
4.31 Average flexural strength as a function of temperature for pure
tungsten . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.32 Flexural force-displacement curves for SELNB specimens for pure
tungsten . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.33 Average fracture toughness as a function of temperature for pure
tungsten . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.34 Fracture surfaces of pure tungsten at 77 and 300 K . . . . . . . . . 85
4.35 Fracture surfaces of pure tungsten at 873 K . . . . . . . . . . . . . 86
4.36 Fracture surfaces of pure tungsten at 1273 K . . . . . . . . . . . . 86
4.37 Transversal sections of pure tungsten after TPB tests at 1473 K in
smooth and notched specimens . . . . . . . . . . . . . . . . . . . . 87
4.38 Flexural σ- curves for W1Y . . . . . . . . . . . . . . . . . . . . . . 88
4.39 Average flexural strength as a function of temperature for W1Y . . 89
4.40 Flexural force-displacement curves for SELNB specimens for W1Y 89
4.41 Average fracture toughness as a function of temperature for W1Y 90
4.42 Fracture surfaces of W1Y at 300 K . . . . . . . . . . . . . . . . . . 91
4.43 Fracture surfaces of W1Y at 873 K . . . . . . . . . . . . . . . . . . 91
4.44 Fracture surfaces of W1Y at 1273 K . . . . . . . . . . . . . . . . . 92
XIII
List of Figures
4.45 Fracture surface for W1Y at 873 K under vacuum with the spectral
data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.46 Flexural σ- curves for W2V0.5Y . . . . . . . . . . . . . . . . . . . 93
4.47 Average flexural strength as a function of temperature for W2V0.5Y 94
4.48 Flexural force-displacement curves for SELNB specimens for
W2V0.5Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.49 Average fracture toughness as a function of temperature for W2V0.5Y 95
4.50 Fracture surfaces of W2V0.5Y at 77 K and 300 K . . . . . . . . . . 96
4.51 Fracture surfaces of W2V0.5Y at 873 K . . . . . . . . . . . . . . . 96
4.53 Fracture surface of W2V0.5Y at 1473 K . . . . . . . . . . . . . . . 97
4.54 Flexural σ- curves for W4V0.5Y . . . . . . . . . . . . . . . . . . . 98
4.55 Average flexural strength as a function of temperature for W4V0.5Y 99
4.56 Average fracture toughness as a function of temperature for W4V0.5Y100
4.57 Fracture surface of W4V0.5Y at 300 K . . . . . . . . . . . . . . . . 100
4.60 Flexural σ- curves for W2Ti1La . . . . . . . . . . . . . . . . . . . 102
4.61 Average flexural strength as a function of temperature for W2Ti1La 103
4.62 Flexural force-displacement curves for the SELNB specimens for
W2Ti1La . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.63 Average fracture toughness as a function of temperature for W2Ti1La104
4.64 Fracture surface of W2Ti1La at 300 K . . . . . . . . . . . . . . . . 105
4.65 Fracture surfaces for W2Ti1La at 873 K . . . . . . . . . . . . . . . 105
4.66 Fracture surfaces of W2Ti1La at 1273 K . . . . . . . . . . . . . . . 106
4.67 Flexural σ- curves for W4Ti1La . . . . . . . . . . . . . . . . . . . 106
4.68 Average flexural strength as a function of temperature for W4Ti1La 107
4.69 Average fracture toughness as a function of temperature for W4Ti1La108
4.70 Fracture surface of W4Ti1La at 300 K . . . . . . . . . . . . . . . . 108
4.73 Microstructure of the W1Y alloy . . . . . . . . . . . . . . . . . . . 110
4.74 TPB test results for the smooth specimens for W1Y . . . . . . . . 111
XIV
4.75 TPB test results for the SELNB specimens for W1Y . . . . . . . . 112
4.76 Average flexural strength as a function of temperature for W0.5Y
and W1Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.77 Evolution of the fracture surfaces of W1Y after TPB tests . . . . . 114
4.78 Microstructure of the W2V0.5Y alloy . . . . . . . . . . . . . . . . . 115
4.79 TPB test results for the smooth specimens for W2V0.5Y . . . . . . 116
4.80 TPB test results for the SELNB specimens for W2V0.5Y . . . . . 117
4.81 Evolution of the fracture surfaces of W2V0.5Y after TPB tests . . 118
4.82 Microstructure of the W4V0.5Y alloy . . . . . . . . . . . . . . . . . 119
4.83 TPB test results for the smooth specimens for W4V0.5Y . . . . . . 120
4.84 TPB test results for the SELNB specimens for W4V0.5Y . . . . . 121
4.85 Evolution of the fracture surfaces of W4V0.5Y after TPB tests . . 122
4.86 TPB test results for the smooth specimens for W2V0.5Y and
W4V0.5Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.87 TPB test results for the SELNB specimens for W2V0.5Y and
W4V0.5Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.88 Microstructure of the W2Ti1La alloy . . . . . . . . . . . . . . . . . 127
4.89 TPB test results for the smooth specimens for W2Ti1La . . . . . . 128
4.90 TPB test results for the SELNB specimens for W2Ti1La . . . . . . 129
4.91 Evolution of the fracture surfaces of W2Ti1La after TPB tests . . 130
4.92 Microstructure of the W4Ti1La alloy . . . . . . . . . . . . . . . . . 131
4.93 TPB test results for the smooth specimens for W4Ti1La . . . . . . 132
4.94 TPB test results for the SELNB specimens for W4Ti1La . . . . . . 133
4.95 Evolution of the fracture surfaces of W4Ti1La after TPB tests . . 134
4.96 TPB tests for the smooth specimens for W2Ti1La and W4Ti1La . 137
4.97 TPB tests for the SELNB specimens for W2Ti1La and W4Ti1La . 138
4.98 Flexural strength of bulk alloys in air . . . . . . . . . . . . . . . . 139
4.99 Flexural strength of bulk alloys under vacuum . . . . . . . . . . . . 140
4.100Fracture toughness of bulk alloys in air . . . . . . . . . . . . . . . . 141
4.101Fracture toughness of bulk alloys under vaccum . . . . . . . . . . . 142
5.1 Specimen dimensions of tungsten foils . . . . . . . . . . . . . . . . 144

5.2 Cross-sections of the foils in the as-received condition . . . . . . . 145
XV
List of Figures
5.3 EBSD cross-section of the foils annealed 1 h at 1173 K . . . . . . . 146

5.6 Evolution of the microstructure with increasing annealing
temperature for the foils . . . . . . . . . . . . . . . . . . . . . . . . 148
5.7 Hardness of the foils . . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.8 FE-SEM image of the double-edge notched specimen before tensile
test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.9 Force-displacement curves for in situ tensile tests of double-edge
notched specimens . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.10 Fracture surface after in situ tensile tests of double-edge notched
specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.11 Force-displacement curves for in situ tensile tests of single-edge
notched specimens . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
5.12 Foils status after TPB tests for pure tungsten and WVM foils . . . 153
5.13 Fracture surfaces at 1273 K . . . . . . . . . . . . . . . . . . . . . . 154
XVI
List of Tables
3.1 Mean values of the measure fracture toughness with the standard
error for the four methods . . . . . . . . . . . . . . . . . . . . . . . 49
4.1 Weight % composition of the bulk alloys . . . . . . . . . . . . . . . 56

4.2 Features of the starting powders . . . . . . . . . . . . . . . . . . . 56
4.3 Density of the constituent powders . . . . . . . . . . . . . . . . . . 72
4.4 Estimated porosity for the bulk materials . . . . . . . . . . . . . . 74
4.5 Volume fraction for the bulk materials . . . . . . . . . . . . . . . . 74
4.6 Thickness of the oxide scale as a function of temperature . . . . . . 77
4.7 Comparison of the physical and micromechanical properties of the
previously reported W2V, W4V and W0.5Y with the pure W,
W2V0.5Y and W4V0.5Y . . . . . . . . . . . . . . . . . . . . . . . . 125
4.8 Comparison of the physical and micromechanical properties of the
previously reported W4Ti and W1La with pure W, W2Ti1La and
W4Ti1La . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
XVII
CHAPTER 1
Introduction
A rise in the worldwide population during last few decades forecasts an increased
energy demand of 37% by 2040. The growth of some economies outside the
Organization for Economic Cooperation and Development such as China or India
plays a decisive role in this increase, which also involves an increase in global
greenhouse-gas emissions. In terms of reducing these CO2 emissions, which are
mainly caused by burning fossil fuels that contribute to a long-term global warming
of 3.6 o C, and satisfying the demand, new sources of energy need to be developed
[IEA, 2014].
Even though renewable energy technologies, helped by global subsidies, are

rapidly gaining ground, other alternative clean sources have been developed in the
last few decades. One of the most promising sources is nuclear fusion power. It will
have the capacity to cover worldwide energy demand with limited environmental
effects and almost unlimited natural sources. But, although several experiments
have been successfully performed, additional research should be conducted before
nuclear fusion is implemented. Up until now, the materials are the most challenging
and unsolved key point because of the extreme conditions that they will have to
support under operation [IAEA, 2001].
1
Chapter 1. Introduction
1.1 Nuclear fusion process and devices
Nuclear fusion is a process in which two light atomic nuclei collide to form one
heavier atom and release large amounts of energy. It is an analogous process than
that produced in the core of the Sun when hydrogen (H) nuclei combine to form
helium (He). But to achieve sufficient fusion reaction rates and to make fusion
useful as an energy source requires that the fuel (Deuterium (D) and Tritium (T))
be heated to temperatures over 100 million ◦ C to overcome the Coulomb barrier
so they can initiate the fusion reaction [CCFE, 2015]. There are several nuclear
reactions, but one of them (eq. 1.1) requires a lower ignition temperature and is
more likely to occur [ITER-India, 2015].
2
1D +31 T !42 He (3.5 MeV) +10 n (14.1 MeV) (1.1)
Although neither deuterium nor tritium are directly available on Earth, they
can be obtained to have a sustainable nuclear fuel cycle. Deuterium can be found in
water, while tritium, a very rare and expensive radioactive hydrogen isotope, can
be produced within the fusion device via neutron irradiation of lithium targets.
This plasma, however, is extremely thin and fragile and must be contained to
minimise contamination and prevent cooling from coming into contact with the
vessel surfaces. Such a confinement system can be gravitational, which is only
possible in stars, inertial or magnetic.
In inertial confinement, the fuel is quickly heated with a laser that produces
a plasma implosion that reaches very high pressures and temperatures. If the fuel
is dense and hot enough, the fusion reaction rate will burn a significant fraction of
the fuel before it is dissipated. However, to achieve these extreme conditions, the
initially cold fuel must be explosively compressed. In magnetic confinement
on the other hand, the plasma is confined by powerful magnetic fields. This is
the most developed and promising confinement system for designing future fusion
reactors and energy production [CCFE, 2015]. The confinement is achieved by
the generation of a strong toroidal doughnut-shaped field and a poloidal field that
finally produce a helicoidal field. The type of fusion reactor – stellarator or tokamak
(fig. 1.1) – is determined by the design of the poloidal field.
In a stellarator reactor (fig. 1.1, left), the poloidal field is produced by

meticulously designed external field coils. This design yields a more stable plasma
with continuous operation, but the cost of achieving this intricate coil system is
high and the proper design of the coils is challenging. An example of stellarator
2
is the currently operational TJ-II in Madrid (Spain) that produced its first
plasma in 1997. This reactor focuses on investigating the physics of the device
and contributing to the study of magnetic confinement for fusion [CIEMAT,
2015]. Another example is the advanced Stellarator Wendelstein 7-X in Greifswald
(Germany) that investigates the suitability of using a stellarator for a power plant
and wants to demonstrate its essential property: continuous operation [IPP-W7-X,
2015].
Fig. 1.1 Types of magnetic confinement configurations: segment of the steel

vessel of W7-X stellarator (left) and cutaway graphic of the tokamak
configuration of the JET vessel (right) [EUROfusion, 2015].
Another type of fusion device design, a tokamak reactor (fig. 1.1, right), is
the most advanced and best investigated configuration. In it, the plasma flow is
induced by an increasing current in the poloidal coils. Once the current reaches its
maximum value, the induction and consequently the pulse will cease. Therefore,
unlike a stellarator reactor, a tokamak design has to operate in a pulsed mode.
It became the dominant concept in the late 1950s after pioneering work in the
Soviet Union [New-Scientist, 2009]. But it was not until the early 1990s when
the first fusion device, the Tokamak Fusion Test Reactor in Princeton, New
Jersey (United States), started to operate with the actual fusion fuel (D+T). It
achieved plasma temperatures of several hundred million ◦ C and major progresses
in confinement time and energy density, but it was shut down in 1997 [PPPL,
2015]. In Japan, the JT-60, in operation since 1985, holds the record for the
highest values of the three key parameters in which fusion depends: density,
temperature and confinement time [JAEA, 2015]. In the meantime, the Tore Supra,
a EURATOM-CEA installation in operation since 1988 in France, has achieved its
goal of creating an exceptionally long-duration plasma [EURATOM-CEA, 2015].
At the moment, however, the Joint European Torus (JET) in Oxfordshire

(United Kingdom) is the largest and most powerful tokamak in the world and the
only one capable of producing fusion energy. JET was designed to study plasma
behaviour in conditions and dimensions approaching those required in a fusion
3
reactor. It produced its first plasma in 1983 and the first controlled release of fusion
power in the world in 1991. Its current primary task is to provide the knowledge for
the construction and operation of ITER, the latest worldwide collaboration project
before the demonstration power plant (DEMO) in the 2030s [EUROfusion, 2015;
New-Scientist, 2009].
Fig. 1.2 Artistic view of JET and ITER by Russel Perry [EUROfusion, 2015].
1.2 ITER
ITER is a large-scale collaboration project for the construction of a tokamak

fusion reactor. It aims to demonstrate the technological and scientific feasibility
of nuclear fusion energy for commercial purposes. The name ITER, originally an
acronym for the International Thermonuclear Experimental Reactor, is the Latin
word meaning ”the way”. This name was chosen by participants in the conceptual
design activities (CDA) between 1988 and 1992 to express their common hope of
creating an international cooperation project for the development of a new form
of energy [IAEA, 1988; ITER, 2015].
The project established its roots towards the end of the Cold War when
Mikhail Gorbachev of the former Soviet Union and President Ronald Reagan
of the United States agreed to develop an international project to create fusion
energy for peaceful purposes. Soon after, Europe and Japan joined the project. In
2003 entered China and South Korea, and finally in 2005 India became a member.
Together, these countries represent over half of the world population [ITER, 2015].
4
Under the support of the International Atomic Energy Agency (IAEA), the
project began its CDA in 1988. Its success lead to the detailed engineering design
activities that produced a complete documentation of the ITER final design report
in 1992 that was approved by its members in 2001 [IAEA, 2001]. This report
includes not only the approval for the construction but also a full analysis of costs
and a safety assessment for the facility now under construction at Cadarache in
southern France [Ikeda, 2010].
The machine itself (fig. 1.3) is a complex device based on the tokamak concept
of plasma magnetic confinement. Inside the vacuum vessel temperatures of
approximately 150 million ◦ C will persist, and a strong helicoidal magnetic field
created by the superconducting magnets and the plasma itself is used to keep the
plasma away from the walls.
v
acuumv
esel
super
conduct
ingmagnet
s
cr
yost
at
t
her
mal
shi
eld
bl
ank
et
di
ver
tor
el
ect
romagnet
Fig. 1.3 ITER cutaway model [New-Scientist, 2009].
5
The scientific goals of ITER are to release 10 times as much energy as it will
need to initiate the fusion reaction, i.e. a ratio of fusion to input power of 10
(Q 10), with an inductive burn duration between 300 and 500 s and a steady
state operation of Q 5. In terms of engineering and testing performance, ITER
will be used to test tritium breeding module concepts with 14 MeV-neutron power
loads on the first wall. The reactor will furthermore demonstrate the availability
and integration of fusion technologies and components for a future commercial
power plant [IAEA, 2001].
The vacuum vessel (figs. 1.3 and 1.4) is a hermetically-sealed steel container
inside a cryostat that houses the plasma and acts as a first safety barrier. Its size
determines the volume of the fusion plasma; in ITER has dimensions of 6 metres
diameter by 11 metres height.
The inner components of the machine (fig. 1.4), commonly known as

plasma-facing components (PFCs), include the blanket system and the
divertor. They directly face the fusion plasma over an area of approximately 850 m2
and are one of the most critical and technically challenging components. PFCs
consist of a PFM mounted onto a heat sink, which acts as armour. This PFM
is supported by a structural material that acts as shielding. According to Merola
et al. (2010), the main functions of the PFC:
Absorb radiation, conduct heat from the plasma and contribute to the
absorption of neutronic heating.
Minimise the impurities from the plasma.
Provide limiting surfaces that define the plasma boundary during start-up
and shutdown.
Contribute to the plasma passive stabilisation.
The blanket system (fig. 1.4) covers the inner surface of the vacuum vessel.
Its basic function is to provide heat and neutron shielding to the vessel but also to
external components. The blanket system includes blanket modules, which provide
the bulk of the shielding, and blanket manifolds, which route the water coolant to
the blanket modules [Merola et al., 2014]. In a future fusion power plant, neutrons
will be slowed down in the blanket and their kinetic energy will be transformed into
heat energy and collected by the coolants for its use as electrical power production.
At a later stage of the project, some of these blanket modules will be also used
to test materials for tritium breeding concepts [ITER, 2015]. Such test blanket
modules will provide the first experimental data of the performance of the breeding
blankets in the integrated fusion nuclear environment since data about tritium
6
self-sustainment are essential to design and predict the performance of DEMO

and other future fusion reactors [Giancarli et al., 2012].
bl
ankets
yst
em
di
ver
tor
Fig. 1.4 Cutaway of the ITER vacuum vessel [ITER, 2015].
The divertor (fig. 1.4) is located at the very bottom of the vacuum vessel.
Its main function is to exhaust the scrape-off layer power from the plasma
volume while maintaining an acceptable level of core plasma contamination due
to α-particles (the so-called helium-ash) and transmutation impurities. However,
at the same time, the divertor must tolerate high heat loads (approximately 15%
of the total fusion power) and provide neutron shielding to the vacuum vessel and
magnet coils in its vicinity [Norajitra et al., 2008; Merola et al., 2010]. Its design,
which should be capable of removing a high heat load of at least 10 MW/m2 ,
is based on a modular construction of a helium-cooled finger unit that helps to
reduce thermal stresses [Norajitra et al., 2007]. The success of the divertor strongly
depends on the effectiveness of the cooling technology and on the availability of
appropriate structural materials. The divertor consists of 54 remotely-removable
cassettes (fig. 1.5), each of which includes a cassette body and three PFCs: the
inner vertical target (IVT), the outer vertical target (OVT) and the dome [Merola
et al., 2014].
7
Fig. 1.5 Three-dimensional view of a divertor cassette [Merola et al., 2014].
The operation of ITER can be explained as follows (fig. 1.6) [New-Scientist, 2009]:
A) Electricity flows through the electromagnet and produce a voltage across
the gas while a blast of deuterium and tritium is injected into the vacuum
vessel. Due to the voltage, the deuterium and tritium atoms are transformed
into charged atoms (ions) within a few microseconds, thereby forming the
plasma.
B) The plasma, locked inside the vacuum vessel due to the magnetic
confinement, is quickly heated by the superconducting magnetic coils to
10 million ◦ C. However, even at this temperature the plasma is not hot
enough for fusion to occur; a temperature of 100–200 million ◦ C is required.
Therefore, radio and microwaves are fired into the plasma along with
high-energy beams of deuterium atoms.
C) High-energy neutrons and helium particles deposit their energy into the
plasma and keep it hot. As these materials turn into ash, they are removed
though the divertor. However, these neutrons and other particles also
bombard the tiles on the PFCs and heat them.
D) Deuterium and tritium must be continuously refuelled to continue the

process. Unburnt fuel is recovered from the gaseous exhaust and the reaction
can be fine-tuned by firing frozen pellets of fusion fuel deep into the plasma.
8
1.3 Materials for divertor applications
As a part of the PFCs, the divertor directly faces the plasma. It is also the
highest thermally loaded component that will have to withstand very demanding
conditions during the reactor operation. The most important requirements of the
divertor are to be resistant, at least up to a certain point, against radiation damage
such as the evolution of transmutation elements and He/H production that leads
to swelling and embrittlement of the materials. Furthermore, it is essential for
the PFMs to maintain their integrity within the operation temperature window
(OTW) of the reactor, which is currently estimated to be 1073–1473 K. Such an
OTW is defined in all body-centred cubic (bcc) and most face-centred cubic metals
by the DBTT at the lower limit and the recrystallisation temperature (RCT) at
the upper limit [Norajitra et al., 2004; Bolt et al., 2004]. Since recrystallisation
in structural materials should be avoided in any case, the DBTT is the main
focus of the design. It is strongly dependent on the chemical composition and
the microstructural state and hence the manufacturing history (particularly for
refractory alloys). However, the determination of the DBTT strongly depends on
the testing method [Nakamura et al., 1994; Rieth and Dafferner, 2005].
Additionally, PFCs conditions beyond ITER (e.g. DEMO) will be more

stringent due to the increase in operational lifetime and also because the surface
power to be removed from the plasma, the neutron fluencies and the operational
temperature will all increase. Furthermore, PFCs should allow the operation
under off-normal events such as plasma disruptions and high transient heat loads
during strong edge localised modes of the plasma [Bolt et al., 2002]. All of these
requirements therefore, demand the development of materials and alloys that
broaden the range of applications and usefulness.
During the 1990s, plasma performance was the most important aspect of fusion
research, therefore the utilization of low-Z PFMs such as beryllium, graphite or
boron was preferred [Noda et al., 1997]. The experience during plasma operation
showed that the use of low-Z PFMs greatly reduced radiation losses from the
plasma compared with medium and high-Z PFMs [Bolt et al., 2002]. Additionally,
low-Z PFMs significantly improved the reactor performance by contributing to a
steady increase in the fusion triple product of density, temperature and energy
confinement [Philipps et al., 2000].
The use of beryllium is not suitable as a PFM because of its low power handling
capacity and high physical sputtering yield at low particle energies [Neu, 2003].
Carbon-based materials meanwhile, have very advantageous thermo-mechanical
9
pl
asma
f
usi
onr
eact
ion
C D
Fig. 1.6 Operation of ITER [New-Scientist, 2009].
10
properties, good availability, good machinability, low activation and low radiation
losses from carbon at plasma temperatures [Garcı́a-Rosales, 1994]. A significant
advantage of these carbon-based materials is that they do not melt under
off-normal heat loads. However, the biggest concern is erosion under normal
operating conditions, which most serious consequence would be the subsequent
redeposition of carbon in remote locations. The solution to these problems is
to develop an efficient cleaning method for layer removal or to omit carbon
as a PFM [Neu, 2003]. Therefore, these carbon-based materials should be used
in combination with special wall conditioning procedures such as boronisation,
siliconisation, lithium injection or beryllium evaporation, which reduce oxygen
impurities and improve density control [Philipps et al., 2000].
Originally, the ITER divertor baseline was slated to use carbon fibre-reinforced
carbon (CFC) for the lower part of the IVTs and OVTs, which would be replaced
with a full-tungsten divertor prior to the start of nuclear operation (hence 3–4 years
of operation). However, due to the potential risk of rapid accumulation of tritium
by codeposition with eroded CFCs or deep migration into the bulk-carbon, CFC
were excluded from the present ITER licensing [Merola et al., 2014]. Consequently,
following a strategy of cost reduction by the utilisation of a single divertor from the
start of operation until the nuclear phase, in September 2011 the ITER Council
proposed to investigate the possibility of beginning operation with a full-tungsten
armoured divertor compatible with nuclear operations. This proposal was analysed
and completed successfully and November 2013 it was decided that ITER would
operate with a full-tungsten divertor [Carpentier-Chouchana et al., 2014].
Recently, some authors such as Brooks et al. (2015) have focused their attention
on broadening the options for high-Z PFM beyond tungsten to avoid relying
on one single material. Five potential alternative materials (zirconium, niobium,
molybdenum, hafnium and tantalum) were examined as full-thickness structural
materials or thinner coating materials. Even so, research is still focused on tungsten
as a PFM for divertor applications.
1.3.1 Tungsten materials
The use of tungsten, a bbc refractory metal, as a PFM is mainly motivated

by its superior thermo-mechanical properties and low erosion under steady
state operation conditions. These favourable properties include, among others,
high melting point (3410 ◦ C), good thermal conductivity (155 W/mK at room
temperature), low vapour pressure (1.3 10−7 Pa at its melting point), relatively low
11
thermal expansion (4.6 10−6 K−1 at room temperature; ASM Handbook Vol.02
(1991)), high creep resistance and high temperature strength, low sputtering
yield, low tritium retention and low hydrogen solubility [Norajitra et al.,
2004]. Nevertheless, many other properties that are relevant in a fusion reactor
environment are required besides these specific advantages. Each property has to
be addressed through specific experimental and theoretical approaches to specify
its behaviour in an integrated approach.
All of the properties noted above make tungsten a suitable PFM. However, its
structural applications are still restricted by its inherent brittleness. The safety
and efficiency of a fusion reactor will depend on its materials since their properties
should be maintained inside the OTW. Therefore it is convenient to use materials
with a DBTT as low as possible and a suitable toughness and ductility at low
temperatures for preventing an accidental brittle fracture when the PFCs are below
the lower bound of the OTW. Pure tungsten, as a material with bbc structure,
has a DBTT that ranges 473–673 K depending on the characterisation method.
Therefore, to enable its use as structural material, its DBTT needs to be decreased.
Many efforts have been made to lower the DBTT of tungsten. The main
strategies that have been employed thus far include conventional techniques such
as alloying with other elements or strengthening with dispersed particles and
thermo-mechanical treatments. Although in general no substantial improvements
have been reported, it was found that oxide-dispersion-strengthened (ODS) or
Al-K-Si doping enhances the mechanical strength and increases the RCT of
tungsten but unfortunately without decreasing its DBTT [Rieth and Dafferner,
2005; Faleschini et al., 2007]. In the last few years, several other methods such
as the sinterising of tungsten composites from tungsten or doped tungsten foils
have been developed with promising improvements [Reiser et al., 2013a]. More
recently, active extrinsic toughening mechanisms of energy dissipation that has
been widely used in ceramic fibre-reinforced ceramics [Jones and Henager, 2005],
has being implemented in tungsten fibre-reinforced tungsten composites with
successful results [Riesch et al., 2014].
In this phD thesis, the main strategies to enhance fracture toughness have been
alloying with elements such as titanium or vanadium and ODS particles for the
case of bulk-tungsten materials (chapter 4) and annealing heat treatment for the
tungsten foils (chapter 5). In the following sections a more specific description of
some of the reported attempts to ductilize tungsten are provided.
12
1.3.2 Tungsten-based alloys
The most common alloying elements used for alloying tungsten are shown in
fig. 1.7. Some of these alloys have been studied because of their declared good
properties (e.g. W-Ta). However, others such as W-Re that were rejected for
economic or practical reasons are still under study because of the transmutation
of tungsten during irradiation conditions that gives rise to W-Re or W-Os alloys.
Sc Ti V Cr Mn Fe
Sc
andi
um Ti
tani
um Vanadi
um Chr
omi
um Manganes
e I
ron
Y Zr Nb Mo Tc Ru
Yt
tri
um Zi
rconi
um Ni
obi
um Mol
ybdenum Tec
hnet
ium Rut
heni
um
La Hf Ta W Re Os
Lant
hanum Haf
nium Tant
alum Tungs
ten Rheni
um Os
mium
Fig. 1.7 Some elements used for alloying tungsten.
1.3.2.1 Tungsten-based alloys with rhenium
The addition of rhenium enhances the ductility of tungsten by shifting its DBTT
to lower values. An improvement in the room temperature fabricability together
with an increased in the high temperature strength of the alloy was first reported
by Geach and Hughes (1956). In addition, an increase in the RCT was reported by
Mutoh et al. (1995). Such characteristics would made W-Re alloys very promising
materials for fusion applications were if not for the fact that transmutation
under irradiation conditions produce drastic embrittlement and hardening by
precipitation and therefore compromises the structural integrity of the PFCs
[Herschitz and Seidman, 1984]. Nonetheless, despite its high price of approximately
$5000 (USD) per kilogram [MetalPrices, 2015], W-Re alloys have drawn significant
attention over the years.
Rhenium, with a hexagonal-close-packed (hcp) crystal structure, has the second

highest melting point after tungsten (3180 ◦ C; ASM Handbook Vol.02 (1991)). Its
complex binary diagram with tungsten (fig. 1.8) exhibits σ and χ phases that when
precipitated are responsible for the embrittlement under neutron irradiation. The
highest solubility of Re in tungsten is pproximately 37 at.% although the highest
W-Re alloy reported is W-26% Re [Nemoto et al., 2000; Wurster et al., 2010].
The hardening effects of neutron damage during operation have been reported by
several authors (e.g. Armstrong et al. (2013) and Xu et al. (2015) in the case of
W-5 wt.%).
13
Fig. 1.8 Binary phase diagram of W-Re [Liu and Chang, 2000].
1.3.2.2 Tungsten-based alloys with tantalum
Tantalum, with a bcc crystal structure, features a combination of properties that

are rarely found in many refractory metals: excellent machinability, a low DBTT
and a high melting point (3020 ◦ C) [ASM Handbook Vol.02, 1991]. In combination
with tungsten, W-Ta alloys present complete solubility in solid and liquid phases
(fig. 1.9). This situation, together with the fact that tantalum does not produce
new phases after transmutation, makes it a widely used alloying element to reduce
the DBTT in low activation ferritic-martensitic steels [Gilbert and Sublet, 2011].
Early studies by Stephens (1970) showed an effective decrease in DBTT for low
alloyed alloys. This decreased in DBTT was very pronounced for W-1 at.% Ta but
was already lost for W-2 at.% Ta. These findings indicate an effective solid solution
softening at temperatures lower than 10% the melting temperature. Gourdin et al.
(1994) meanwhile, have reported a good combination of mechanical properties such
as elasticity and corrosion resistance (from the tantalum) and strength (from the
tungsten) for tantalum-based alloys with 2.5, 5 and 10% tungsten content. In
accordance with these results, Wurster et al. (2011) recently reported a rather
good performance of the industrial alloy W-1% Ta with a high decrease in the
workability and fracture toughness with the addition of 5 and 10 wt.% Ta. These
14
results are also in agreement with data reported by Tejado et al. (2015) from
laboratory scale W-5% Ta and W-15% Ta processed using MA and HIP.
Fig. 1.9 Binary phase diagram of W-Ta [ASM Handbook Vol.03, 1992].
All these results together with the data of Rieth et al. (2010), suggest the
optimisation of the chemical composition of W-Ta since the alloys that have
been already studied did not exhibit improved DBTT values compared with pure
tungsten.
1.3.2.3 Tungsten-based alloys with vanadium
W-V alloys are characterised by an isomorphous phase diagram (fig. 1.10) with a
continuous range of solid solution. However, the formation of intermetallic phases
even at high temperatures was predicted via ab initio calculations [Muzyk et al.,
2013]. Such compounds would involve significant changes in their elastic properties
compared with the isotropic elastic properties of pure tungsten.
Vanadium-based alloys have been typically used in the aerospace industry,

because of their low density, high strength and resistance to high operating
temperatures and stress [Shyrokov et al., 2009]; however, little research have
been conducted on W-V alloys. In recent years, W-V composites have been
investigated using numerical simulation that suggested an enhancement of the
15
OTW: an improved creep resistance at high temperatures and a reasonable fracture

toughness at low temperatures [Hohe and Gumbsch, 2010]. Nonetheless, one of the
main problems with vanadium is its low erosion resistance.
Fig. 1.10 Binary phase diagram of W-V [Okamoto, 2010].
1.3.2.4 Tungsten-based alloys with titanium
Titanium has been used as a sintering activator because of its extreme corrosion
resistance and high strength. The W-Ti binary phase diagram (fig. 1.11) shows
limited solid solution above 740 ◦ C and complete solid solution at 1255 ◦ C.
Although the prerequisite for the formation of a complete solid solution is that
the elements possess the same crystal structure and titanium is hcp. However, the
formation of a complete solid solution is also affected by other factors, such as the
difference in atom size between the alloying elements [Wang et al., 2009].
The mechanical properties of W-Ti alloys have been scarcely investigated.

Savoini et al. (2013) reported the formation of some metastable titanium phase
such as α0 , α00 or ω that could influence the mechanical behaviour of W-Ti alloys.
Meanwhile, Garcı́a-Rosales et al. (2014) studied self-passivating tungsten-based
alloys (WCr12Ti2.5) that minimise the formation of volatile, radioactive oxides
and the creation of protective barriers in case of a loss of coolant.
16
Fig. 1.11 Binary phase diagram of W-Ti [ASM Handbook Vol.03, 1992].
1.3.3 Dispersion-strengthened tungsten alloys
Another method to increase the high temperature performance of tungsten by

raising the RCT significantly is doping or the addition of dispersion-strengthened
particles. Such an improvement is almost independent of the type of dispersoid
[Rieth and Dafferner, 2005]. However, ODS particles inhibit grain growth either
during the recrystallisation process or during sintering and contribute to an
increase in RCT without decrease in DBTT [Ryu and Hong, 2003]. The dispersion
of strengthening TiC particles is the only one that decreases the DBTT [Kitsunai
et al., 1999]. tungsten-TiC alloys exhibit superplasticity at high temperatures
based on grain boundary sliding above 1673 K, which is strongly dependent on
the TiC content as well as the mechanical alloying atmosphere [Veleva, 2011].
The dispersion of thorium dioxide (ThO2 ) particles has been studied from
the beginning because these particles possess the highest melting temperature
(3473 K) among all metallic oxides [Luo et al., 1991]. Unfortunately, their use has
been dismissed due to their radioactive potential, which is an obvious obstacle
to nuclear applications [King and Sell, 1965]. Nowadays, the addition of yttria
(Y2 O3 ) or lanthana (La2 O3 ) as a second-phase particle is a promising attempt
17
under study. Although lanthana has been only rarely investigated, it improves the
processability, suppresses recrystallisation and exhibits a slight strengthening in
creep [Rieth and Dafferner, 2005]. However, the inherently brittle behaviour of
pure tungsten increases [Rieth et al., 2010].
1.3.3.1 ODS tungsten with yttria
The addition of yttria particles provides excellent chemical and thermal stability
[Swamy et al., 1998] and creep resistance at high temperatures [Kim et al., 2006].
Additionally, since yttrium possess a high chemical affinity to oxygen, W-Y2 O3
alloys have good corrosion resistance [Itoh and Ishiwata, 1996; Veleva et al., 2009].
The content of the second-phase should be carefully optimised since the

addition of yttria to a tungsten matrix generates some contradictory effects in
terms of alloy behaviour. Although increased quantities of yttria contribute to
increasing the corrosion resistance, yttria also lowers the strength and ductility of
the resulting alloy [Itoh and Ishiwata, 1996]. In addition, higher contents of yttria
compromise the use of the alloy for structural applications at high temperatures
because the melting temperature of yttria (2711 K; Swamy et al. (1998)) is much
lower that of tungsten (figs. 1.12 and 1.13).
Fig. 1.12 Binary phase diagram of Y-O [Okamoto, 2011].
18
Fig. 1.13 Binary phase diagram of W-Y [Okamoto, 2000].
These alloys are therefore expected to combine a high strength at high

temperatures of the ceramic particles with a high ductility as compared to ceramics
due to the metallic matrix.
19
CHAPTER 2
Experimental Methods
Microstructural, mechanical and fractographical characterisation were performed

in the tungsten-based materials and alloys using experimental procedures on
the nano, micro and macro scales. In this chapter, the experimental techniques
employed in both the bulk and the foils together with the procedures used for the
sample preparation, are exposed in detail. Nevertheless, particular specifications
of each type of material will be clarified in the corresponding chapter (see chapters
4 and 5 for the bulk and foil tungsten materials, respectively).
2.1 Microstructural characterisation
Microstructural and compositional characterisation of the specimens was

performed by analysing transversal sections. Although the standard preparation
involves embedding the samples in resin, this method was discarded for the bulk
specimens since they were simply placed on an aluminium holder. This preparation
allowed a higher conductivity of the samples, which resulted in better quality
imagining with the field emission scanning electron microscope (FE-SEM). On the
other hand, the transversal sections of foils were embedded in hard resin to produce
a tough and stiff compound that prevented the cross-sections from delaminating
during the grinding and mechanical polishing [Reiser et al., 2012].
21
Chapter 2. Experimental Methods
Afterwards, the sample surfaces were prepared for observation using a

Struers Abramin (Germany) universal grinding and polishing machine. They
were progressively more finely ground using 400, 600 and 1000 grit SiC paper
and were then polished using a diamond suspension of 3 and 1 µm. Finally,
the samples were etched using Murakami reagent (10 g KOH, 10 g K3 Fe(CN)6
in 100 ml distilled water) for 1 min to reveal the microstructure of the
materials more clearly, as necessary. High resolution imagining analysis of the
microstructure was performed using a FE-SEM from Zeiss (AURIGA with
GEMINI electron beam column, Germany) equipped with secondary electrons
(SE2, InLens), backscattered electrons (BSE, EBSD) and X-ray (EDX) detectors.
For observations of the foils, optical microscopy at very low magnification (up to
200 ) was used due to the large microstructure observed on the samples after the
annealing treatment.
FE-SEM utilises a very strong magnetic field to draw electrons from a very
fine metallic tip to the sample. When the electron beam strikes the specimen, it
interacts with some of the material (the interaction volume; fig. 2.1) to produce
different signals. The penetration length of the electron beam into the sample is
affected by many factors; the most significant are the accelerating voltage (kV)
that is applied and the atomic number (Z) of the material. The higher the kV, the
greater the penetration; whereas the higher the Z, the lower the penetration. In
general, this electron penetration ranged from 1 to 5 µm with the incident beam
perpendicular to the sample.
e
lec
tron
beam
S
E2
S
E1
BS
E X-
ray
S
E
BS
E
c
har
act
eri
sc
X
-ra
ys
s
ampl
e
Fig. 2.1 Interaction volume of high energy electrons with solid materials
[AMMRF, 2014].
22
The majority of the incident electrons interact with the specimen atoms and are
scattered instead of penetrating the sample. These scattering events can be elastic
or inelastic. In elastic scattering, the electron trajectory changes but its kinetic
energy and velocity remain essentially constant because of the large difference
in the mass of the electron and the masses of the atomic nuclei. Alternatively,
in inelastic scattering, the trajectory of the incident electron may be only slightly
perturbed but energy is lost via interactions with the orbital electrons of the atoms
in the specimen.
The most common interaction products used for generating images in SEM are
SE1, SE2, BSE and X-rays. They are collected by different detectors to obtain a
wide range of information about the materials being studied.
The SE2 detector collects low-energy secondary electrons (SE2) ejected from
a thin layer near the surface of the sample by the BSEs (fig. 2.1) that have returned
to the surface after several inelastic scattering events. Since the electrons come from
a surface area that is larger than the spot associated with the incoming electrons,
this is a low-resolution signal. The detector is used to show the morphology and
surface topography of the sample. The contrast of the image depends on the
number of SE2 that reach the detector from each spot of the sample. Such contrast
is dominated by the so-called edge effect (fig. 2.2); a higher brightness is produced
at the edges because more SE2 are able to leave the sample. In the same way, faces
oriented towards the detector will be brighter than those on the opposite side.
Fig. 2.2 Effect of topography on SE2 detector signal [ETH, 2014].
23
The BSE detector collects high-energy backscattered electrons (fig. 2.1)

coming out of the material via elastic scattering interactions with the specimen’s
atoms. This detector is used to detect contrast between areas with different
chemical compositions based on the fact that elements with high Z backscatter
electrons more strongly (and appear brighter in the image) than low Z elements.
However, BSEs are also used to determine the crystallographic orientation of grains
using the EBSD detector, as will be explained in section 2.1.2.
The InLens detector is an efficiency tool for high-resolution secondary

electrons (SE1) imagining. It is located above the objective lens (fig. 2.3) and
detects directly in the beam path. SE1 are generated by the incoming electron
beam as they enter the surface, and the resolution signal of the InLens detector
is only limited by the electron beam diameter. Its efficiency is mainly determined
by the electric field of the electrostatic lens, which decreases exponentially with
distance. Particularly at low kV and small working distances, it is possible to
obtain images with high contrast besides morphology and surface topography
information. This capability is possible because the InLens detector features a
high lateral resolution unlike images recorded by a conventional SE2 detector in
which topographic information is dominant.
E
lec
troma
g ni
c
a
pert
urechar
ger
F
iel
dle
ns
I
nLe
nsS
Ede
tec
tor
Ma
gne cl
ens
E
lec
tros
tacl
ens
S
pec
ime
n
Fig. 2.3 Schematic of the GEMINI lens design from Zeiss with the InLens
detector [Zeiss, 2012].
24
2.1.1 Energy dispersive X-ray spectroscopy
Energy dispersive X-ray spectroscopy (EDX) is a non-destructive microanalytical

technique that provides qualitative and quantitative X-ray analysis information
about chemical composition of a material. Electrons from the beam penetrate
the sample and interact with the atoms. In this work, the diffracted X-rays
(background and characteristic) were captured by the detector (QUANTAX EDX
from BRUKER), and the received signal was displayed using the SPRIT software
as a spectrum of intensity. The energies of the characteristic X-rays allowed to
identify the elements in the sample, while the intensities of the characteristic X-ray
peaks provided their quantification.
Special sample preparation, other than that required to image inside the
FE-SEM, was not required. The different outputs that can be obtained from
the EDX analysis (spectrum, mapping, quantitative analysis) were used for
microstructural characterisation of the materials.
2.1.2 Electron backscatter diffraction
Electron backscatter diffraction (EBSD) is a powerful crystallographic SEM tool

for microstructural characterisation in which a crystalline sample tilted 70o is
struck by an incident electron beam (fig. 2.4). The beam interacts with the
crystal lattice within a small volume and coherently diffract BSEs out of the
sample (fig. 2.1). The diffracted electrons generate a Kikuchi pattern (fig. 2.5)
on the phosphor screen that is characteristic of each crystal’s structure and
orientation. This pattern yields information about the absolute crystal orientation
with sub-micron resolution [Wilkinson and Britton, 2012].
The Kikuchi patterns were processed and analysed using the ESPRIT
software of the QUANTAX CrystAlign EBSD analysis system from BRUKER.
These patterns can be used to determine grain size, misorientation, texture,
grain boundaries, deformed grains or phases. The detector is equipped with
forescattered electron (FSE) detectors and BSE detectors (fig. 2.4) that provide
valuable additional information that can be used for efficient microstructure
characterisation. The BSE detectors are located to improve the image quality,
whereas the FSE make visible the slightest signal variation due to orientation
change when a polycrystalline sample is scanned [Bruker, 2014].
25
e
lect
ron
beam
BS
Edet
ect
ors
70ºlt
ed
s
peci
men
F
SEdet
ect
ors
Fig. 2.4 Schematic set-up of the EBSD the detector used inside the FE-SEM
chamber [Bruker, 2014].
Fig. 2.5 High-resolution Kikuchi pattern of a single crystal.
26
The preparation of samples for EBSD analysis include an additional final

polishing since this technique requires free of relief surfaces to obtain good quality
and reliable results. Although this step can be achieved via electropolishing, ion
milling or colloidal silica, this last method was preferred because it is the most
suitable for high Z materials [Chen et al., 2012]. The step was performed with
a VibroMetTM 2 Vibratory Polisher from Buehler (United States) designed to
prepare highly reflective polished surfaces with exceptional flatnesses. Its low
amplitude vibratory action produces less deformation, flatter surfaces and reduced
edge rounding. It also yields a stress-free surface without the use of dangerous
electrolytes (often a mixture of acids) associated with electro-polishing. Special
care should be taken with the polishing time; it should not be prolonged but rather
just sufficient to achieve the desired surface finish without causing excessive relief
[Oxford-Instruments, 2014; Buehler, 2014].
2.2 Nanomechanical characterisation
2.2.1 Nanoindentation tests
Nanoindentation tests were performed at room temperature using a NanoIndenter

XP from former MTS Systems Corporation (United States) equipped with a
continuous stiffness measurement (CSM) module. This module makes it possible to
measure the contact stiffness at any point along the loading curve and not just at
the point of unloading, which is what is possible using conventional measurements
according to Li and Bhushan (2002). The CSM module therefore results in a
significant improvement in terms of the information that can be obtained.
The nanoindentations were performed using a standard Berkovich tip

calibrated with fused silica. Although several indenter geometries can be used
(cube corner, spherical, conical, etc), the selected Berkovich tip (fig. 2.6) is the most
frequently used for instrumented indentation testing. This three-sided pyramid
indenter is preferred over the four-sided Vickers indenter because of its smaller
angle between faces that minimises the influence of friction and makes it easier to
ground to a sharp point [Hay and Pharr, 2000].
With this equipment, instrumented indentation tests can be performed under

load or displacement control. The displacement control tests were conducted at
the beginning to determine the optimum test load to avoid the indentation size
effect that would affect the mechanical properties results. The force, determined
27
to be 0.6 N, was applied to the indenter through a calibrated electromagnetic

coil with a resolution of 50 nN. Meanwhile, the displacement was measured using
an electrostatic transducer with a resolution of 0.01 nm. During the tests, the
control unit continuously registered the force and displacement experimented by
the indenter when the surface of the material was penetrated.
Fig. 2.6 Geometry of the Berkovich indenter showing its 65.3◦ angle between
the centre line and each face (left) and the indentation produced in the material
(right).
Based on the load-displacement data obtained (curves as in fig. 2.7), the

average values of nanohardness and elastic modulus were calculated according
to the method of Oliver and Pharr (1992). Nanoindentation hardness (Hn ) of the
materials was defined as the ratio between the peak of applied load (Pmax ) and
the projected contact area of the indentation (A) as follows:
Pmax
Hn = (2.1)
A
The elastic modulus (En ) was inferred based on the initial unloading contact
stiffness (S, fig. 2.7) and the following the formula:
p
1 π S
Er = p (2.2)
β 2 A
28
where β is a constant that depends on the geometry of the indenter (β=1.034 for
the Berkovich tip) and Er is the reduced elastic modulus that accounts for the
fact that elastic deformation occurs in both the sample and the indenter and is
given by:
1 1 ν2 1 νi2
= + (2.3)
Er E Ei
where E and ν are the elastic modulus and Poisson’s ratio for the material tested,
respectively, and Ei and νi the values for the indenter; in the case of diamond Ei
and νi are 1141 and 0.07 GPa, respectively [Tabor, 1951].
Fig. 2.7 Schematic of the typical load-displacement curve obtained by

nanoindentation [Oliver and Pharr, 1992].
29
2.3 Micromechanical characterisation
2.3.1 Microindentation tests
The microhardness of the materials was obtained using microindentation tests at

room temperature using a durometer AKASHI MVK-EIII (Japan) equipped with
a Vickers indenter and following the ASTM-E384 (2011).
During the tests, a four-sided diamond pyramid indenter with an angle of 136◦
between opposite faces i.e Vickers indenter (fig. 2.8, left) was pressed into the
specimen under a test force (F ) for 12 s. After the indenter was removed from
the surface, the diagonal lengths (d1 , d2 ) of the indentation (fig. 2.8, right) were
measured using optical microscopy. The hardness number (HV ) was obtained by
dividing the applied force by the area of contact (A) between the indenter and
specimen with a correction factor as follows:
F F
HV = = 0.1891 (2.4)
A d1 d2
In this work, tungsten foils were tested under 0.1 Kg (0.98 N), whereas
bulk-tungsten alloys were also tested under 1 Kg (9.8 N) to determine the influence
of the applied load on the microhardness.
Fig. 2.8 Geometry of the Vickers indenter showing its 136◦ angle between
opposite faces (left) and the indentation produced in the material (right).
30
2.3.2 Microtensile tests
In situ tensile tests were performed at room temperature on notched foils using
a miniaturised tensile testing machine (fig. 2.9) from Kammrath & Weiss GmbH
(Germany) installed inside the FE-SEM.
Fig. 2.9 Miniaturised tensile testing machine used for the in situ tensile tests.
Specimens of 30 mm in length and 2 mm in width were cut from foils of

thickness 100 µm by refrigerated electro-discharge machining (EDM). After that,
sharp laser notches were introduced into the specimens, as explained in chapter 3.
These tensile tests on the notched foils were performed using a load cell of
10 kN and a displacement control of the cross-head rate of 0.05 µm/s until
failure. During the tests, videos of the fracture process and the load-displacement
data were recorded. The fracture toughness values were calculated as explained in
chapter 3 using the formulas of Tada et al. (2000).
31
2.4 Macromechanical characterisation
2.4.1 Impulse excitation technique
The impulse excitation technique (IET) is a non-destructive method appropriate

for determining the elastic dynamic modulus of slender bars and rods of materials
that are elastic, homogeneous and isotropic. The specimens are subjected to little
strains so the modulus is measured at or near the origin of the stress-strain curve
where there is only a remote possibility of fracture. Depending on the geometry
of the display used, the elastic modulus (E), shear modulus (G) and Poisson
modulus (ν) can be obtained.
To determine the elastic modulus at room temperature, materials were tested

using prismatic samples in flexion with a Grindosonic MK5 Industrial (Belgium).
The samples were supported on two knife-edges placed at the standing wave nodes
that occur at 0.224 of the length (L) from each end. This type of support allow the
freely flexion of the bar in and out of its plane. To produce the natural vibration
frequencies (ff ) of the material, the prismatic bars were knocked with a low impact
tool that resulted in an excitation with a wide spectrum of associated frequencies.
The sample, due to the geometrical configuration of both the testing fixture and
the specimen, only oscillates at its flexure frequency since the entire system acts
as a filter for the rest of the frequencies.
The frequencies in flexure (ff ) that were detected using a piezoelectric

or acoustic detector were inserted in the following formula proposed by the
ASTM-E1876 (2009) to obtain the elastic modulus of the material:
mff2 L3
E = 0.9465 T (2.5)
b t3
where m is the mass, b is the width, t is the thickness of the sample and T is a
correction factor that depends on the relationship L/t and ν and can be obtained
from ASTM-C1259 (2014).
32
In addition, a model of maximum values [Alger, 1997] was used under the
hypothesis of isodeformation to obtain an approximation of the theoretical elastic
modulus (Eth ):
Eth = Ew Vw + Ea Va + Eb Vb + Eair Vair (2.6)
where Ew , Ea and Eb the elastic moduli and Vw , Va and Vb are the volumetric
fraction of the corresponding component in the alloys. The subscript a refers to
vanadium and titanium and the subscript b refers to the oxides Y2 O3 or La2 O3 .
2.4.2 Three-point bending tests
Non-standard TPB tests were performed on smooth and laser-notched bend bar
specimens to determine flexural strength and fracture toughness respectively. Bulk
specimens with nominal dimensions of 1.6 1.6 25 mm3 were tested in air and
under vacuum environment over an extended high temperature range. Meanwhile,
foils with nominal dimensions of 10 15 0.2 mm3 were tested in air at room
temperature. TPB tests were preferred to four-point bending tests because of the
small amount of material provided. Furthermore, the maximum bending moment
in TPB tests is concentrated at the specimen mid-point, which reduces the stress
on the rest of the surface.
The bending fixture of the electromechanical universal testing machines was

connected to an actuator and the frame using alumina load bars. The tests were
performed with a load cell of 1 kN and displacement control at a constant
cross-head speed of 100 µm/min. High temperature tests were performed with a
heating rate of 30 K/min and a dwell time of 10 min to thermally stabilise the
system.
Tests at 77 K were performed in an Instron 3369 (United Kingdom) equipped

with a device surrounded by a polyurethane isolator to immerse the samples in
liquid nitrogen under load (fig. 2.10).
Tests in air (oxidising atmosphere) were performed in a universal testing

machine (Instron 3369) equipped with a small furnace to perform tests between
300 and 1273 K (fig. 2.11, left). Tests under vacuum (inert atmosphere) were
performed in an Instron 8501 equipped with a high temperature furnace
(fig. 2.11, right) at a pressure of approximately 10−6 mbar and temperatures
33
between 673 and 1473 K. Three to five samples were tested for each material and
condition to ensure repeatability of the results.
Fig. 2.10 Experimental set-up for TPB tests at 77 K (liquid nitrogen

immersion).
Fig. 2.11 Experimental set-up for TPB tests at high temperature in air (left)
and under vacuum (right).
34
2.4.2.1 Flexural strength tests
To measure the flexural strength of the materials, TPB tests were performed on
smooth specimens with a load span (Ls ) of 16 mm and the dimensions shown in
fig. 2.12. The tests were conducted in air and under vacuum atmosphere over a
temperature range between 77 and 1473 K, as explained above (fig. 2.11).
L = 25 mm
Ls = 16 mm
B = 1.6 mm
D = 1.6 mm
Fig. 2.12 Schematic illustration and nominal dimensions of the smooth TPB
test specimen for flexural strength measurements.
Based on the force-displacement data obtained from the TPB tests and the
standard Euler-Bernoulli equations for slender beams up until failure [Timoshenko,
1955], σ- curves were computed to obtain the maximum tensile bending strength
at the fracture point:
3 Fmax Ls 6δD
σf = ; = (2.7)
2 BD2 L2s
where σf is the flexural strength, Fmax is the maximum applied load, Ls is the
support span, is the deformation and δ is the measured displacement. The values
of B and D, the width and height of the specimen, respectively, were measured
for each sample.
Because this method is restricted to the elastic bending regime, the 0.2% yield
strength offset criteria was reported when the material exhibited ductile behaviour,
particularly at higher temperatures.
35
2.4.2.2 Fracture toughness tests
TPB tests were performed on laser-notched samples with a 8.5 mm load span
(Ls ) (fig. 2.13) in air and under vacuum atmospheres over a temperature range
between 77 and 1473 K.
L = 25 mm
Ls = 8.5 mm
B = 1.6 mm
D = 1.6 mm
a 250 µm
Fig. 2.13 Schematic illustration and nominal dimensions of the SELNB TPB
test specimen for fracture toughness measurements.
As reported in chapter 3, single edge laser-notched beam (SELNB) specimens

with notch root radii of approximately 5–20 nm and crack lengths (a) of around
250 µm were used. However, the precise notch length was measured for each
specimen using FE-SEM after testing.
The formula 2.8 provided by Guinea et al. (1998) was applied to calculate the
stress intensity factor in mode I and the resultant fracture toughness values (KIC )
from the failure load and the beam section of each specimen.
a a 2 a 3 a 4
3 Fmax Ls p A − C + E − F + G
KIC = πa D D D D (2.8)
2 3
2 D a 2
2a p
1− 1+ π
D D
where Fmax corresponds to the failure load, D is the height of the specimen and
A, C, E, F and G are calculated as follows:
D D D
A = 1.989 − 0.356 ; C = 1.217 − 0.315 ; E = 3.212 − 0.705
Ls Ls Ls
D D
F = 3.222 − 0.020 ; G = 1.226 − 0.015
Ls Ls
36
2.5 Other physical characterisations
2.5.1 Density
The experimental density (ρexp ) was measured via Archimedes’ method with
immersion in high purity ethanol at room temperature. Furthermore, the
theoretical density (ρth ) was calculated using the following formula:
ρw ρa ρb
ρth = 100 (2.9)
(100 A B)ρa ρb + Aρw ρb + Bρw ρa
where ρw , ρa and ρb are the density of the corresponding component in the alloy
and A and B are the weight % (wt.%). The subscript a refers to vanadium or
titanium and the subscript b refers to the oxides Y2 O3 or La2 O3 .
Based on previous results of experimental and theoretical density, the relative

density (ρr ) was calculated using the relationship:
ρexp
ρr (%) = 100 (2.10)
ρth
And it was therefore possible to estimate the porosity of the materials:
porous (%) = 100 ρr (2.11)
2.5.2 Oxidation tests
Supplementary high temperature tests were performed in dry air to determine the
oxidation behaviour of the bulk-tungsten alloys. Although thermal degradation
due to oxide evaporation was not expected until 873 K, the temperature at which
tungsten is highly reactive with oxygen [Ulick, 1976], the tests were performed at
temperatures ranging from 673 to 1273 K.
To fulfil these experiments, the target prismatic samples were polished and
cleaned with ethanol before the initial measurement and weight. The samples were
37
then heated inside a furnace until the set temperature was reached and held for
10 min. This methodology was intended to emulate the performance of the TPB
tests at high temperatures. Once the samples were cooled in air they were weighed
again. Given the dimensions of the exposed area and the initial and final weights,
the mass gain was determined for each temperature and material.
Finally, the cross-sections were embedded in resin to protect the grown oxide
scale, and were prepared by conventional metallographic techniques to be examined
by FE-SEM.
38
CHAPTER 3
Advanced Experimental
Methods: Fracture
Toughness Measurements
The purpose of the work presented in this chapter is to implement a reliable method
for accurately measuring the real values of the critical stress intensity factor i.e.
the fracture toughness. As previously explained, one of the main drawbacks that
compromises the application of tungsten as structural material is its inherent
brittleness. Therefore, improving this critical feature relies on accurately measuring
its fracture toughness as a real property independent of geometrical parameters
and the method used to introduce the crack. Here, a study of the effects of the notch
geometry on the fracture toughness is performed in both bulk and foil tungsten
materials.
3.1 Bulk-tungsten samples
A nanostructured bulk-tungsten alloy (fig. 3.1) with a composition of

W-4 wt.% Ti-1 wt.% La2 O3 (hereafter referred to as W4Ti1La) was used for the
experiments.
39
Chapter 3. Advanced Experimental Methods: Fracture Toughness Measurements
The first technique considered to measure fracture toughness was mechanical

fatigue. For metals, this is the most common and reliable method to initiate
and grow controlled and reproducible cracks. This technique, so-called single
edge precraked beam (SEPB), is performed from straight notched specimens;
the methodology prevents plastic damage while producing real cracks [Anderson,
2004]. Unfortunately, SEPB has certain restrictions when it refers to extremely
brittle materials such as tungsten and its alloys. The main problem encountered is
that the fatigue stress necessary to nucleate the crack is very close to the fracture
threshold stress and an accurate control is very dependent on the preparation
of the specimen. As a result, the probability of introducing reproducible sharp
notches close to a crack without a high loss of material is reduced [Morrell, 2006;
Achilles and Brøndsted, 2007] and a more functional and reproducible alternative
is required to obtain accurate measurements.
1μm
Fig. 3.1 Microstructure of the W4Ti1La nanostructured alloy.
Numerous authors have studied over the years alternative techniques to

measure fracture toughness in ceramics and hard materials. However, these
methods are not as well established for brittle metals. A common method
for determining fracture toughness in such brittle materials is indentation
microfracture [Shetty et al., 1985]. Nevertheless, based on previous experience,
the application of this method is not possible. Although tungsten is a very brittle
material, cracks did not growth evenly from the indentation corners even at high
40
loads (fig. 3.2). The result is often chipping due to annular subsuperficial cracks
around the indentation or irregular crack growth in some of the corners.
5μm
Fig. 3.2 Vickers microindentation at a load of 9.8 N. The cracks irregularly

grew from some of the corners.
Another alternative method for conductive materials is single-edge notched

beam (SENB) specimens introduced with EDM. However, this method was
discarded a priori because it induces extend thermal damage and significantly
modifies the microstructure [Muthuramalingam and Mohan, 2014].
3.1.1 Techniques to introduce the notch
Four types of methods that gradually approach a crack-like notch were used to
introduce the notches on bend bar specimens of a nanostructured W4Ti1La alloy.
The methods were compared to analyse the influence of the notch root radius on
the measured fracture toughness.
41
3.1.1.1 Single edge notched beam specimens using a diamond disc
In this method, SENB specimens were produced using a refrigerated diamond disc
(named SENB-D specimens) of 400 µm thick and operated at 1800 rpm [Wang
et al., 1992]. This method has been used commonly over time for brittle metals
and ceramics because it is easy, quick and inexpensive. However, it also involves
significant disadvantages:
The introduced notch root radius is approximately 250 µm so the notch
appearance is far from being crack-like (fig. 3.3, left).
The grains just below the notch root radius are severely deformed as a
consequence of the disc friction and pressure over the specimen during the
cutting process (fig. 3.3, right).
For these reasons, the SENB-D method is only suitable for rough fracture
toughness evaluations.
not
chr
oot
200μm 1μm
Fig. 3.3 Micrographs of a broken SENB-D specimen (left) along with its
fracture surface (right). Deformation of the grains around the notch root can be
observed.
3.1.1.2 Single edge notched beam specimens using a diamond wire
In this second method, SENB specimens were produced using a diamond wire
(named SENB-W specimens) of 130 µm diameter and a lower cutting speed than
that of the diamond disc. The method is meant to reduce the notch root radius
42
while preserving the simplicity of the technique and its speed. Unfortunately, it
has also some drawbacks:
The introduced notch root radius is approximately 70–80 µm so the notch
appearance is slightly reduced but is still far from being crack-like (fig. 3.4,
left).
Deformation of the grains below the notch root radius is reduced but can be
still observed (fig. 3.4, right).
A slightly smaller thread wire diameter can also be employed. Even so, the feed
rate will be greatly reduced along with the lifespan of the wire, which will made
the process too expensive and sluggish. The feed speed in this case is controlled
manually by adjusting the force of the wire over the specimen. This process should
be done carefully since excessive pressure will sag the wire and produce a not
straight notch. Special care should be taken to avoid this effect, since it can result
in inaccurate measurements of the fracture toughness. Therefore, this method is
suitable, as in the previous case, for rough fracture toughness evaluations.
not
chr
oot
def
ormedgr
ains
200μm
Fig. 3.4 Micrographs of a broken SENB-W specimen (left) along with its
fracture surface (right). Some deformation around the notch root can be still
observed.
3.1.1.3 Single edge V-notched beam specimens
Single edge V-notched beam (SEVNB) specimens were experimentally produced.

In this method, a sharp razor blade of 150 µm thick that was regularly impregnated
with diamond paste of 1 µm grain size wears the material down because of the
43
friction [Kübler, 1999]. Although the use and efficiency of this method is well
established and verified among ceramics [Nose, 1988; Kübler, 1999; Gogotsi, 2003;
Morrell, 2006; Picard et al., 2006; Fischer et al., 2008; Garcı́a-Prieto and Baudı́n,
2012], only Rupp et al. (2010) and Palacios et al. (2013) have applied the method
to metals, specifically tungsten. The success of the method with ceramics has
been attributed to the fact that these type of materials are more abrasive than
the metallic blade. As a result, the blade wears away with the abrasive diamond
paste during notching, which results in progressive sharpening of the blade and the
generation of a notch tip radius around a few microns. In metals, however, that
materials in contact with the blade experience wear at roughly the same level; this
process blunts the blade and consequently increases the notch radius. Additionally,
the blunted blade wears faster than the sample and inhibits the creation of sharp
notches [Palacios et al., 2013].
To achieve these V-notches, a razor blade machine that operates like a

tribometer was specially designed (fig. 3.5). In this machine, the specimen is
placed in a clamp on a ground plate and the razor blade lays within a guide-way
above (fig. 3.5, right). The upper part can move up and down to apply a normal
force against the mounted specimen that moves linearly back and forth with
the adequate frequency to create the V-notch. Meanwhile, the lower part moves
linearly backwards and forwards with the adequate frequency selected by the
operator.
To obtain the best results, i.e. the shortest preparation time and the sharpest
notch radius, several factors were considered:
The microstructure of the material (e.g. porosity) and physical properties
such as elastics modulus, hardness and toughness. They vary for each
material and strongly influence the rest of factors.
An excessive reciprocating frequency of the linear movement may cause a
widening of the notch because of the vibrations of the machine or even the
breakage of the material; a deficient frequency increases the processing time.
The normal force, together with frequency, is the most important factor. An
overload leads to the breakage of the material or the razor blade; a deficient
force leads to an increase in the processing time.
The sharpness of the razor blade decreases over time; blunt razor blades need
to be changed several times during the operation (typically every 15–20 min).
The diamond paste used as an abrasive must be replenished several times
throughout the operation. For the first steps, diamond paste of 1 µm grain
size was used. However, in the last change of razor blade it was proved that
44
the sharpest notch root radius was obtained when the diamond paste grain
size was reduced and its concentration was increased.
A proper cleaning between razor blade changes is critical to remove leftover
material that may blunt the notch root radius or even impede its progress.
Special care should be taken before the last step of the machining. Once the
process is fully completed, the specimens should be cleaned in an ultrasonic
bath to remove any diamond grit still embedded in the notch tip.
r
azorbl
ade
s
peci
men
nor
mal
for
ce
l
i
nearmov
ement
Fig. 3.5 Razor blade machine to create the SEVNB specimens. Details of the
position of the razor blade and the specimen are shown on the right side.
45
Although the process is time-consuming because these parameters need to be

optimised for each material and the subsequent notching process may last one or
two days, the results were successful:
The introduced notch root radius are very sharp, approximately 5—7 µm
(fig. 3.6, left).
Evidences of grains deformation below the notch root radius were not found
(fig. 3.6, right).
In the case of coarse grain materials, the fracture toughness values obtained
using the SEVNB specimens determine the real fracture toughness values in
agreement with the SEPB data [Gogotsi, 2003]. Therefore, the SEVNB method
cannot be applied for materials with grain sizes smaller than the notch root radius
(5 µm), as in this case.
For materials with fine microstructure in which the grain size is much smaller
than 1 µm, the SEVNB method overestimates the fracture toughness values.
The critical notch root radius should be roughly the same size as the relevant
microstructural features [Damani et al., 1996; Fischer et al., 2008] and the
dependence on the notch root radius increases as the grain size decreases [Carolan
et al., 2011]. For these reasons, the application of this method is unacceptable for
nanostructured materials.
not
chr
oot
100μm
Fig. 3.6 Micrographs of a SEVNB specimen (left) along with its fracture
surface (right). No evidence of deformation around the notch root was observed.
46
3.1.1.4 Single edge laser-notched beam specimens
In the SELNB method, a femtosecond laser is used to machine ultra-sharp notches

on the specimens [Pastor, 2002; Pastor et al., 2008]. The laser conditions were
optimised to obtain the sharpest notch root radius together with minimal thermal
damage prior to implementation of the method. Although the selection of the
proper parameters for the pulsed laser requires time, the process is less tedious
than the SEVNB method since once the parameter optimisation is set, the notching
process is quite fast and its use is appropriate for similar materials [Palacios et al.,
2014; Tejado et al., 2015].
As a result of the notching process, extraordinarily results were achieved:

Straight notches with a length of approximately 250 µm (fig. 3.7) and sharp
notch root radii around 5–20 nm (fig. 3.8, left).
No evidence of a thermally affected zone close to the notch root radius
(fig. 3.8, right).
50μm
Fig. 3.7 Micrograph of a SELNB notch.
Some authors such as Turon-Vinas and Anglada (2015) have reported that
this method applied to ceramics results in an affected zone with microcracks and
pores in the vicinity of the laser notch. Thus, the authors added the length of
the affected area to the initial value that was measured to calculate the fracture
toughness. For the studied materials here, the thermal damage was minimal or
even non-existent as a result of the optimisation process and the good conductivity
of tungsten. This is because the femtosecond laser works in non-linear optical
conditions and the action of the laser transforms the material directly into plasma
by an ablation process. Therefore, the transfer of heat energy to the solid is
47
minimal. As a result, evidences of an affected zone were not found (fig. 3.8, right).
The increase in thermal conductivity of tungsten compared with ceramics may
be a determining factor to avoid an affected zone. Further evidence of material
without thermally affected zone, is that the initial notch length measured on the
side and the final notch length from the fracture surfaces have the same value.
2μm
Fig. 3.8 Image of a SELNB notch root based on profilometer analysis (left)
along with its fracture surface (right).
This method exhibits obvious advantages such as good reproducibility, high

accuracy and precision for reliable fracture toughness testing as in the case of
structural ceramics studied by Zhao et al. (2014). The notches that have been
produced are to all intents like real cracks, with the drawback of their higher
price.
3.1.2 Influence of the notch root radius
After the notching process, TPB tests were performed at room temperature on
a universal testing machine (Instron 3369) using a displacement control rate of
100 µm/min and a load span of 8.5 mm. Additional details about the testing as
well as the applied stress intensity factor formula can be found in chapter 2.
The mean values obtained from the TPB tests (Table 3.1) reveal the strong
dependence of the notch root radius on the fracture toughness. While values from
p
blunt SENB specimens are nearly the same (around 6 MPa m), the results for
48
p
SEVNB specimens decreased by 13% to 5.2 0.3 MPa m; the results are even
p
more reduced for the SELNB specimens (to 4.4 0.1 MPa m). This influence
may be a consequence of the interaction of a distorted stress field with material
flaws in front of the notch [Damani et al., 1996].
Table 3.1 Mean values of the measure fracture toughness with the standard
error for the four methods.
SENB-D SENB-W SEVNB SELNB

radius (µm) 200 70–80 5–7 <1
p
KIQ (MPa m) 6.1 0.6 6.0 0.7 5.2 0.3 4.4 0.1
The SELNB specimens have a notch root radius of the same order of magnitude
as the grain size of the W4Ti1La alloy. Therefore, the measured fracture toughness
is considered to be a theoretical limit for this material according to the relation
exposed by Damani et al. (1996). These authors stated that if the notch root radii
are close to infinitesimal values, the fracture toughness will be approximately equal
to the theoretical value.
Consequently, the results obtained from the SELNB specimens yield the real
fracture toughness (KIC ) value. With it, the fracture toughness (KIQ ) results for
the SENB-D, SENB-W and SEVNB specimens were normalised (fig. 3.9). The
curve fit shows a clear tendency towards the ratio KIQ /KIC being equal to one,
in accordance with the results of Damani et al. (1996).
Another difference in the results is the progressive reduction in the dispersion

of the values. Such dispersion stats at 10% for the SENB specimens, becomes 6%
for the SEVNB specimens and is then only 2% for the SELNB specimens; this
trend follows the same pattern as the notch root radius progressive sharpening. It
is attributed to the fact that the SELNB method is less invasive and harmful to
the material. As a result, repetitive notches are easier to perform.
Considering therefore that the values of the SELNB specimens are the real
fracture toughnesses, the SENB specimens can be only considered for rough
toughness evaluations since the results are approximately 40% overestimated.
The results of the SEVNB specimens, however, exhibited an overestimation of
20%. Even so, despite the difficulties with the razor blade control, the SEVNB
technique produced results with low dispersion. Thus, the SEVNB method can
be used to evaluate materials with coarse microstructures that do not require
the use of more advanced and expensive techniques such as SELNB. As was
49
discussed in Palacios and Pastor (2015), the SELNB method exhibits obvious
advantages such as high accuracy and speed, good reproducibility and precision
for reliable fracture toughness measurements of nanostructure brittle materials.
Unfortunately, this technique also involve higher costs. For these reasons, and
taking into account the fine microstructure of the studied tungsten materials and
alloys, the SELNB method was used throughout the entire investigation.
Fig. 3.9 Influence of the notch root radius on the fracture toughness.
50
3.2 Tungsten foils
In this section, the geometry used to measure the fracture toughness through in
situ tensile tests with a micromachine (section 2.3.2) is discussed. tungsten foil
specimens with dimensions of 30 2 0.1 mm3 were cut by EDM. Afterwards
the femtosecond laser notches were introduced into the samples. The notches had
lengths of approximately 250–300 µm and very fine notch root radii of less than
1 µm.
Since the fracture toughness measurements were not easy to perform due to
the small size of the samples and their slenderness, two different geometries were
used to check which one provided the most accurate results.
3.2.1 Geometry selection
3.2.1.1 The double-edge notched specimen
The geometry of the double-edge notched specimen (fig. 3.10) was used to measure
the fracture toughness of the tungsten foils because it enables more stable testing
since the specimens are symmetric.
a a
h
b b
σ
Fig. 3.10 Geometry of the double-edge notched specimen.
51
The introduced femtosecond laser notches exhibited very similar notch lengths
of approximately 300 µm. However, aligning two notches at such small sizes is
challenging; in most of the cases both of the notches were slightly misaligned.
Another important factor that hinder alignment is that the specimens are small
and very slender.
Based on the force-displacement data obtained from the tensile tests and the
specific dimensions of the specimens, the following equations from Tada et al.
(2000) for double-edge notched test specimens were applied:
p a
KI = σ πa F (3.1)
b
where F(a/b) is a parameter that depends on the geometry of the specimens.

a
For these specimens 0.3, therefore the modification of Benthem’s formula
b
(eq. 3.2) was applied. This modification achieves an accuracy of 0.5% for any a/b.
a a 2 a 3 a 4
a 1.122 0.561 0.205 + 0.471 0.190
F = b rb b b (3.2)
b a
1
b
3.2.1.2 The single-edge notched specimen
The geometry of the single-edge notched specimen (fig. 3.11) was used because it
was easier to introduce in the specimen. The introduced femtosecond laser notches
exhibited lengths of approximately 250 µm. From the force-displacement data
obtained from the tensile tests and the specific dimensions of the specimens, the
following equations from Tada et al. (2000) for single-edge notched test specimens
were applied:
p a
KI = σ πa F (3.3)
b
52
where F(a/b) is a parameter dependent on the geometry of the specimens. For

a
these specimens 0.3, therefore the least squares fitting method from Gross
b
(1964) was applied. This method achieves an accuracy of 0.5% for a/b 0.6.
a a a 2 a 3 a 4
F = 1.122 0.231 + 10.550 21.710 + 30.382 (3.4)
b b b b b
b
h
σ
Fig. 3.11 Geometry of the single-edge notched specimen.
53
CHAPTER 4
Bulk-Tungsten Materials
This chapter evaluates the microstructural, mechanical and fractographical

behaviour of five bulk tungsten-based alloys and a pure tungsten material used
as a reference. Mechanical tests, among other micromechanical and physical
characterisations, were performed in air and under vacuum atmosphere over a
temperature range between 77 K (via immersion in liquid nitrogen) to 1473 K. The
fracture surfaces were thoroughly analysed for each test condition to correlate the
macroscopic behaviour with the micromechanisms of failure involved and finally
determined the influence of each alloying element on pure tungsten.
All of the materials were provided by the Physics Department of the Carlos III
University of Madrid (UC3M) for their characterisation. Although the fabrication
process of the materials will be briefly described in the following section, further
details can be found in Muñoz et al. (2011) since the manufacturing process is not
the primary concern of this thesis.
4.1 Materials and samples
Five bulk tungsten alloys (W1Y, W2V0.5Y, W4V0.5Y, W2Ti1La and W4Ti1La)
with the weight (wt.%) target compositions listed in table 4.1 and a pure tungsten
material were processed following a powder metallurgical route.
55
Chapter 4. Bulk-Tungsten Materials
Table 4.1 Weight % composition of the bulk alloys.
alloy W V Ti Y2 O3 La2 O3
pure W 100 – – – –
W1Y 99 – – 1 –
W2V0.5Y 97.5 2 – 0.5 –
W4V0.5Y 95.5 4 – 0.5 –
W2Ti1La 97 – 2 – 1
W4Ti1La 95 – 4 – 1
Each alloy was prepared by mixing together the starting powders (table 4.2)
of every target composition for 4 h. The powder mixtures were then processed by
MA inside a sealed WC vessel with a high purity argon atmosphere for 20 h in
a high-energy planetary mill. WC balls of 10 mm diameter were used as grinding
media with a ball-powder ratio of 4:3. After that, the alloyed powders were
canned and degassed at 673 K for 24 h under vacuum before sealing. All of the
powders (starting, blended and alloyed) were manipulated under a high-purity Ar
atmosphere using a glove box. Finally, HIP was conducted at 1573 K for 2 h at a
pressure of 195 MPa to consolidate the canned powders.
Table 4.2 Features of the starting powders.
average particle
powder purity (%)
size (µm)
W 99.9 12
V 99.5 20
Ti 99.8 20
Y2 O3 99.5 0.01–0.05
La2 O3 99.9 0.01–0.03
As a result of this process, cylinders with dimensions of 30 mm in diameter and

50 mm in length were obtained. These cylinders were cut by refrigerated EDM to
obtain bend bar specimens with nominal dimensions of 1.6 1.6 25 mm3 that
were used to perform all measurements and experimental methods.
56
4.2 Microstructure at room temperature
A microstructural analysis of the bulk alloys at room temperature was performed

by using high resolution microscopy. To that end, the sample surfaces were
prepared as explained in section 2.1. In this part, information about the chemical
composition and grain size distribution from EDX and EBSD microanalyses,
respectively, was also displayed for some of the alloys.
The EBSD analysis was performed by applying a criterion of discrimination

of crystallographic misorientation more than 5◦ between adjacent crystalline
domains. Only results with a high percentage of indexed patterns, i.e. a high
image quality, were considered. It was accordingly possible to apply the noise
reduction procedure to suppress the zero solution produced by unindexed patters
and to conduct reliable microstructural characterisation, otherwise, low indexed
patterns led to an unreliable characterisation. For the grain size measurements, a
minimum of 1000 complete grains per single map were considered. This method
to measure grain size is more favourable than ASTM-E112 (2013) because the
distribution size rather than mean value can be more precisely measured. This
method is also preferable to image analysis because although both methods can
provide information about grain morphology, EBSD is more accurate and yields
the volume fraction for each population of grains. Furthermore, the analysis of
ultrafine grained alloys like these ones is unfeasible using previous methods.
4.2.1 Microstructure of pure tungsten
The microstructure of pure tungsten consists on a relatively well-mixed

heterogeneous distribution of equiaxial grains and intergranular porosity (fig. 4.1).
The volume fraction distribution for each population, based on reconstructed

EBSD data, exhibits a mean size of approximately 1 µm (fig. 4.2) that also
represents the most probable population (20%). In addition, even bigger grains
occupy 40% of the sample. Meanwhile, the percentage of grains smaller than
100 nm is negligible (less than 3%). The grain morphology (fig. 4.3) is in accordance
with the micrographs (fig. 4.1) although the grains look coarser in the micrographs.
In the image of the EBSD data, the crystallographic misorientation was taken into
account to show the real grain size and morphology.
57
5μm
Fig. 4.1 Microstructure of pure tungsten.
Fig. 4.2 Histogram of the equivalent grain size distribution for pure tungsten
[Palacios et al., 2015].
58
10μm
Fig. 4.3 EBSD measured tungsten grains as a function of their circle equivalent
diameter for pure tungsten [Palacios et al., 2015]. The white regions correspond
to cavities or V-rich regions.
4.2.2 Microstructure of W1Y
The W1Y alloy exhibits polyhedral tungsten grains with yttria dispersion and
intergranular porosity (fig. 4.4). The tungsten grains exhibit a relatively well-mixed
bimodal size morphology and a distribution of submicron and coarse grain
populations centred at 420 nm and 5.3 µm, respectively [Palacios et al., 2014].
Although the yttria nanoparticles are too small to be visible under low
magnification SEM (fig. 4.4), Y-rich grains were identified. The EDX elemental
analysis performed on the darker grains revealed an increasing yttrium and
oxygen content. In particular, for point A, a composition of 41.8 wt.% Y,
37.6 wt.% W and 20.6 wt.% O was obtained; brighter grains are mainly composed
of tungsten. Such Y-rich grains can also be distinguished because they exhibit a
more rounded morphology. Nonetheless, observations of the yttria nanoparticles
are clearer under high magnification FE-SEM as revealed the black patches
shown in fig. 4.5. The EDX analysis reveals the differential concentration of
yttrium within the tungsten grains (fig. 4.6). Yttrium nanoparticles are mainly
located at the grain boundaries as a consequence of the growth of Y-rich
oxides after the HIP process [Veleva, 2011]. However, some conglomeration of
59
nanoparticles was detected with transmission electron microscopy, probably as

a consequence of their deficient dispersion within the matrix [Palacios et al., 2014].
Fig. 4.4 Microstructure of W1Y. A was the point at which EDX elemental
analysis was performed.
200nm
Fig. 4.5 High magnification BSE-SEM image for W1Y (left) with the EDX
spectral data (right). The solid line corresponds to a representative EDX
spectrum for the black areas, and the dotted line corresponds to the
tungsten-matrix [Palacios et al., 2014].
60
*1 Spectrum Y W
*3 1 1.28 98.72
*2
2 1.58 98.42
3 0.52 99.48
4 0.56 99.44
*4
1μm
Fig. 4.6 Micrograph of W1Y (left) with the points from which the EDX
spectral data were obtained (right).
Despite the addition of 1% Y2 O3 , the microstructure of the W1Y alloy did

not substantially change compared with that of pure tungsten (fig. 4.1). This
finding is inconsistent with the results of Ryu and Hong (2003) who reported an
inhibition of the coarsening of tungsten grains during sintering with the addition
of 0.1–5% Y2 O3 . A high porosity and a large population of grains on the order
of a few microns were found for both materials, probably as a consequence of the
deficient dispersion of the ODS nanoparticles within the matrix.
4.2.3 Microstructure of W2V0.5Y
The W2V0.5Y alloy exhibits a fine microstructure that is difficult to distinguish

under low magnification microscopy (fig. 4.7) unlike pure tungsten and W1Y
(figs. 4.1 and 4.4, respectively). However, diverse features can be observed; black
areas with different sizes and morphologies, some of which are micro-sized but
most are submicron, and a grey area distributed around the entire material. At
higher magnifications, however (fig. 4.8), the grey area reveals fine nanostructured
grains with small populations of micro-sized grains. It was also observed that some
grains, regardless of their size, are aligned to form elongated shapes.
After the analysis of the reconstructed EBSD data, the refinement of the
microstructure become more evident (figs. 4.9 and 4.10). The mean grain size is
around 200 nm and a significant fraction of the grains, around 20%, exhibit sizes
below 100 nm in contrast with the residual amount of nano-sized grains found in
pure tungsten (less than 3%). Although some coarse micro-sized grains are still
observed, they represent only the 10% of the volume fraction (i.e. four times lower
than that of pure tungsten).
61
5μm
Fig. 4.7 Microstructure of W2V0.5Y under low magnification.
Fig. 4.8 Microstructure of W2V0.5Y.
62
Fig. 4.9 Histogram of the equivalent grain size distribution for W2V0.5Y
5μm
Fig. 4.10 EBSD measured tungsten grains as a function of their circle

equivalent diameter for W2V0.5Y [Palacios et al., 2015]. The white regions
correspond to cavities or V-rich regions.
63
The EDX microanalysis revealed the composition of the different parts within
the material (fig. 4.11). The big black patches largely correspond to vanadium
pools although some oxygen was also found. Some of them also correspond to
residual porosity since vanadium was not detected. Meanwhile, the grey areas are
mainly composed of tungsten and some dispersed vanadium and yttrium. Since
the W-V equilibrium diagram features total solubility (fig. 1.10), such vanadium
pools should not be emerged; however, they are distributed around the entire
material (fig. 4.7). Their presence can be attributed to small deviations from
equilibrium during processing, presumably due to impurities (such as oxygen)
during processing, as the case for earlier tungsten-alloys reported by Aguirre
(2014).
Fig. 4.11 Micrograph of W2V0.5Y with the corresponding EDX elemental

mapping of W, V, Y and O elements, respectively.
4.2.4 Microstructure of W4V0.5Y
The W4V0.5Y alloy shows a similar microstructure to that of W2V0.5Y. Some

small differences can be found basically in the amount of vanadium pools, which are
now larger and more numerous (fig. 4.12). At higher resolutions, the morphology
of the nano-sized grains can be clearly distinguished (fig. 4.13).
The grain size distribution after EBSD analysis (figs. 4.14 and 4.15) displays an
even more refined microstructure. Although the mean grain size is 200 nm and the
population below 100 nm is around 20% (the same as for W2V0.5Y), the volume
fraction of grains on the order of microns is reduced to 3%, much smaller than
that of W2V0.5Y and even more than that of pure tungsten.
64
5μm
Fig. 4.12 Microstructure of W4V0.5Y under low magnification.
200nm
Fig. 4.13 High resolution micrograph of W4V0.5Y.
65
Fig. 4.14 Histogram of the equivalent grain size distribution for W4V0.5Y
Fig. 4.15 EBSD measured tungsten grains as a function of their circle

equivalent diameter for W4V0.5Y [Palacios et al., 2015]. The white regions
correspond to cavities or V-rich regions.
66
Considering that the W4V0.5Y alloy presents more numerous and larger
vanadium pools, an EDX linescan analysis was performed to clarify its chemical
composition. It was observed from the results (fig. 4.16) that in the black areas
the vanadium content radically increases to more than 90%, it was also detected a
slight increase in the oxygen content, but in any case larger than 10%. Meanwhile,
the presence of tungsten and yttrium was almost negligible.
Fig. 4.16 EDX linescan with the profile content of a vanadium pool of
W4V0.5Y.
4.2.5 Microstructure of W2Ti1La
The microstructure of the W2Ti1La (fig. 4.17) exhibits numerous black areas, some
of which correspond to pores. Even through the black areas cannot be clearly
distinguished in the etched surface (fig. 4.18), the micrograph reveals the fine
microstructure with some coarse micro-sized grains and fine aligned grains. The
large black patches cannot be observed as a consequence of the etching. Similar to
the yttria nanoparticles in W1Y, the lanthana nanoparticles are either dispersed
in the tungsten grains or located at the grain boundaries (fig. 4.19).
67
Fig. 4.17 Microstructure of W2Ti1La under low magnification.
Fig. 4.18 Etched micrograph of W2Ti1La under low magnification.
68
500nm
Fig. 4.19 High resolution micrograph of W-1%La [Muñoz et al., 2012]. The
black areas correspond to La-rich areas.
The EDX mapping microanalysis of the surface (fig. 4.20) reveals that most
of the black areas correspond to titanium pools that have been segregated filling
the pores between the tungsten grains, as in the case of W-V-Y2 O3 alloys with
vanadium pools. The W-Ti binary diagram (fig. 1.11) features complete solubility
in liquid state at temperatures above the HIP temperature and solid solution
solubility around the processing temperatures. For that reason, a complete solid
solution should occur, however, a segregation of titanium is observed, which could
be due to irregularities during the manufacturing process.
4.2.6 Microstructure of W4Ti1La
The W4Ti1La microstructure is very similar to that of the W2Ti1La alloy. It is

not well defined under low magnification (fig. 4.21), although some differences
can be found. The big black areas are composed by titanium as revealed from
EDX chemical microanalysis, but many small black zones around larger grey
areas represent the porosity that has experimented a high increase in this alloy.
At higher magnification, the nano-sized microstructure is revealed (figs. 4.22 and
4.19) and many grains below 100 nm are observed.
69
Fig. 4.20 Micrograph of W2Ti1La with the corresponding EDX elemental

mapping of W, Ti, La and O elements, respectively.
70
5μm
Fig. 4.21 Microstructure of W4Ti1La under low magnification.
1μm
Fig. 4.22 Microstructure of W4Ti1La.
71
4.3 Micromechanical and physical

characterisation
4.3.1 Density
The experimental density (ρexp ) measured via Archimedes’ method, the theoretical
density (ρth ) calculated using eq. 2.9 and the data from table 4.3, and the relative
density (ρr ) calculated using the ratio exposed in eq. 2.10, are compared for the
materials.
Table 4.3 Density of the constituent powders (g/cm3 ).
ρW ρV ρTi ρY2 O3 ρLa2 O3

19.25 6.11 4.51 5.03 6.51
The porosity of the materials was estimated from the values of ρr using
eq. 2.11. The results (figs. 4.23 and 4.24 and table 4.4) show evidence of the porous
materials. This porosity can be attributed to the processing conditions since the
HIP temperature is rather low for pure tungsten and solid phase sintering might
occur [Blaschko et al., 1996].
Fig. 4.23 Experimental (ρexp ), theoretical (ρth ) and relative (ρr ) density of the
bulk materials.
72
pur
eW W1Y
W2V0.
5Y W4V0.
5Y
W2Ti
1La W4Ti
1La
Fig. 4.24 Microstructure of the bulk materials at low magnification. The black
areas correspond to porosity, but also to the metallic components in some of the
alloys (vanadium and titanium pools).
73
The ratio between the measured density and the experimental density estimates
a porosity of approximately 8% for pure tungsten. This rather high porosity is
reduced by the addition of the alloying elements, with the exception of the W1Y
and W4Ti1La alloys. In general, the addition of metallic elements (vanadium or
titanium) decreases the porosity. However the increase in the content of these
metallic elements to 4%, leads to a decay of the densification for these alloys, in
disagreement with previous studies.
Table 4.4 Estimated porosity (%) for the bulk materials.
pure W W1Y W2V0.5Y W4V0.5Y W2Ti1La W4Ti1La

8.4 9.6 2.6 5.4 4.7 11.0
In particular, the increased porosity for W1Y is in disagreement with the results
of Veleva et al. (2009) who reported an increase in the relative density with the
addition of yttria. Meanwhile, the addition of yttria and also vanadium yields an
improvement in the densification. The best value is achieved in W2V0.5Y (2.6%);
although major porosity is observed with an increasing vanadium content, the
result is still lower than that of pure tungsten. W2Ti1La exhibits a porosity lower
than 5%, almost half the porosity of pure tungsten. However, with increasing
titanium content the densification strongly decreases to more than double the
porosity, as in the case of the W-V-Y2 O3 alloys.
4.3.2 Elastic modulus
The elastic modulus was measured using the IET technique (EIET ) and also
nanoindentation tests (En ). Besides, the theoretical elastic modulus (Eth ) was
calculated by applying eq. 2.6 with the corresponding values from table 4.5.
Table 4.5 Volume fraction for the bulk materials, Vx (%).
alloy W V Ti Y2 O3 La2 O3
pure W 100 – – – –
W1Y 96.26 – – 3.74 –
W2V0.5Y 92.23 5.96 – 1.81 –
W4V0.5Y 86.80 11.46 – 1.74 –
W2Ti1La 89.40 – 7.90 – 2.70
W4Ti1La 82.60 – 14.84 – 2.56
74
The elastic modulus (fig. 4.25) shows that the addition of yttria significantly
decreases the modulus since yttria has a much lower elastic modulus (171.5 GPa;
Yeheskel and Tevet (1999)).
Fig. 4.25 Elastic modulus measured via IET (EIET ), instrumented indentation
(En ) and theoretically calculated (Eth ) for the bulk materials.
Even though the elastic modulus for pure vanadium also exhibits low values
(128 GPa; ASM Handbook Vol.02 (1991)), the W-V-Y alloys reveal higher modulii.
The obtained values are slightly higher for W2V0.5Y than for W4V0.5Y with all
of the methods. This result is probably not only because of the mayor vanadium
content but also because of the increased porosity of the alloy. In general, the
discrepancy between W2V0.5Y and W4V0.5Y alloys is slightly more pronounced
than for previously reported W-2 wt.% V and W-4 wt.% V alloys because of the
increased porosity of W4V0.5Y [Palacios, 2010].
The elastic modulii values for the W-Ti-La alloys are similar to those obtained
for pure tungsten. No significant differences were found for either the methods or
the alloy composition. However, the theoretical elastic modulus was not calculated
because the elastic modulus for lanthana was not found.
75
Above all, the differences among the techniques are very small and the
agreement between methods is considered to be good in all the cases.
4.3.3 Hardness
To analyse the influence of the applied load (if any) on the hardness, Vickers
microindentations with two applied loads and Berkovich nanoindentations were
performed.
The results (fig. 4.26) show a small influence of the applied load in the Vickers
microhardness tests; in general, higher values were achieved using the lower load
(0.98 N). These results are comparable to those obtained with instrumented
Berkovich indentations. Approximately 10% higher values were obtained for
nanoindentations with 0.6 N, and also a larger dispersion in the results that may
be due to the porosity of the different phases of the alloys.
Fig. 4.26 Vickers microhardness (HV ) with two different applied loads and
Berkovich nanohardness (nH) for the bulk materials.
In particular, the addition of yttria slightly increases the hardness; however,

the values are still very low, around 3.5 GPa. Meanwhile, with the simultaneous
addition of vanadium and yttria, the alloys exhibit much higher values than
76
those obtained for pure tungsten. No significant differences are found between
both alloys. In contrast, although W2Ti1La experiments an improvement in the
hardness that is slightly higher than that found for the W-V-Y2 O3 alloys, the
hardness is reduced to approximately 8 GPa with increasing titanium content.
4.3.4 High temperature stability in air
The oxidation phenomenon with tungsten materials and alloys is complex due
to the wide variety of oxides that form with the increasing temperature. In this
section, a general overview of the processes is presented for a better understanding
of the mechanical properties at high temperatures in an oxidizing atmosphere.
The mass gain balance after high temperature exposure (fig. 4.27) clearly shows
that the oxidation resistance of tungsten materials decreases above 673 K and
becomes fairly non-existent at 1273 K. The mass gain at this temperature reaches,
in most cases, higher values than at 673 K. Pure tungsten is highly reactive with
oxygen above 673 K, but the oxidation decreases with the addition of some alloying
elements. As expected, the W1Y alloy exhibits an improvement in the oxidation
resistance despite its increased porosity, showing a mass gain at least three times
lower than that of pure tungsten. The rest of the alloys, however, exhibit diverse
behaviour due to the formation of complex oxides but also due to the lack of
densification. In W2Ti1La, unlike the general tendency, mass loss is observed.
This result and the increasing thickness of the oxide scale (table 4.6) suggest that
volatile oxides developed.
Table 4.6 Thickness (µm) of the oxide scale as a function of

temperature.a
alloy 673 K 873 K 1073 K 1273 K

pure W – 5 40–70 200–450
W1Y – – 15–20 50–70
W2V0.5Y – – 100 250
W4V0.5Y – – 150 450
W2Ti1La – – 10–15 45–65
W4Ti1La – – 10–15 55–65
a The values were roughly measured over the micrographs and should be
used only as guidance.
77
Fig. 4.27 Mass gain curves versus temperature for samples oxidised for 10 min
in air. The standard error for each temperature and alloy are also shown but are
too small to be distinguished.
Measurements of the oxide scale on the cross sections (table 4.6) also
evidence that the W1Y and W-Ti-La2 O3 alloys possessed higher protection against
oxidation despite the high porosity reported for W4Ti1La (approximately 10%).
An oxidation layer was not found in any of the materials at 673 K. At 873 K,
however, a thin layer was observed in pure tungsten and a very slight layer in the
W4Ti1La alloy. Above 1073 K the thickness of the oxide scale become sizeable
and at 1273 K is significantly thick (fig. 4.28). The micrographs of the oxide scale
at 1273 K (fig. 4.28) evidence not only a difference in thickness but also in the
structure of the scales.
The oxidation becomes appreciable for pure tungsten at 873 K. At this

temperature, it develops a dense, thin, adherent black layer that provides
some protection, but its formation rate quickly increases with temperature
and gradually becomes overlaid with a bright yellow deposit (WO3 ) [Kellett
78
and Rogers, 1963]. This layer becomes thicker with increasing temperature and
develops into a coarse, non-protective oxide scale that cracks to release stresses
from oxide growth (fig. 4.28) [Cifuentes et al., 2012a]. This yellowish, porous,
powdery layer is highly volatile due to its low sublimation temperature at 1973 K.
Differences between these two oxides (black and yellowish) seem to be due to
structural perfection and orientation rather than composition [Kellett and Rogers,
1963].
Fig. 4.28 Oxide scale developed for the bulk materials in air after 10 min of
exposure at 1273 K. The magnification of the micrographs differs depending on
the degree of degradation of the alloy.
79
W1Y develops a highly adherent, coherent protective scale that reduces further
oxidation at high temperature and passivates the surface against chemical and
electrochemical corrosion at room temperature [Borchardt, 1964]. Although this
protective layer was observed at higher temperatures, it becomes thicker and
oxygen solubility become appreciable at 873 K. Cifuentes et al. (2012b) reported
a protective effect only up to 873 K as well, because microcracking of the scale at
higher temperatures allows oxygen transportation inwards the material. Even so,
the results show an increase in the mass gain that is still four times lower than
that of pure tungsten with fewer microcracking of the oxide scale.
W-V-Y2 O3 alloys do not exhibit an oxide scale at 673 K nor even at 873 K.
However, at 1073 K they exhibit a non-adherent layer that is thicker than in
the rest of materials. Such a layer should not be very dense because oxidation is
produced in a greater extent in W4V0.5Y with a higher porosity. The evolution
of the oxide layer in the W2V0.5Y alloy (fig. 4.29) shows that oxidation occurs in
a preferential shape at the edges, which are progressively split and curved as the
temperature increases. Such behaviour is typical for pure tungsten [Gulbransen
and Andrew, 1960] and was also observed for the W4V0.5Y, W2Ti1La and
W4Ti1La alloys. Although the images were obtained by FE-SEM and colours
cannot be distinguished, they show the typical powdery, yellowish oxide scale.
500μm
873K 1073K 1273K 1373K
Fig. 4.29 FE-SEM images of the evolution of the oxide scale in the W2V0.5Y
alloy fracture surfaces with increasing temperature.
Finally, the W-Ti-La alloys exhibit a peculiar phenomenon that is also observed
in pure tungsten at high temperatures (around 1500 K): the high volatility of WO3
lead to mass loss [Gulbransen and Andrew, 1960]. As a consequence, the outer
yellow scales of the alloys remain thinner and the mass gain is lower across the
entire the temperature range. At 873 K the scale is quite dense in contrast with
that of the W-V-Y2 O3 alloys. At 1073 K it remains thin and cracks are observed
due to the stresses produced during the oxide growth.
80
4.4 Fracture behaviour
The fracture behaviour was studied using TPB tests performed on smooth and
SELNB specimens to obtain flexural strength and fracture toughness, respectively,
and the subsequent analysis of their fracture surfaces. Such tests were performed
in air and under vacuum atmospheres over a temperature range from 77 to 1473 K
following the conditions explained in section 2.4.2. The results for pure tungsten,
used as a reference and those of each alloy (W1Y, W2V0.5Y, W4V0.5Y, W2Ti1La,
W4Ti1La) are presented following the scheme:
i) Flexural strength. The TPB test results for the smooth samples and the
specific measurements for each specimen were used to calculate σ- curves
applying the Euller-Bernoulli equations (eq. 2.7). In the first plots, σ- curves
for each test temperature in air and under vacuum are shown. After that, the
average flexural strength values with their corresponding standard error are
plotted along the studied temperature range for the air and vacuum tests.
The curves are considered to exhibit ductile behaviour when the 0.2% yield
strength offset criterion is satisfied. In such cases, the values are represented
using open symbols and dashed lines, otherwise the curves will be consider
as brittle fracture. The DBTT is then defined as when the σ- curves change
their behaviour from linearly elastic to plastic for a given strain rate that in
this case is 100 µm/min.
ii) Fracture toughness. The TPB test results for the SELNB specimens and
the specific measurements of each sample were used to calculate the resultant
fracture toughness by applying eq. 2.8. The flexural force-displacement
curves obtained from each temperature in air and under vacuum are only
shown in some cases because they look very similar. Afterwards, the average
fracture toughness with its corresponding standard error is shown for the
studied temperature range for the air and vacuum tests.
iii) Fracture surfaces. After the TPB tests, every fracture surface was analysed
with a FE-SEM, although only selected temperatures are presented. In
general, surfaces at 300 K, 873 K (the temperature at which thermal
degradation is usually present) and 1273 K are shown, although additional
fractographies are also displayed in certain cases. On the whole, the
micromechanisms of failure involved are intergranular and transgranular
fracture. Intergranular fracture, caused by grain boundary decohesion, was
observed in most cases due to the fragile nature of the materials. Meanwhile
transgranular fracture, produced through the grains, was less common but
was also observed under the DBTT in some fracture surfaces. Transgranular
81
fracture produced along the crystalline planes is known as cleavage; in the

case of many bcc metals cleavage is mostly produced along the f100g plain
[Gumbsch, 2003].
4.4.1 Fracture behaviour of pure tungsten
i) Flexural strength
The flexural σ- curves (fig. 4.30) showed linear elastic behaviour in both air and
under vacuum until fracture over almost the entire temperature range. Although
the results at 1273 K in air exhibited plastic behaviour, the strain did not
satisfy the 0.2% yield strength offset. In contrast, the specimens tested at 1473 K
under vacuum exhibited plastic behaviour and macroscopic deformation without
breakage.
Fig. 4.30 Flexural σ- curves at different temperatures for pure tungsten in air
(left) and under vacuum (right). Note the different scales of the axis for visual
clarity.
The average flexural strength as a function of temperature in air (fig. 4.31, left)
revealed degradation with increasing temperature. Although at 473 K there was an
unexpected increase in the flexural strength, several tests and the low dispersion
support this value. Above that temperature, thermal degradation processes
became more extensive (fig. 4.27) and the flexural strength experimented a
further decrease. Although the aim was to maintain equal conditions for each
82
test, small variances during the temperature dwell time can lead to an irregular
formation of oxides that may cause a larger dispersion of results. This degradation
of the values continued to suffer a strong decrease at 1273 K, with a complete loss
of mechanical properties.
Fig. 4.31 Average flexural strength as a function of temperature for pure

tungsten in air (left) and under vacuum (right).
The results under vacuum (fig. 4.31, right) exhibited a continuous degradation
in the flexural strength values with a fairly linear tendency up until 1473 K. Since
the formation of oxides is not possible due to the inert atmosphere, the dispersion
of results was reduced.
In general, the maximum flexural strength values in both atmospheres were

very low and did not even reach 250 MPa in the best of cases at lower temperatures.
With the exception of the higher temperatures (1273 and 1473 K) in which the
specimens in air suffered from strong thermal degradation, the results in air yielded
higher strengths than those under vacuum. This fact can be attributed to the
blunting of surface defects due to the formation of superficial oxides [Palacios
et al., 2014].
The DBTT was observed above 1273 K under vacuum. Although tests were
not performed above 1273 K in air, evidence of plastic behaviour was not found
below that temperature (fig. 4.30).
83
ii) Fracture toughness

The force-displacement data from the TPB tests (fig. 4.32) showed a linearly
elastic until fracture behaviour over the entire temperature range. Despite the
SELNB notches had nearly the same dimensions, small differences caused the
variations between the slopes.
Fig. 4.32 Flexural force-displacement curves at different temperatures for

SELNB specimens for pure tungsten in air (left) and under vacuum (right).
The average fracture toughness in air (fig. 4.33, left) showed a progressive
embrittlement from 77 to 473 K. From that point, results experimented an increase
p
in the fracture toughness until its maximum value of 4.5 0.3 MPa m at 1073 K.
At higher temperatures, the specimens experimented a sharp reduction of the
fracture toughness until they reached similar values as at low temperatures.
The results under vacuum nevertheless (fig. 4.33, right) looked very different
since they exhibited an increase from 300 to 673 K. Pure tungsten reached its
p
maximum value of 3.6 0.9 MPa m at this temperature, which was lower than
the maximum value in air, but slightly over than the values obtained at the same
temperature. Above 673 K, pure tungsten became progressively more brittle up
until 1273 K, at which point it stabilises, although at very low fracture toughness
values.
84
Fig. 4.33 Average fracture toughness as a function of temperature for pure

tungsten in air (left) and under vacuum (right).
iii) Fracture surfaces

The fracture surfaces at 77 and 300 K look very similar (fig. 4.34). They exhibited
brittle fracture, in accordance with the results of the TPB tests (fig. 4.30), with
flat breakage in a plane perpendicular to the maximum traction stress without
plastic deformation on the macroscopic scale. At the microscopic, surfaces
also showed brittle fracture with grain boundary decohesion. This fracture
micromechanism is supported by the high intergranular porosity reported, higher
than 8% (table 4.4), which promoted such decohesion.
2μm 2μm
Fig. 4.34 Fracture surface of pure tungsten at 77 K during immersion in liquid

nitrogen (left) and 300 K in air (right).
85
The tests performed in air maintained the same fracture micromechanism of

intergranular decohesion at least up until 673 K. Above this temperature thermal
degradation started and an oxide layer masked the fracture surfaces. At 873 K, the
entire surface was already covered by the oxide scale (fig. 4.35, left). This oxide
scale was maintained at fairly the same thickness at least until 1073 K, above
which the material experimented a significant increase in mass gain (fig. 4.27).
Although the colour cannot be distinguished on the fractography at 1273 K
(fig. 4.36, left), the surface was completely covered with a yellowish dense oxide
scale of WO3 with a thickness of approximately 0.5 µm (fig. 4.28).
2μm 2μm
Fig. 4.35 Fracture surfaces of pure tungsten at 873 K in air (left) and under
vacuum (right).
2μm 2μm
Fig. 4.36 Fracture surfaces of pure tungsten at 1273 K in air (left) and under
vacuum (right).
86
In contrast, the micromechanisms of failure can be seen in the fracture surfaces

of tests performed under vacuum. On the macroscopic scale, the fracture surfaces
exhibited brittle behaviour with flat breakage. On the microscopic scale, they
showed faceted equiaxial grains with intergranular decohesion at least until 873 K
(fig. 4.35, right). At that temperature, although the surfaces were still broken by
grain decohesion, the grains became rounded because of the activation of diffusion
processes with increasing temperature (fig. 4.36, right).
At 1473 K under vacuum, TPB tests on smooth samples exhibited ductile

behaviour (fig. 4.30, right). Although evidence of ductile micromechanisms on
fracture surfaces was not found since samples did not break, specimen bending
was observed on the macroscopic scale. At the microscopic, the samples exhibited
secondary cracks along the surface, mainly in the middle, the point of the
maximum applied stress (fig. 4.37, top). Notched samples at this temperature
exhibited brittle behaviour with intergranular crack growth (fig. 4.37, bottom).
Fig. 4.37 Transversal sections of pure tungsten after TPB tests at 1473 K in
smooth (top) and notched specimens (bottom).
87
4.4.2 Fracture behaviour of W1Y
The flexural σ- curves for W1Y in air and under vacuum (fig. 4.38) showed linear
elastic behaviour until fracture over almost the entire temperature range. The σ-
curve at 1273 K in air exhibited some plastic behaviour, but not enough to satisfy
the 0.2% yield strength offset. Under vacuum however, results for specimens tested
at 1473 K exhibited clear ductile behaviour with macroscopic deformation without
sample breakage in the same way as pure tungsten.
Fig. 4.38 Flexural σ- curves at different temperatures for the W1Y alloy in air
(left) and under vacuum (right).
With the exception of the results at 77 K, which exhibited values in the

same range as at 200 K, the average flexural strength results presented the same
performance in both atmospheres over the entire temperature range (fig. 4.39).
Above 300 K the values increased until they reached a maximum at 873 K of around
450 MPa. From that point on to higher temperatures, a loss of flexural strength was
observed due to massive thermal degradation (fig. 4.27) [Palacios et al., 2014]. At
1473 K, the properties were completely degraded even under vacuum; nonetheless,
the samples did not break. Even so, the addition of 1 wt.% Y2 O3 to pure tungsten
evidenced an increase in the average flexural strength over the entire temperature
range due to the thermal stability provided by the yttria nanoparticles. This finding
is in accordance with literature data [Itoh and Ishiwata, 1996; Swamy et al., 1998].
88
Fig. 4.39 Average flexural strength as a function of temperature for the W1Y
alloy in air (left) and under vacuum (right).
According to the definition of the DBTT, no improvement was observed with

the addition of yttria since ductile behaviour was not observed in air even at
1273 K. Under vacuum, ductile behaviour was presented above this temperature.

Brittle behaviour was observed during the TPB tests since the force-displacement
curves exhibited linear elastic behaviour just like in the case of pure tungsten
(fig. 4.40).

SELNB specimens for the W1Y alloy in air (left) and under vacuum (right).
89
In air (fig. 4.41, left), the W1Y alloy improved its fracture toughness at low
temperatures and at 1473 K; the results were not consistent with the protective
action of yttria at high temperatures. It is likely that the higher porosity enable
an easier inward transportation of oxygen, which promote massive degradation. In
any case, the fracture toughness at 1273 K was doubled that of the pure tungsten
p
(4.4 0.2 MPa m).
Fig. 4.41 Average fracture toughness as a function of temperature for the W1Y
alloy in air (left) and under vacuum (right).
Under vacuum, however (fig. 4.41, right), the results were a bit more promising
since the fracture toughness was above or in the same range as the values for pure
tungsten. In general, the values were in the same range in both atmospheres and
a significant improvement was not observed.

The fracture surfaces at 300 K (fig. 4.42) exhibited features typical of brittle
fracture by grain boundary decohesion and flat breakage on the macroscopic scale.
At higher temperatures, the surfaces in air (fig. 4.43, left) exhibited an oxide
layer covering the entire surface. However, the edges of the oxide were split, as
was macroscopically observed for other alloys after TPB tests (for example in
fig. 4.29). Although this effect was induced by stresses during the scale growth,
the roughness of the fracture surface could be still distinguished (unlike on pure
tungsten surfaces). At 1273 K the differences between atmospheres and pure
tungsten are clear (fig. 4.44, left): whereas pure tungsten was completely covered
90
with a dense oxide scale, the surface of the W1Y alloy was perfectly recognisable.
The EDX data confirmed that rounded grains were Y-rich, while polyhedral grains
were mainly composed of tungsten. The linear oxidation kinetics of pure tungsten
observed over 1073 K likely changed in the alloy to a protective oxide scale
with dense oxide needles [Cifuentes et al., 2012a] that were concentrated around
tungsten grains.
5μm 1μm
Fig. 4.42 Fracture surfaces of W1Y at 300 K in air.
2μm 2μm
Fig. 4.43 Fracture surfaces of W1Y at 873 K in air (left) and under vacuum
(right).
At 873 K the specimens tested under vacuum, exhibited a faceted surface with
rough, small areas (fig. 4.45). The EDX microanalysis revealed that these areas
were composed of tungsten and oxygen, whereas the polyhedral grains were mainly
tungsten. Since these tests were performed under vacuum and the same features
were observed on every fracture surface under these test conditions, the most
probable explanation is contamination during some point of the process. At 1273 K
(fig. 4.44, right), however, the surfaces exhibited a bimodal grain morphology with
round and polyhedral grains, similar to tests in air without the oxide needles.
91
2μm 2μm
Fig. 4.44 Fracture surfaces of W1Y at 1273 K in air (left) and under vacuum
(right).
*A
Spectrum W O
A 88.67 11.33 *B
B 75.97 24.03 *C
C 73.68 26.32
Fig. 4.45 Spectral data (left) of the points in fracture surface at 873 K under
vacuum (right).
4.4.3 Fracture behaviour of W2V0.5Y
The flexural σ- curves exhibited linear elastic behaviour over almost the entire
temperature range. The results in air (fig. 4.46, left) showed a slight ductilisation at
high temperatures. At 1073 K data from the TPB tests experimented some strain,
but it was in the limit to satisfy the 0.2% yield strength offset criterion. Over
that range temperature (1373 and 1473 K), however, the specimens behaved in a
ductile manner. Meanwhile, the tests performed under vacuum (fig. 4.46, right)
did not yield any ductile behaviour up to 1473 K, however, they exhibited 20%
less strain than the tests in air.
92
Fig. 4.46 Flexural σ- curves at different temperatures for the W2V0.5Y alloy
in air (left) and under vacuum (right).
The average flexural strength, as a general rule, exhibited an increase up to

a maximum at a high temperature and then a decrease. In tests performed in
air (fig. 4.47, left), the maximum flexural strength value, 1 GPa, was obtained at
1073 K, the temperature at which the alloy exhibited some plastic behaviour. The
values for the alloy were nearly three times higher than those obtained for pure
tungsten. Even at 1273 K, when pure tungsten had already lost all of its strength,
the W2V0.5Y alloy yielded more than 800 MPa.
The results under vacuum (fig. 4.47, right) did not exhibit such high values,
but still displayed an increase in the flexural strength that double those of pure
tungsten. The highest strength was over 700 MPa at 873 K. The vacuum results
yielded outcomes that were lower than those in air with the exception of data
at 1473 K, temperature at which oxidation in samples tested in air was already
massive. These higher values obtained in air, despite thermal degradation, were
attributed to the blunting of superficial defects, as was previously explained.
The DBTT yielded above 1073 K in air and 1273 K under vacuum. The
temperature under vacuum was the same as for pure tungsten, but in air it was
slightly improved since ductile fracture was not found even at 1273 K for pure
tungsten (even though the σ- data experimented some ductile deformation).
93
Fig. 4.47 Average flexural strength as a function of temperature for the

W2V0.5Y alloy in air (left) and under vacuum (right).

The flexural force-displacement curves for the alloys exhibited linear elastic
behaviour until fracture not even having a slight ductile deformation at any
temperature (fig. 4.48).

SELNB specimens for the W2V0.5Y alloy in air (left) and under vacuum (right).
94
The average fracture toughness results for the SELNB specimens after the TPB
tests (fig. 4.49) were less brittle than pure tungsten. Furthermore, these values
remained fairly constant over the entire temperature range. In general, the values
were two times higher than those for pure tungsten. However, at 873 K in both
atmospheres a considerable increase in the dispersion was observed. This increase
was likely due to some microstructural change caused by the temperature but not
associated with thermal degradation since the increase was also observed in inert
atmosphere tests.
Fig. 4.49 Average fracture toughness as a function of temperature for the


The fracture surfaces at 77 and 300 K looked very similar (fig. 4.50). On the
macroscopic level, the surfaces exhibited flat breakage in a plane perpendicular
to the maximum stress without plastic deformation. On the microscopic scale,
they showed a rough nanostructured area around coarse grains with mainly
intergranular breakage and some transgranular breakage in accordance with the
data from the σ- curves. Some coarse grains were aligned forming elongated shapes
at all temperatures.
The tests performed in air maintained fracture micromechanisms at least up

until 673 K. Above this temperature, thermal degradation started and a layer of
oxide masked the fracture surfaces. At 873 K, the oxide scale was large enough
to cover the entire surface (table 4.6 and fig. 4.51, left). Oxidation became more
evident with increasing temperature and at 1273 K (fig. 4.52, left) the surfaces
were completely covered with oxides. Therefore, the micromechanisms of failure
95
cannot be distinguished but differences can be clearly observed between these

specimens and pure tungsten (fig. 4.36, left). On the pure tungsten surface,
non-protective columnar oxides grew from the surface to the outside [Webb
et al., 1956; Kellett and Rogers, 1963]; the W2V0.5Y alloy exhibited round
particles covering the surface (probably vanadium oxides that became larger with
increasing temperature).
500nm 500nm
Fig. 4.50 Fracture surfaces of W2V0.5Y at 77 K during immersion in liquid

nitrogen (left) and 300 K in air (right).
500nm 5
0 0n
m
Fig. 4.51 Fracture surfaces of W2V0.5Y at 873 K in air (left) and under
vacuum (right).
The fracture surfaces of the tests under vacuum (figs. 4.51 and 4.52, right)
exhibited similar micromechanisms as the tests conducted at 300 K with flat
surfaces on the macroscopic level. At the microscopic, the specimens showed grain
boundary decohesion and some cleavage in the bigger grains, mechanisms that
were maintained up to 1473 K (fig. 4.53). It is also worthwhile to notice the high
number of aligned grains in the fracture surfaces at all temperatures.
96
500nm 500nm
vacuum (right).
500nm
Fig. 4.53 Fracture surface of W2V0.5Y at 1473 K under vacuum.
4.4.4 Fracture behaviour of W4V0.5Y
The flexural σ- curves in air (fig. 4.54, left) exhibited linear elastic behaviour until
fracture over almost the entire temperature range with the exception of 1273 K.
At this temperature, ductile behaviour was observed with a strain of about 1%,
less than in previous cases. Tests under vacuum (fig. 4.54, right) also exhibited
brittle fracture even up to 1273 K; at 1473 K ductile fracture with 1.5% strain
after breakage were exhibited.
97
Fig. 4.54 Flexural σ- curves at different temperatures for the W4V0.5Y alloy
in air (left) and under vacuum (right).
The average flexural strength in air (fig. 4.55, left) showed an increase in
mechanical properties until 1073 K. After that temperature, degradation with
plastic behaviour that was similar to the values obtained for the tests conducted
in liquid nitrogen. In general, the values were more favourable than those of pure
tungsten, particularly at high temperatures: the flexural strength was three times
higher and had a maximum of 700 MPa at 1073 K.
The results under vacuum (fig. 4.55, right) exhibited values in the same range
and an earlier degradation of the strength; the maximum value, at 873 K, was just
below 700 MPa. At 1073 K there was a decrease in the data that was maintained
up until 1273 K. However, these flexural strength results quickly decrease to values
than those obtained at 300 K and exhibited ductile fracture. In any case, the results
were much higher than those obtained for pure tungsten, although at 1473 K there
was a severe loss of strength.
The BDTT was slightly higher for tests performed in air since ductile behaviour
was observed at 1273 K and any changes were observed under vacuum.
98
Fig. 4.55 Average flexural strength as a function of temperature for the


As in previous cases, linear elastic behaviour was observed until fracture over
the entire temperature range. In this alloy, the fracture toughness results in air
(fig. 4.56, left) exhibited a peculiar behaviour compared with previous materials
and alloys. Oddly, the results during immersion in liquid nitrogen were less brittle
than at 300, 473 or even 673 K; higher results were only obtained for a maximum
p
of 9.6 0.5 MPa m at 873 K. Above this temperature, the W4V0.5Y alloy
became very brittle and exhibited the same values as pure tungsten with a high
dispersion. Additional analyses of this values should be completed to understand
this odd phenomenon, although a determining factor is the high porosity reported
(table 4.4).
The results of the tests under vacuum revealed an increase in the fracture
toughness approximately a factor of two over the entire temperature range
(fig. 4.56, right). Since the odd result disappeared, the implication of the high
porosity as being responsible for such behaviour of the material in air is confirmed.
Interconnected pores lead to oxygen penetration inside the material, which leads
to a degradation of mechanical properties. The results in vacuum were slightly
lower than those in air, but the increase of ductility remained constant over the
p
temperature range and exhibited values of approximately 6 MPa m at 1473 K.
99
Fig. 4.56 Average fracture toughness as a function of temperature for the


The fracture surfaces exhibited brittle fracture with flat breakage. At 300 K, many
coarse particles with cleavage and elongated shapes were observed (fig. 4.57), but
in general the surfaces were very similar to that of W2V0.5Y.
500nm
Fig. 4.57 Fracture surface of W4V0.5Y at 300 K in air.
With increasing temperature, in air, oxides covered the surfaces and the
failure mechanisms could not be cleared. At 873 K (fig. 4.58, left), small round
oxides formed, but they were too small to be distinguished. However, at 1273 K
(fig. 4.59, left) the oxide grains grew and masked the entire surface.
100
Under vacuum, the surfaces maintained the micromechanisms of brittle

fracture. At 873 K (fig. 4.58, right) vanadium pools with brittle breakage were
observed. No changes were recorded with increasing temperature (fig. 4.59, left).
Differences between atmospheres are clear; at higher temperatures the
nanostructured grains maintained their brittle fracture but also their size
and morphology.
500nm
vacuum (right).
5
0 0nm
vacuum (right).
101

Teresa Palacios Garcia 1 PDF

Uploaded by

Copyright:

Available Formats

Teresa Palacios Garcia 1 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Teresa Palacios Garcia 1 PDF

Uploaded by

Copyright:

Available Formats

Universidad Politécnica de Madrid

Escuela Técnica Superior de Ingenieros de Caminos,

Departamento de Ciencia de Materiales

Microstructural and Mechanical

Realizado el acto de defensa y lectura de la Tesis el dı́a ..... de ............. de 2015 en la

EL PRESIDENTE LOS VOCALES

The microstructural and mechanical characterisation of different

To continue the search of a tungsten material with lowered DBTT, a

El wolframio (W) y sus aleaciones se consideran los mejores candidatos para la

Durante éste trabajo, se realizó la caracterización microstructural y mecánica

This work is driven by the necessity of finding tungsten materials with

Motivation of the Work V

List of Figures XVI

List of Tables XVII

3 Advanced Experimental Methods: Fracture Toughness

5 Tungsten Foils 143

6 Conclusions and Perspectives 157

1.1 Types of magnetic confinement configurations . . . . . . . . . . . . 3

2.1 Interaction volume of high-energy electrons with solid materials . . 22

2.10 Experimental set-up for TPB tests at 77 K . . . . . . . . . . . . . 34

3.1 Microstructure of the W4Ti1La nanostructured alloy . . . . . . . . 40

4.1 Microstructure of pure tungsten . . . . . . . . . . . . . . . . . . . . 58

4.17 Microstructure of W2Ti1La under low magnification . . . . . . . . 68

5.1 Specimen dimensions of tungsten foils . . . . . . . . . . . . . . . . 144

5.3 EBSD cross-section of the foils annealed 1 h at 1173 K . . . . . . . 146

4.1 Weight % composition of the bulk alloys . . . . . . . . . . . . . . . 56

Even though renewable energy technologies, helped by global subsidies, are

1.1 Nuclear fusion process and devices

In a stellarator reactor (fig. 1.1, left), the poloidal field is produced by

Fig. 1.1 Types of magnetic confinement configurations: segment of the steel

At the moment, however, the Joint European Torus (JET) in Oxfordshire

ITER is a large-scale collaboration project for the construction of a tokamak

Fig. 1.3 ITER cutaway model [New-Scientist, 2009].

The inner components of the machine (fig. 1.4), commonly known as

self-sustainment are essential to design and predict the performance of DEMO

Fig. 1.4 Cutaway of the ITER vacuum vessel [ITER, 2015].

Fig. 1.5 Three-dimensional view of a divertor cassette [Merola et al., 2014].

D) Deuterium and tritium must be continuously refuelled to continue the

1.3 Materials for divertor applications

Additionally, PFCs conditions beyond ITER (e.g. DEMO) will be more

Fig. 1.6 Operation of ITER [New-Scientist, 2009].

1.3.1 Tungsten materials

The use of tungsten, a bbc refractory metal, as a PFM is mainly motivated

1.3.2 Tungsten-based alloys

Fig. 1.7 Some elements used for alloying tungsten.

1.3.2.1 Tungsten-based alloys with rhenium

Rhenium, with a hexagonal-close-packed (hcp) crystal structure, has the second

1.3.2.2 Tungsten-based alloys with tantalum

Tantalum, with a bcc crystal structure, features a combination of properties that

1.3.2.3 Tungsten-based alloys with vanadium

Vanadium-based alloys have been typically used in the aerospace industry,

OTW: an improved creep resistance at high temperatures and a reasonable fracture

Fig. 1.10 Binary phase diagram of W-V [Okamoto, 2010].

1.3.2.4 Tungsten-based alloys with titanium

The mechanical properties of W-Ti alloys have been scarcely investigated.

1.3.3 Dispersion-strengthened tungsten alloys

Another method to increase the high temperature performance of tungsten by

1.3.3.1 ODS tungsten with yttria

The content of the second-phase should be carefully optimised since the