Nanoscale Structure Damage in Irradiated W-Ta Alloys for Nuclear Fusion Reactors

NANOSCALE STRUCTURE DAMAGE IN IRRADIATED W-TA ALLOYS FOR NUCLEAR FUSION REACTORS

A thesis submitted to the University of Manchester

for the degree of

Doctor of Philosophy

in the

Faculty of Science and Engineering

2018

Iuliia Ipatova

School of Materials TABLE OF CONTENTS

List of figures ...... 4 List of tables ...... 12 Abstract ...... 14 Declaration ...... 15 Copyright Statement ...... 16 Acknowledgements ...... 17 1 Introduction ...... 18 1.1 Nuclear fusion reactors ...... 18 1.1.1 Plasma facing materials (PFM) ...... 26 1.2 Challenges of current material selection for fusion ...... 32 1.3 Material degradation during reactor operation ...... 35 1.4 References ...... 40 2 Material properties and degradation mechanisms of W, Ta and their alloys ...... 44 2.1 Fundamental properties ...... 44 2.1.1 Crystallography ...... 44 2.1.2 Phase diagram ...... 47 2.1.3 Diffusion ...... 48 2.1.4 Physical properties ...... 54 2.2 Microstructure ...... 57 2.2.1 Recovery, recrystallization and grain growth ...... 57 2.2.2 Dislocation structures ...... 60 2.3 Mechanical properties ...... 66 2.4 Radiation damage ...... 69 2.4.1 Radiation-matter interaction ...... 69 2.4.2 Quantification of cascade damage ...... 72 2.4.3 Correlating neutron and ion irradiation damage ...... 74 2.4.4 Radiation-induced hardening and embrittlement ...... 78 2.4.5 Radiation-induced void swelling ...... 82 2.4.6 Nuclear transmutation reactions ...... 88 2.5 Main research tasks, limitations found in literature and objectives of the project ...... 91 2.5.1 Current status of the work in the literature ...... 91 2.5.2 Limitations ...... 92

2.5.3 Thesis objectives ...... 93 2.6 References ...... 94 3 Experimental methods ...... 102 3.1 Sample preparation ...... 102 3.2 Ion irradiation experiments ...... 106 3.3 Characterisation methods ...... 110 3.3.1 Optical microscopy ...... 110 3.3.2 Electron-matter interaction ...... 111 3.3.3 Scanning electron microscopy ...... 113 3.3.4 Transmission electron microscopy ...... 117 3.3.4.1 Diffraction contrast mechanisms ...... 119 3.3.4.2 Contrast of dislocation loops and voids ...... 123 3.3.4.3 Burgers vector analysis ...... 124 3.3.4.4 Convergent beam electron diffraction technique (CBED) ...... 126 3.3.5 Hardness measurements ...... 129 3.4 References ...... 132 4 Structural defect accumulation in tungsten and tungsten-5wt.% tantalum under incremental proton damage ...... 134 5 Thermal evolution of the proton-irradiated structure in tungsten-5wt.% tantalum ...... 152 6 Lattice damage formation in irradiated tantalum at variable temperatures ...... 163 7 Radiation-induced void formation and ordering in Ta–W alloys ...... 181 8 Overall discussion ...... 200 9 Conclusions ...... 206 10 Future work ...... 208

List of Figures

1.1 Binding energy per nucleon as a function of the mass number ...... 19 1.2 Design of an inertial fusion confinement (IFC) reactor ...... 21 1.3 Diagram of a beamline at NIF ...... 22 1.4 Simplified diagram for one LMJ beam line showing the main equipment involved during the alignment phase prior to the shot ...... 22 1.5 Schematic of fast ignition fusion ...... 23 1.6 Two concepts of toroidal MCF ...... 23 1.7 The main ITER superconducting magnet system view ...... 24 1.8 Power path to the commercial fusion reactor ...... 25 1.9 Schematic representation of ITER with key parts highlighted ...... 25 1.10 View of the vacuum vessel with selected positions of blanket and divertor ...... 26 1.11 Cross-sectional view of the fusion nuclear components in tokamak wall ...... 27 1.12 Overview of the operating temperatures and radiation damage levels expected after 60 years of operation for different fission and fusion nuclear reactors ...... 27 1.13 Plasma-facing components corresponding to the ITER cross-section; and view of the divertor assembly ...... 28 1.14 Operating temperature windows of candidate fusion materials based on radiation damage and thermal creep considerations ...... 30 1.15 SEM images of cracks formed on the tungsten surface after different numbers of plasma pulses at 0.45 MJ m−2 ...... 32 1.16 Schematic illustration of risk development vs measure of the attractiveness of the structural materials considered for the test blanket modules ...... 34 1.17 Model of selected materials for divertor ...... 34 1.18 Cross-section of the vacuum vessel of fusion reactor showing blanket and divertor ...... 35 1.19 A schematic view of the collision cascade when an “incoming neutron” interacts with the target material ...... 36 1.20 Schematic illustration of time and length scales of the 14.1 MeV neutron induced damage processes in plasma facing materials and components, that can cause microstructural changes and consequently property degradation ...... 37

1.21 Examples of representative microstructures in fusion-related materials as a function of irradiation temperature ...... 38 1.22 Summary of five radiation damage processed that can affect the performance of structural materials in a nuclear fusion reactor ...... 38 2.1 Unit cell model of the different crystallographic phases of Ta and W ...... 44 2.2 Variation of the lattice parameter with composition in Ta–W alloys ...... 45 2.3 Predicted compositional change with time due to 14.1 MeV neutron irradiation of tungsten with a wall heat loading of 2 MW m-2 ...... 46 2.4 TEM images of W after neutron irradiation to 0.98 dpa at 800°C ...... 46 2.5 Models of unit cell: B2 type and D03 type ...... 47 2.6 The calculated equilibrium phase diagram and melting temperatures of Ta–W alloys .... 47 2.7 Experimentally determined W–Re phase diagram and Helmholtz energy difference ΔF at T=1225°C ...... 48 2.8 Radiotracer diffusion demonstrating four basic steps ...... 50 2.9 Arrhenius plots for self-diffusion in tungsten ...... 51 2.10 Potential positions of interstitial H atoms in the bcc unit cell of W ...... 52 2.11 Measured surface self-diffusion coefficient of tantalum with comparison data for W and Mo ...... 53 2.12 The curve of the dependence of tungsten diffusion coefficient on its concentration in the W–Ta system ...... 54 2.13 Self-sputtering yield of candidate high-Z materials as a function of ion energy ...... 55 2.14 SEM images of various bursting features of blisters in the recrystallized tungsten exposed to a fluence of 1026 D m−2 at 250°C and then heated to over 800°C ...... 56 2.15 Cross-sectional SEM micrographs of W targets exposed to pure He plasma for 1 h ...... 56 2.16 Microstructure of tantalum after different thermo-mechanical processes ...... 58 2.17 Recrystallization temperatures versus heating rate for cold-worked tantalum ...... 59 2.18 Effect of annealing temperature on the W microstructure ...... 59 2.19 Ex-situ TEM micrograph of a 0.1 mm-thick tungsten foil ...... 60 2.20 Possible point defects in a crystal lattice ...... 60

2.21 Schematic representation of the slip systems {110}<111>, {112}<111>, {123}<111> in bcc metals ...... 61 2.22 TEM images of annealed Ta samples showing regions of low dislocation density, long straight dislocation lines, and dislocation tangles ...... 62 2.23 Schematic representation of applying a tensile stress along the long axis of a cylindrical single crystal sample with cross-sectional area A ...... 63 2.24 Dislocation structures formed in Ta–W alloys tested at room temperature at a strain rate of 0.1 s-1 ...... 63 2.25 Schematic representation of potential interstitial configurations in a bcc lattice ...... 64 2.26 Schematic representation of interstitial and vacancy-type dislocation loops ...... 65 2.27 Relative formation energies of selected self-interstitial atom configurations for bcc transition metals ...... 65 2.28 The dependence of the ductile-brittle transition temperature on strain rate ε in recrystallized tungsten with a mean grain diameter of 50 μm ...... 67

2.29 Bulk modulus (B), shear modulus (G), and Young‟s modulus (E) of bcc W–Ta alloys as function of Ta concentration ...... 68 2.30 Stress-strain curves for tantalum and Ta–W alloys tested at RT ...... 68 2.31 Schematics of neutron-matter interaction ...... 69 2.32 Schematic representation of the radiation cascade damage ...... 71 2.33 Model by Kinchin-Pease showing the number of displacements ν as a function of PKA energy (T) ...... 73 2.34 Damage profiles for 1 MeV neutrons, 3.2 MeV protons, and 5 MeV Ni11 ions in stainless steel ...... 75 2.35 The temperature shift from 473K (200°C) required at constant dose in order to maintain the same point defect absorption at sinks as a function of dose rate, normalized to initial dose rate ...... 77 2.36 The temperature shift from 473K (200°C) required at a constant dose rate in order to maintain the same point defect absorption at sinks as a function of dose, normalized to the initial dose ...... 78 2.37 Loop number density in W and W–5Ta as a function of irradiation temperature for selected radiation damage levels ...... 80

2.38 Selected micrographs illustrating the radiation damage microstructure in W and W–5Ta alloy as a function of irradiation temperature and dose ...... 81 2.39 Stress-strain curve for Ta–1W and Ta–10W tested at room temperature after neutron+proton exposure ...... 82 2.40 Voids observed in neutron-irradiated tungsten ...... 83 2.41 Incipient void lattice formation observed in neutron-irradiated tungsten up to a fluence of 12×1022 neutrons/cm2 and at a temperature of 750°C ...... 84

2.42 Ordered array of voids in tungsten irradiated at ~550°C to a fast neutron fluence of  1×1022 n/cm2 ...... 84 2.43 Void ordering observed in tantalum after neutron irradiation up to 2.5×1022 neutrons/cm2 and at 585°C ...... 86 2.44 Dissolution of a void in the „interstitial‟ position due to the absorption of breathers coming from larger distances as compared to „regular‟ voids ...... 87

2.45 Crowdion-supply cylinders (length Lc) of a void which run parallel to the close-packed directions of the host lattice ...... 87 2.46 The crowdion-supply overlap of the crowdion-supply cylinders ...... 88 2.47 Pathways of rhenium and osmium generation from tungsten under neutron irradiation . 89 2.48 Tantalum transmutation characteristics as a function of irradiation time ...... 89 3.1 Electric current as a function of potential during electropolishing ...... 104 3.2 A sketch of the procedure used for the preparation of irradiated samples for TEM analysis ...... 105 3.3 SRIM calculated displacement damage profile as a function of depth for 3 MeV proton beam irradiation of pure Ta at current of 9.5 mA during 36 hours with accumulated charge of 1.39 Coulombs ...... 106 3.4 Platform for the radiation experiments in the Dalton Cumbrian Facility ...... 107 3.5 Slit system used to optimise the beam position on the sample and to determine the area of the sample to be irradiated uniformly by the ion beam ...... 108 3.6 Schematics of the slit system ...... 108 3.7 JEOL JEM-2000FX TEM placed at the end of the ion beam line of the MIAMI-1 ion irradiation facility at the University of Huddersfield ...... 109

3.8 Sketch of ion beam production and transport systems at the MIAMI-1 ion irradiation facility ...... 110 3.9 Optical microscope Olympus GX71 ...... 110 3.10 The generation of various detectable signals, arising from the interaction of an electron with the atoms in a solid ...... 111 3.11 Signals emitted from different parts of the electron-sample interaction volume ...... 112 3.12 Diagram of the main components of a typical SEM ...... 114 3.13 Position of the EDX and EBSD detectors in the SEM ...... 115 3.14 Example of X-ray emission in sodium atomic nucleus; and electron shells and main electronic transitions ...... 116 3.15 Schematic setup of an EBSD detector and sample orientation in the SEM ...... 117 3.16 Schematic representation of the main components of TEM and TEM FEI Tecnai G2 20 with its main components labelled ...... 118 3.17 Schematic representation of the Ewald sphere ...... 120 3.18 Formation of a bright field, displaced aperture dark field and centred aperture dark field image ...... 121 3.19 Selected area diffraction pattern (SADP) formation ...... 122 3.20 HR STEM image along the <111> zone axis in Tantalum; and the FFT of that image that shows the cubic symmetry along the <111> zone axis ...... 122 3.21 TEM analysis of the as-received Ta–10W material showing the different contrast of dislocations along different zone axis ...... 123 3.22 Bright field images of voids in tantalum proton-irradiated at 350oC and at 1.55 dpa, taken in TEM using the out-of-focus imaging technique ...... 124 3.23 Bright-field TEM imaging of interstitial-type dislocation loops in tantalum proton irradiated At 180(2)°C and 590(5)°C for different two-beam conditions ...... 125 3.24 TEM data of dislocation loops showing outside contrast in a proton-irradiated Ta–10W sample ...... 126 3.25 The formation of selected-area electron diffraction patterns and convergent-beam electron diffraction (CBED) patterns ...... 127

3.26 Intensity profile (Ig) as a function of s (deviation parameter) and 2ϴB (the separation of the (000) and (hkl) discs); and CBED image of 000 and 200 reflections with the parallel Kossel-Mollenstedt fringes in tantalum ...... 128

2 2 3.27 Graphical representation of (si/nk) vs (1/nk) to obtain the foil thickness ...... 129 3.28 Nanoindenter system from NanoIndenter XP System, MTS; and schematic of the indentation and scratch instrument ...... 130 3.29 Scheme of the indenter – surface contact: under force F , where α is the angle between

indenter surface and sample surface, Ac is the projected area of contact, radius a, which

occurs at a distance hc from the indenter tip, hmax – the maximum indentation depth and

hs is the bowing of the surface (top); After force removal where h0 is the residue indentation depth [17] ...... 131 4.1 EBSD map of the W and W–5Ta microstructures prior to proton irradiation ...... 138 4.2 Simulated damage profile using the SRIM software with the quick Kinchin – Pease approach and the total current deposited on the sample during the ex-situ or in-situ (inset) proton irradiation experiment. The asterisks indicate the regions of the sample from where TEM foils were extracted for ex-situ analysis ...... 141 4.3 In-situ observation of dislocation loop formation and evolution in W–5Ta at 350°C, under increasing damage levels induced by a 40keV proton beam, close to the [111] zone axis ...... 142 4.4 In-situ observation of dislocation loop formation and evolution in W at 350°C, under increasing damage levels induced by a 40keV proton beam, close to the [111] zone axis ...... 142 4.5 Average dislocation loop size and number density in W and W–5Ta as a function of damage level, as derived from the in-situ irradiation data ...... 143 4.6 Secondary electron images of ex-situ W and W-5Ta samples irradiated at 350(4)°C….151 5.1 Dislocation loop evolution in W–5Ta alloy during annealing, after proton irradiation at a temperature of 350℃ and a damage level of 0.7dpa ...... 155 5.2 Variation in the number density of dislocation loops in W–5Ta alloy during post- irradiation annealing up to 1300℃. The coloured region of the graph represents the temperature range where voids are present in the microstructure during annealing. Inset TEM images show evidence of dislocation loop absorption with increasing temperatures ...... 157

5.3 Occurrence and evolution of voids in proton irradiated W–5Ta alloy during annealing...158 5.4 Number density and average diameter of voids formed in irradiated W–5Ta alloy as a function of annealing temperature. The inset in (b) shows the dependence of the average number of vacancies per void on the annealing temperature (see text) ...... 159 6.1 Simulated damage profile using the SRIM software, together with the nano-hardness values measured along the cross-section of the tantalum sample irradiated at 180(2)°C. The asterisks indicate the regions of the sample from where TEM foils were extracted for analysis ...... 167 6.2 Evolution of dislocation structure in tantalum with damage level and temperature. TEM data taken along the [111], [113], [0 ̅2] zone axis ...... 169 6.3 Average dislocation loop size and number density in tantalum as a function of temperature at 0.15dpa and 0.55dpa ...... 170 6.4 Determination of the nature of the dislocation loops by the inside-outside technique in proton-irradiated tantalum at 590(5)°C. The loop show inside contrast in the g= ̅21 reflection and outside contrast in the g=12 ̅ which belong to the [0 ̅2] zone axis. The Burgers vector of the loops was determined to be a<001> (see text) ...... 171 6.5 Bright-field TEM imaging of interstitial-type dislocation loops in tantalum proton irradiated at: left – 180(2) °C; right – 590(5)°C, shown for different two-beam conditions. These TEM data was used for the dislocation analysis using the g.b=0 invisibility criterion, where g denotes the scattering vector and b the Burgers vector, and revealed the a/2<111> (top – 180(2)°C) and a<001> (bottom – 590(5)°C) Burgers vector ...... 173 6.6 TEM BF image showing the presence of radiation-induced voids in tantalum at 345(3)°C, and their absence at 180(2)°C and 590(5)°C ...... 174 7.1 (a) Ion accelerator beamline used for the proton irradiation experiments of Ta–W alloys; (b) main components of the sample stage; (c) top view of the stage with four samples mounted simultaneously and clamped with a tantalum shim; (d) emissivity-corrected temperature distribution over the four samples under proton irradiation, recorded on a thermal camera mounted at a 30° angle with respect to the sample surface; (c) histogram of temperature distribution corresponding to 24 hours of continuous irradiation ...... 186 7.2 (a) Backscattered electron image of the tantalum cross-section, after having been irradiated at 345±3°C using 3MeV proton beam; (b) simulated damage profile using the

SRIM software with the quick Kinchin – Pease approach and the total current deposited on the sample during the irradiation experiment, together with the nano-hardness values along the cross-section of the irradiated tantalum sample ...... 188 7.3 Formation and evolution of void (top row) and dislocation (middle and bottom row) structures in Ta irradiated at 345±3ºC using a 3 MeV proton beam. All data were obtained using the [111] zone axis ...... 189 7.4 Variation with damage level of (a) void size, (b) number density and (c) average void lattice distance in Ta, proton irradiated at 345±3 °C. The inset in (b) corresponds to the number of vacancies per void as a function of damage level (see text). The stripped region denotes the region where the voids are not ordered into a bcc lattice ...... 190 7.5 Evolution of dislocation structures in Ta–W alloys at 345±3°C with damage level ...... 191 7.6 Dislocation loop size and number density as a function of damage level in Ta–W alloys irradiated at 345±3°C ...... 191 7.7 Radiation-induced void formation in Ta–10W alloy, proton irradiated at 345±3°C to a damage level of 1.55 dpa ...... 192

7.8 Radiation-induced change in hardness (ΔHirr.) observed in Ta–W alloys at 345±3°C with increasing damage level ...... 193

List of Tables 1 1.1 Expected operational conditions in Plasma Facing Components in ITER during normal reactor operation ...... 29 1.2 Decay heat and surface dose rate for different potential plasma facing materials ...... 30 1.3 Candidate high-Z plasma-facing surface materials ...... 31

2.1 Diffusion activation energy (Qs) of bcc metals ...... 50 2.2 Thermo-physical properties of tungsten and tantalum as high-Z candidates for plasma facing components in nuclear fusion reactors ...... 55 2.3 Physical properties of materials and the relation to nanostructure growth, where ∆T is the temperature window in which fuzz formation can occur. The shear modulus given corresponds to room temperature, without any effect of He content ...... 57 2.4 Various kinds of radiation damage phenomena and the expected material degradation ... 71

2.5 Barrier strength factor () as a function of defect type and size used in the dispersed barrier hardening model ...... 79 2.6 Hardness data for W-irradiated W and W–5Ta materials ...... 80 2.7 Proton and neutron-irradiation-induced voids in W ...... 83 2.8 Comparison of the reported irradiation conditions for void lattice formation in tungsten 85 2.9 Comparison of the reported irradiation conditions for void lattice formation in tantalum 85 2.10 Summary of the power-plant transmutation products for W and Ta ...... 90 2.11 Neutronic response of first-wall/blanket structural material candidates ...... 90 3 3.1 Sample preparation procedure adopted for tantalum- and tungsten-based materials ...... 103 3.2 Electrolytes and temperature used for electropolishing Ta– and W– based alloys ...... 104 4 4.1 Main parameters used during the proton irradiation of W and W–Ta samples. The damage level of the in-situ samples corresponds to the average value over the disc thickness of 100nm studied by TEM (see inset of Figure 4.1). In this case, the samples were studied in-situ by TEM at selected incremental damage levels up to 0.7 dpa. In contrast, the damage levels mentioned for the ex-situ irradiation corresponds to the value from where the foils were extracted for TEM analysis, using the samples irradiated to a fluence of 1.2×1019 protons/cm2 ...... 139

6 6.1 Summary of irradiation conditions used for tantalum at selected irradiation temperatures and radiation damage levels ...... 167 6.2 Principal characteristics of the radiation-induced dislocation loops in tantalum ...... 173 7 7.1 Summary of irradiation conditions used for the Ta–W samples in this study ...... 186

ABSTRACT

In this project, we have assessed the structural tolerance of advanced refractory alloys to simulated nuclear fusion reactor environments, by using intense proton beams to mimic fusion neutron damage and analysing the proton damaged structures using in-situ/ex-situ transmission electron microscopy and nano-hardness measurements. Refractory metals such as tungsten or tantalum, and their binary alloy combinations, are considered as promising structural materials to withstand the unprecedented high heat loads and fast neutron/helium fluxes expected in future magnetically-confined fusion reactors. Tungsten is currently the frontrunner for the production of plasma-facing components for fusion reactors. The attractiveness of tungsten as structural material lies in its high resistance to plasma-induced sputtering, erosion and radiation-induced void swelling, together with its thermal conductivity and high-temperature strength. Unfortunately, the brittle nature of tungsten hampers the manufacture of reactor components and can also lead to catastrophic failure during reactor operations.

We have focused on two potential routes to enhance the ductility of tungsten-containing materials, namely alloying tungsten with controlled amounts of tantalum, and using alternatively tantalum-based alloys containing specific tungsten additions, either as a full- thickness structural facing material or as a coating of first wall reactor components. The aim was to investigate the formation and evolution of radiation-induced damaged structures in these material solutions and the impact of those structures on the hardness of the material. The main results of this work are: (1) the addition of 5wt%Ta to W leads to saturation in the number density and average dimensions of the radiation-induced a/2<111> dislocation loops formed at 350C, whereas in W the loop length increases progressively and evolves into dislocation strings, and later into hydrogen bubbles and surface blisters, (2) the recovery behaviour of proton irradiated W–5wt.%Ta alloy is characterized by dislocation loop growth at 600-900C, whereas voids form at 1000C by either vacancy absorption or loop collapse, (3) the presence of radiation-induced a<100> loops at 590C in Ta hinders the formation and ordering of voids observed with increasing damage levels at 345C, (4) the addition of 5-10wt.%W to Ta delays the evolution of a/2<111> dislocation loops with increasing damage levels, and therefore the appearance of random voids. These results expand the composition palette available for the safe selection of refractory alloys for plasma facing components with enhanced, or at least predictable, tolerance to the heat-radiation flux combinations expected in future nuclear fusion plants.

14 DECLARATION

I, Iuliia Ipatova, declare that no portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or another institute of learning.

COPYRIGHT STATEMENT I. The author of this thesis (including any appendices and/or schedules to this thesis) owns the certain copyright or related rights in it (the “Copyright”) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. II. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. III. The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. IV. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property University IP Policy (see http://documents.manchester.ac.uk/display.aspx?DocID=24420), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy on Presentation of Theses.

ACKNOWLEDGEMENTS For his supervision, I would like to acknowledge my supervisor Dr. Enrique Jimenez-Melero for having given me an opportunity to learn new things and to grow up as a scientist and a person by giving me independence in the beginning and an essential feedback during writing our joint publications. I acknowledge the rest of my team and every person that during my Ph.D. was so kind to share a part of their time. I want to express my sincere gratitude to my mum, Vera Ipatova, for her never-ending support and belief in me, continuous encouragement, advice, and words of love and the rest of my family and my friends who were always present to endorse me. Also, I want to send my greetings and a massive thanks to Prof. Boris Burakov, who was always there to help me with an advice, consultation or with his kind word. He constantly was in touch to inspire me to chase my dreams, to stay focused and persistent, and do not fear a failure. I would like to acknowledge the Dalton Cumbrian Facility (DCF), where I was based most of my Ph.D. time, and particularly Prof. Simon Pimblott for having funded this project. I also thank people from the Dalton Cumbrian Facility for having a good time and new experience during all these years. A huge thankfulness is expressed to all the staff of the University of Manchester, especially from the Material performance center, who helped me through my Ph.D. and let me work with the most advanced and impressive equipment in the material world. I would also like to heartily thank the University of Huddersfield, and particularly Prof Stephen Donnelly and Dr. Robert Harrison for our collaborative work and productive outcome. Without their contributions and feedback to my work, this project would have been impossible.

Manchester, March 2018

“The release of atomic energy has not created a new problem. It has merely made more urgent the necessity of solving an existing one.” Albert Einstein (1879 – 1955) 1 Introduction 1.1 Nuclear fusion energy Nowadays, one of the biggest challenges of humanity is how to produce the energy required for the ever-growing population on Earth. It has been predicted that the energy demand will be doubled by 2050 and global energy consumption is set to triple by the end of the century [1]. A fusion reactor mimics the energy supply of the sun in a controlled environment. But the most important aspects of fusion energy are that this large-scale energy provided during fusion is safe, does not produce CO2 or other greenhouse gases, does not lead to long-lived radioactive waste generation, and hence, has minimal negative impact on our environment [2]. The implementation of a controlled thermonuclear reaction is one of the most promising ways to solve the increasing energy demand of mankind. Since the early 1950s, the attention of the international scientific community has been drawn to this powerful energy source and have sought to recreate it on earth [3]. The first idea proposed for a fusion machine was based on the thermal insulation of high-temperature plasma using a high voltage electric field. Since the first tokamak, TMP (USSR), was built in 1954, in subsequent years at least 200 more tokamaks have been constructed and tested in the USSR, USA, Europe and Japan [4]. With the accumulated experience in the fusion field, it has become clear that a bigger machine is needed for fusion to take place. At present, the main efforts of the international thermonuclear community are concentrated on the further development and improvement of the tokamak type facilities. The world‘s largest fusion experimental facility ever is currently being built in France is ITER (International Thermonuclear Experimental Reactor) which will demonstrate the technical and scientific feasibility of fusion power as a source of energy. From 50 MW of input heating power, ITER is expected to generate 500 MW of thermal power for at least 300 seconds. Through the ITER project, six partners (China, Korea, India, Russia, Japan, the USA and the European Union) have agreed to apply their scientific and financial sources to prove the viability of fusion as a sustainable energy source. In order to understand fusion, the nature of nuclear reactions should be taken into account. Energy output is obtained in two types of nuclear transformations: (1) the fission or split of a heavy nucleus into lighter fragments and (2) the fusion or merger of two or more light nuclei into a heavier nucleus. In both cases, nuclei with a lower binding energy per nucleon are

18 transformed into energetically more stable nuclei, as seen from Figure 1.1, and the mass difference between reactants and products is released as the kinetic energy of the reaction products according to Einstein‘s energy-mass equivalence formula: E = Δmc2 (1) where E is the change in energy, Δm is the total change in the rest mass of the nucleons involved in the reaction, and c denotes the velocity of light. The plot in Figure 1.1 shows the binding energy per nucleon with respect to the mass number of the elements. Fission occurs at relatively large values of the mass number (A>60), where the fusion of nuclides would require an external energy to take place. 56Fe is at the curve peak and has the highest binding energy per nucleon. For nuclei with an atomic number lower than that of 56Fe energy is released during fusion reactions, for heavier nuclei energy is released during fission reactions.

Figure 1.1 Binding energy per nucleon as a function of the mass number [5].

In order to overcome the Coulomb repulsion between the positively charged nuclei, energy to the very high temperatures of the fuel mixture is needed, about 100 million degrees, which, in this way, is in a plasma state. Radio and microwaves hit the plasma simultaneously with highly energetic deuterium and tritium beams until the moment when plasma achieves a temperature close to 150×106 °C and fusion occurs. The reaction products are alpha particles which are removed via the divertor and stay in the plasma to keep it at high temperatures and neutrons which escape the magnetic confinement and hit the surrounding metal wall. Once they have slowed down, the energy of 14.1 MeV is released. There are several possible fusion

19 reactions between light atomic nuclei with atomic mass A<12÷16. For the future fusion reactor technology, the most promising and perspective nuclear reaction involves two isotopes of hydrogen: deuterium ( ) and tritium ( ):

( ) ( ) (2) During this process, due to the energy and momentum conservation, neutrons with an energy content of 14.1 MeV are released. The energy generated from the reaction above can be used to heat water which will rotate a turbine as it is in a common nuclear (fission) power plant. This heat is finally can be converted into electricity. Fusion is being considered as a perspective ideal source of sustainable energy due to several advantages [2]: 1. In fact, fusion power is safe – the reaction in a fusion reactor is under control and the limited storage of radioactive isotopes means that even in the event of a worst-case scenario, evacuation of surrounding areas would not be necessary as it does not cause long-term contamination. 2. Produced low-level wastes will have a minimal impact on the environment. In the fusion process itself, no radioactive wastes are produced. However, the metallic components of the fusion plant close to the plasma are expected to become radioactive with time, due to the bombardment of neutrons produced by the fusion reactions. By choosing low- activation structural materials, the level of radioactivity should decrease quickly.

3. The fuel will consist of hydrogen isotopes: deuterium ( ) and tritium ( ). Its supply is enough for millions of years of energy production as deuterium is present in sea water and tritium can be produced in-situ by irradiating lithium-6 with fusion neutrons. To breed tritium, lithium-6 (Li) is being used through the following nuclear reaction:

(3) Lithium can be extracted from rocks or soil. The current lithium reserves will be enough to sustain fusion power for 250-600 years, additionally, lithium supply by its extraction from the sea water could be potentially unlimited [6]. Therefore, there is potentially a sufficient amount of fusion fuel to produce enough fusion energy. There are two major concepts to initiate fusion reactions. Firstly, inertial confinement fusion (ICF), where the fusion reaction is achieved by laser assisted heating and compression of a small deuterium/tritium droplet, and, secondly, magnetic confinement fusion (MCF). ICF resembles the formation of a dense miniature Sun. The deuterium/tritium fuel in the pellet must be heated to tens of millions of degrees and simultaneously compressed to ten times the

20 density of lead and remain in this state long enough for most of it to burn. These extreme conditions can be achieved by hitting the pellet, which is placed in the centre of a high- vacuum chamber, with an intense pulse of laser light with a power of several hundreds of TW, which uniformly heats the outer layer of the pellet (see Figure 1.2). This rapid surface heating causes the material to contract. The outer layer rockets away, forcing the remaining pellet wall inward, compressing and heating the fuel. At this stage, the hot dense fuel is being confined in this state for less than a billionth of a second. In the compressed core, deuterium and tritium nuclei encounter each other at high velocity and react.

Figure 1.2 Design of an inertial fusion confinement (IFC) reactor [7].

The largest currently working experimental ICF site is National Ignition Facility (NIF) located at the Lawrence Livermore National Laboratory in Livermore, USA. NIF is the world‘s highest energy laser system, containing 192 laser beamlines. Figure 1.3 demonstrates the 300-meter path of a NIF laser beam, one of 192 similar beamlines. As each laser beam comes out of the preamplifier in the centre, it is routed first to the left, where it passes back and forth through the main amplifier four times before heading back to the target chamber. Le Laser Megajoule (LMJ), laser-based ICF research facility, constructed near Bordeaux, France, is an analogue of NIF. The working principles are the same, but there are 176 laser beams in LMJ. A laser is being used to generate a pulse of infrared light that continues a few billionths of a second with billionths of a joule of energy. A schematic of one LMJ beamline is shown Figure 1.4.

Figure 1.3 Diagram of a beamline at NIF [8].

Figure 1.4 Simplified diagram for one LMJ beam line showing the main equipment involved during the alignment phase prior to the shot [9].

Both, NIF and its European prototype LMJ are based on ―central hot spot scenario‖ which lies on compression and ignition of a spherical fuel capsule happening simultaneously. Another considerable approach called ―fast ignition‖ is to separate the compression phase from the ignition. Fast ignition uses a high-intensity, ultra short-pulse laser to generate the ―spark‖ that initiates ignition. A deuterium-tritium pellet is firstly compressed by lasers to a very high density, and afterwards the short-pulse laser beam supplies energy to ignite the compressed core — as a spark plug in an internal combustion engine. There are several projects currently underway to study the fast ignition approach, including the OMEGA laser at the University of Rochester and the GEKKO XII device in Japan. Additionally, the High Power laser Energy Research facility (HiPER) is the first proposed experiment designed for

22 possible construction in Europe to study the "fast ignition" approach of inertial fusion, which uses much smaller lasers than conventional designs. There are three stages of fast ignition fusion, shown in Figure 1.5: a bunch of lasers to compress the fuel, one super intense laser to ignite the fuel and final step when fusion occurs.

Figure 1.5 Schematic of fast ignition fusion [10].

The magnetic confinement (MCF) attempts to create the fusion energy conditions by using the hot plasma, which is conductive, confined in a vacuum chamber by strong magnetic fields. The pressure is usually on the order of 1 bar with the confinement time up to a few seconds. Whereas in contrast, ICF has higher pressure but ICF confinement process is pulsed as it was mentioned before, and one pulse takes less than a billionth of a second. Deuterium and tritium fill the vacuum of a toroidal chamber in a gas state, being subsequently heated by electromagnetic waves. Once the ignition conditions are reached, the plasma is formed in the chamber. There are two kinds of magnetic confinement devices: tokamaks and stellarators (see Figure 1.6).

Figure 1.6 Two concepts of toroidal MCF [11].

In tokamaks, a doughnut shape of plasma is produced by external superconducting magnets. In the vertical direction, the plasma is confined due to the magnetic field induced by the plasma current, which in turn is induced by the changing current in a central solenoid. In a

23 stellarator the confinement in both toroidal and poloidal directions is ensured by the configuration of external magnets, allowing continuous operation of the device. ITER is based on the tokamak concept of magnetic confinement. First, the fusion fuel of deuterium and tritium is injected into a vacuum vessel, which has a doughnut shape. A powerful magnet is turned on inducing a current across the gaseous mixture. The iso-view of the ITER magnets is shown in Figure 1.7. The voltage rips electrons from the atoms, and the plasma is formed. The superconducting toroidal together with poloidal coils and central solenoid are ramped up keeping the plasma inside the vacuum vessel. These coils generate a magnetic field which shapes the plasma into a torus, where charged particles will start circulating without hitting the reactor wall in an ideal scenario.

Figure 1.7 The main ITER superconducting magnet system view [12].

The commercial fusion path involves confining D-T plasma inside a tokamak, as demonstrated in Figure 1.8. Currently the largest and most powerful tokamak in operation is JET (Joint European Torus), located at the Culham Centre for Fusion Energy, UK [13]. The aim of JET has been to prepare the technical performance scenario close to the one planned for ITER. Thirty times more powerful than JET, ITER will also have a plasma volume ~9 times larger [14]. The construction of ITER has begun in the south of France, a pilot reactor, where experiments will be conducted in order to solve key issues such as to maintain a fusion plasma up to 8 minutes, to test and to verify the tritium breeding technology, to produce a steady-state plasma with a Q value (the fusion energy gain factor) close to 10, to reach

24 deuterium-tritium plasma in which the reaction is sustained through internal heating, to prove the safety aspects of the fusion reactor technology. The results from ITER will aid in the construction of DEMO, a demonstration commercial reactor, projected by the year 2050 [1]. The idea for ITER is to have output ten times more energy than to input, which means to reach Q≥10, and to become the first fusion machine in history which can obtain a net energy output, for pulses between 300 and 5000 seconds in length [15]. A schematic of ITER is given in Figure 1.9. Complementarily, ITER is to prove tritium breeding to be implementable via the lithium blanket technology, to produce a D-T plasma that is sustained through the internal heating, and to demonstrate the safe working mode of high-scale fusion plant.

Figure 1.8 Power path to the commercial fusion reactor [2].

Figure 1.9 Schematic representation of ITER with key parts highlighted [16].

1.1.1 Plasma-facing materials (PFM) For the operational conditions of the future fusion reactor, advanced materials with specific properties are needed. For successful and safe reactor performance it is crucial to select the right structural compositions. There are specific combinations of properties that are required for consideration of structural materials depending on the part of the reactor they will be used for. Among the main requirements attributed to the structural materials are the adequate high- temperature mechanical properties, resistance to irradiation, compatibility with other materials and cooling environments and low activation under irradiation [17].

Figure 1.10 View of the vacuum vessel with selected positions of blanket and divertor [18].

As can be seen from Figure 1.10, the blanket and divertor are the main plasma-facing components. It should be noted here, that the first wall is an integral part of the blanket. The idea of the separation of these two components was discarded in the 1980s and scientists came to the unification of the first wall - blanket structure for plasma device applications. Figure 1.11 shows in a more visible manner the location of all parts from the edge of the plasma to the toroidal field coils, including the first wall and tritium breeding zone as parts of the blanket.

Figure 1.11 Cross-sectional view of the fusion nuclear components in tokamak wall [18].

Figure 1.12 Overview of the operating temperatures and radiation damage levels expected after 60 years of operation for different fission and fusion nuclear reactors. Fusion denotes DEMO fusion reactor. The abbreviations for the fission reactors are: GEN II, Generation II; GFR, gas-cooled fast reactor; LFR, lead-cooled fast reactor; LWR, light-water-cooled fission reactor; MSR, molten salt–cooled reactor; SCWR, supercritical water reactor; SFR, sodium- cooled fast reactor; VHTR, very high temperature reactor [19].

Figure 1.1.2 shows a comparison of the expected operating temperatures and displacement damage levels for structural materials in the six Generation IV concepts and two fusion energy systems (ITER, DEMO). Fusion reactors are intended to operate under more extreme conditions than fission reactors. In general, the fusion materials are called high heat flux materials. Additionally to the heat loading requirements that limit the material‘s choice for a fusion reactor, another demand is to use materials with reduced activation properties. For example, blanket materials are required to shield the vacuum vessel and the superconducting magnets from high-energy neutron fluxes and heat loads of 0.25 – 0.5 MWm-2, as well as to successfully perform the breeding of tritium [20-22]. The plasma facing materials for the divertor will also be required to withstand the high fluxes of 14.1MeV neutrons with higher heat loads [23]. The most difficult task to implement the fusion reaction is to sustain hot fuel – plasma inside the reactor without contacting its walls. The divertor will undergo extreme operating conditions combining the radiation damage induced by the 14.1 MeV neutrons and high heat fluxes of 10 MWm-2 during steady state and up to 20 MWm-2 during slow transients in ITER [16].

Figure 1.13 Plasma facing components corresponding to the ITER cross-section (right); view of the divertor assembly (left) [24], [25].

The main function of divertor is to extract heat and to protect the surrounding walls from thermal and neutronic loads, as well as to remove the fusion reaction impurities (α-particles), unburned fuel and eroded particles from the reactor wall which deteriorate the plasma quality. The divertor in ITER shown in Figure 1.13 will be made up of 54 cassettes and the choice of

28 surface material for it is very important. In 2011 the ITER Organization proposed to start the operation with a tungsten armoured divertor at the vertical targets instead of carbon fibre composite (CFC) tile which was considered for ITER at the end of 20th century. After two years of intense design and successful technology, the full-tungsten armoured divertor was developed as the baseline since the end of 2013 [26]. Ideally, the plasma facing materials would, therefore, have low activation and high strength at high temperatures, be ductile and therefore offer high fabricability, have a high thermal conductivity and have a high resistance to thermal shock and fatigue as well as to radiation/plasma bombardment. However, the ideal material that can completely meet these criteria does not currently exist, since some of the desirable properties cannot be in principle achieved at the same time, for instance, ductility and strength [20]. That is why it is of paramount importance to develop such advanced engineering alloys that will hopefully meet these criteria. Table 1.1 contains the design parameters and working conditions for ITER reactor operation, whereas Figure 1.14 summarizes the operating temperature window of several candidate materials which may be used to fabricate plasma-facing components.

Table 1.1 Expected operational conditions in Plasma Facing Components in ITER during normal reactor operation [24]

Figure 1.14 Operating temperature windows of candidate fusion materials based on radiation damage and thermal creep considerations. The lightly shaded bands on either side of the dark bands represent the uncertainties in the minimum and maximum temperature limits [27].

Refractory metals are being considered as more resistant materials to sputtering than the others [28]. However, some of those high Z materials can negatively impact the plasma as they may cause strong radiation loss from the plasma centre [29]. For example, molybdenum as a PFM was used in Alcator C-Mod and TEXTOR tokamaks [30, 31]. Molybdenum has a high melting point (2623 °C) and sputtering threshold, as well as proven suitability as a plasma facing material. However, Mo was discarded for ITER because the radioactive products that would be generated after the 14.1 MeV neutron irradiation would be too long lived to meet the 100-year limit at which any waste should be appropriate to be recycled, as has been suggested for ITER. For molybdenum after 2.5 years of irradiation at a flux of 5 MWm-2, the decay time to reach a surface dose rate of 10-2 Sv would be 2×105 years, see Table 1.2 [32].

Table 1.2 Decay heat and surface dose rate for different potential plasma facing materials Decay heat after 1 day Decay time to reach (kW/kg) surface dose rate of (irrad. 4.15 MW/m2, 2.5y.) 10-2 Sv (years) (irrad. 5 MW/m2,2.5y.) W 2×10-1 150 Mo 3×10-2 2×105 Fe 10-2 60 SiC 10-3 10

Currently, high-Z materials are again being reconsidered. Tungsten is proposed as a plasma facing armour of the divertor of ITER and also tungsten is currently considered the most suitable candidate for the first wall in DEMO and future fusion reactors. W has the advantage of low or negligible sputtering at low plasma temperatures, low vapour pressure (10-7 Pa) and the highest melting point of 3422°C [33, 34]. Properties of tungsten and alternative material to tungsten are listed in Table 1.3.

Table 1.3 Candidate high-Z plasma-facing surface materials [35] Element Atomic № Melting Atomic density Thermal Specific heat of point (°C) (1028 atoms/m3) conductivity vaporization (W/mK) (KJ/g) Zr 40 1855 4.25 26 6.4 Nb 41 2477 5.4 67 7.5 Mo 42 2623 6.36 100 6.2

Hf 72 2233 4.39 20 3.2 Ta 73 3107 5.48 61 4.1 W 74 3422 6.28 110 4.5

Other potential material candidates such as beryllium or carbon fibre composites (CFC) are being considered for use as plasma-facing components [33, 35]. An alternative choice for plasma facing materials in a future fusion reactor was liquid metals as lithium or a tin-lithium mixture [36]. Flowing liquids could potentially withstand heat loads of up to 50 MWm-2, and would allow for the extraction of heat and particles simultaneously. But this option requires additional investigations devoted to assessing the physical implementation of liquid metals safely in a real fusion reactor. Tantalum is also reported as a candidate material for an application as a plasma-facing component in future fusion reactors either as bulk material or as coating, due to its relatively high fluence threshold for nanostructure formation on the surface [35, 37]. Due to high ductile to brittle transition temperature (DBTT), which lies between 150oC and 400oC, tungsten surface may start cracking which will lead to rapid degradation of the material [34]. The irradiation of tungsten samples at plasma heat loads of 0.45 MJ m−2 has shown that major cracks start forming after 10 pulses and propagating at 100 pulses, see Figure 1.15. Thus, alloying tungsten with tantalum is one of the potential approaches to improve the ductility of W and hence, to enhance the material performance under irradiation.

Figure 1.15 SEM images of cracks formed on the tungsten surface after different numbers of plasma pulses at 0.45 MJ m−2: (a) after 10 pulses; (b) after 100 pulses [38].

1.2 Challenges of current material selection for fusion Considering the requirements for high material performance and reduced activation, the radiation products occurred following irradiation by the 14.1 MeV neutrons should meet the 100-year limit by which time waste will be suitable for recycling or fit for ‗clearance‘ that has been suggested for ITER [32]. This rule is based on the main aim which is set for a fusion commercial power plant – low-activity level waste production with limited impact on the environment. This requirement was applied as the main structural components becoming active will contain radioactive isotopes with half-lives of the order of 10 years, and, hence, recycling will occur in a time frame of 100 years. Therefore, fusion will be focused on structural materials which composition will be based on elements with low-activation properties such as carbon, silicon, titanium, chromium, iron, vanadium, tungsten [39]. Here we talk about alloyed compositions which need to be optimized, as none of the pure candidates will meet all criteria related to plasma-facing components due to some controversial requirement for a material to be ductile, strong and stable at high temperatures and neutron loads. Talking about the main parts of the fusion reactor, blanket which protects the vacuum vessel and its facing plasma part, so-called fist wall, are going to deal with high heat-radiation fluxes as well as the bottom of the vacuum vessel – divertor. There are four main candidates for these roles: vanadium alloys, reduced activation ferritic-martensitic steels, SiC/SiC composites, and tungsten and tungsten-based alloys. Advanced vanadium and its alloys with particular emphasis on the V-4Cr-4Ti alloy are being considered as prospective candidates for the fusion blanket due to its low thermal expansion together with a low elastic modulus which makes these alloys to stand high heat fluxes

32 keeping low thermal stresses. Low activation of vanadium will allow the material to not become radioactive and after the reactor shut down the blanket and its 440 modules will be considered as low-level waste. A reactor using a vanadium-based alloy for the blanket would utilise liquid Li as a coolant as well as a breeder for tritium. Recently extensively studied ferritic/martensitic steels for the blanket, namely F28H steel, contain 7.65 wt.%Cr, 2 wt.%W, Mn, V, Ta, Si and C at below 1 wt.% in total, and Fe as a base. The thermophysical properties of the steel are proved to be good as well as resistance to swelling and embrittlement at high temperatures at damage levels higher than 100 dpa [40]. These steels are considered also for their qualified fabrication routes, developed technologies for welding and available general industrial experience. Current upper operating temperature of ferritic martensitic steels is ~550°C, but the use of oxide-dispersion strengthened (ODS) steels could allow for a maximum temperature greater than 700°C [41]. Probably least developed candidate for fusion application SiC and SiC composites have also low activation properties due to the low neutron cross-section, but eventually, it results in a higher volume of material that is active due to increased neutron penetration, compared with a material with a larger neutron cross-section. There are still issues in manufacturing SiC or SiC composites as some of the additives that are used in manufacture may lead to an increased material activation. Additionally, neutron irradiation causes swelling of SiC/SiC composites which later results in a significant reduction in thermal conductivity. As discussed, the pallet of candidate structural materials for a blanket in future fusion reactors mainly includes reduced activation ferritic/martensitic steels, V-alloys and SiC/SiC composites. But in the future commercial fusion reactor, the liquid LiPb test blanket module becomes one of the most attractive designs. LiPb eutectic as both tritium breeder and coolant of the blankets plus Pb has higher neutron economy and can enhance the tritium breeding performance, and then no additional neutron multiplier such as Be will be needed. And it will be less complicated to extract tritium from liquid LiPb. Figure 1.16 reflects the potential risks of development against a measure of the material attractiveness within the selection of proposed materials for the blanket modules in a future fusion reactor.

Figure 1.16 Schematic illustration of risk development vs measure of the attractiveness of the structural materials considered for the test blanket modules.

Divertor as the main component in a fusion reactor to remove fusion reaction impurities such as α-particles will be also exposed to major neutron loads. Tungsten as a premier frontrunner will be used for a plasma-facing part of the divertor capable to operate up to 3000°C. The ITER divertor model, which can meet the ITER divertor requirement such as heat flux = 5 MW/m2 at 3,000 cycles has successfully been developed in Japanese atomic energy agency. This mock-up is shown in Figure 1.17.

Figure 1.17 Model of selected materials for divertor

Herein, material for the support structure has been chosen as stainless steel (316SS-LN alloy).

Challenging material selection for materials mainly considered as structural for fusion reactors described above as well as their main proposed characteristics are summarized in Figure 1.18. Even though these materials are used in the nuclear industry since decades, there is still a lack of ability to model and, eventually, predict their response to irradiation.

Figure 1.18 Cross-section of the vacuum vessel of fusion reactor showing blanket and divertor.

1.3 Material degradation during reactor operation Material degradation, in terms of thermal shock, thermal fatigue cracks, recrystallization, melting, is expected through the steady state of the reactor operation, as well as transient heat loads [42]. The type of structural defects created is mainly dependent on the material itself, the type of irradiation particle, the radiation dose, dose rate, temperature, solute additions, transmutant elements such as H and He. There are three main processes that may occur when an energetic particle bombards a target material: (1) nuclear reactions are important for safety reasons, since the nuclear transmutation processes may involve radioactive decays with emission of neutrons, protons and/or αparticles, and may extend over several years propagating nuclear wastes and having a harmful impact on the environment; (2) inelastic collisions potentially generate atomic displacements and hence, induce changes in the mechanical properties of the structural material; (3) electronic excitation/ionization which is of limited significance for metals and the irradiation damage process. Excitation happens

35 when charged particle transfers energy to the atom and pushes electrons to the levels of higher energy. During ionization, an electron can be removed if a charged particle has enough energy. In more details, neutron/charged particle-matter interactions are described in Chapter 2. The primary lattice defects generated in the displacement cascade are point defects, namely vacancies and self-interstitial atoms (SIA). They form so-called Frenkel pairs. Interstitials and vacancies move in the crystal lattice through thermally or radiation activated diffusion and may eventually cluster together, annihilate or become trapped at defect sinks such as pre- existing dislocations, grain boundaries, secondary phase precipitates [43]. When a bombarding particle collides with an atom of the material and transfers energy higher than threshold displacement energy, this atom will be displaced from its equilibrium lattice site. By this, vacancy-interstitial pair known as a Frenkel pair (see Figure 1.19) is created. The first shifted atom is called the primary knocked-on atom (PKA). The PKA formation is very rapid and takes less than 10-15 seconds [44]. Afterwards, the PKA may transfer energy to its local environment displacing secondary atoms (if EPKA > Ed, where Ed denotes to displacement energy) and creating additional point defects that constitute the displacement cascade. Displacement per atom unit is used to characterise the damage level achieved in the target material and refers to how many times on average an atom in the irradiated volume of the material has been shifted from its equilibrium position in the crystal lattice.

Figure 1.19 A schematic view of the collision cascade when an ―incoming neutron‖ interacts with the target material [20].

As shown in Figure 1.20 the initial damage happens in a very short time frame and length scales, and it evolves with time leading to changes in material properties at longer timescales. As a consequence of the evolution of the radiation-induced self-interstitial atoms and vacancies and their clustering through the damage cascade, the formation of dense and dilute zones takes place. There may be a higher dislocation density, plus radiation-induced phase transformations and precipitation reactions may occur [45]. In addition to this, the formation of diluted zones or local vacancy enrichment may lead to void growth and material swelling. The consequences of these and other processes involving radiation-induced SIAs and vacancies which can compromise the material stability in a radiation environment are shown in Figures 1.21 and 1.22. Besides that, Figure 1.21 shows the types of temperature- dependent microstructures produced in irradiated materials

Figure 1.20 Schematic illustration of time and length scales of the 14.1 MeV neutron induced damage processes in plasma facing materials and components, that can cause microstructural changes and consequently property degradation [20].

Figure 1.21 Examples of representative microstructures in fusion-related materials as a function of irradiation temperature [19].

Figure 1.22 Summary of five radiation damage processed that can affect the performance of structural materials in a nuclear fusion reactor [46].

In order to predict the material damage in future fusion reactors, detailed comparative expertise from the scientific community is needed. This expertise will help to develop advanced engineering compositions which will be able to withstand unprecedented amounts of radiation at elevated temperatures. To summarize, the main processes which can modify and in the end determine material microstructure during/post-irradiation are: 1 Generation of single self-interstitial atoms and vacancies (through the damage cascade)

tending to annihilate at ~0.1 Tmelt., where Tmelt. denotes the melting point of the material (for high-temperature materials such as tantalum and tungsten). 2 Formation of interstitial or vacancy-type dislocation loops and voids, as a result of SIAs or vacancy agglomeration. 3 Nucleation of voids or gas bubbles due to vacancy clustering and accumulation of He atoms. 4 Capture of lattice defects at different traps in the microstructure, which can be dislocations, impurity or alloying atoms, grain boundary or the free surface of the material. 5 The effect of high fluencies of H and He and its clustering as expected in future devices may cause a number of (near-) surface modifications, related to the formation of bubbles, blisters, voids or cavities.

1.4 References

1. EFDA, Fusion Electricity A roadmap to the realisation of fusion energy. 2012. 2. C. Llewellyn Smith, D.W., Fusion. Energy Policy, 2008. 36: p. 4331-4334. 3. Tamm, I.E., Theory of the magnetic thermonuclear reactor, part I Plasma Physics and the Problem of Controlled Thermonuclear Reactions (transl) 1959: p. 3-19. 4. Balshaw, N. All-the-World‘s Tokamaks. 2006 08.08.2017]. 5. A. Lépine-Szily, P.D., Nuclear astrophysics: Nucleosynthesis in the Universe. International Journal of Astrobiology, 2012. 11(4): p. 243-250. 6. D. Fasel, M.Q.T., Availability of lithium in the context of future D–T fusion reactors Fusion Engineering and Design, 2005. 75-79: p. 1163-1168. 7. https://lasers.llnl.gov. National Ignition Facility & Photon Science. [cited 2017 02.08]. 8. https://pubs.acs.org. [cited 2017 02.08]. 9. M Luttmann, V.D., P Gendeau, C Lanternier, P Baudon, Y Schiano, L Espert, F Seguineau, O Henry, and E.C. D Raffestin, L Delage, C Lecot, N Chapron, Overview of LMJ alignment to target chamber centre and very first results Journal of Physics: Conference Series, 2016. 717: p. 012106. 10. http://www.osaka-u.ac.jp/. [cited 2017 05.08]. 11. Y.Xu, A general comparison between tokamak and stellarator plasmas. Matter and Radiation at Extremes, 2016. 1: p. 192-200. 12. C. Sborchia , A.B.O., T. Boutboul , K. Chan , A. Devred , S. Egorov , K. Kim , N. Koizumi , S. Lelekhov, P. Libeyre, The ITER magnet systems: progress on construction. Nuclear Fusion, 2014. 54: p. 013006. 13. Tokamak‖, C.-F.E.-T. http://www.ccfe.ac.uk/tokamak.aspx. [cited 2017 10.08.2017]. 14. Eurofusion. https://www.euro-fusion.org/2014/01/comparison-of-jet-and-iter/. 2014 21.08.2017]. 15. J.B.Lister, B.P.D.X.L., CRPP-EPFL, Association EURATOM-Confédération Suisse. The ITER project and its data handling requirements. in ICALEPCS. 2003.

16. "Divertor", I.O. 2016 10.08.2017]. 17. Zinkle, S.J., Advanced materials for fusion technology. Fusion Engineering and Design, 2005. 74(31-40). 18. Hassan, M., Introduction to fusion technology issues. 2014, Alexandria University.

19. S.J. Zinkle, L.L.S., Designing Radiation Resistance in Materials for Fusion Energy The Annual Review of Materials Research, 2014. 44: p. 241. 20. J. Knaster, A.M., T. and Muroga, Materials research for fusion. Nature Physics, 2016. 5(2): p. 77-80. 21. L. Giancarli, V.C., M. Abdou, M. Akiba, B.G. Hong, R. Lasser, C. Pan, Y. Strebkov, Test blanket modules in ITER: An overview on proposed designs and required DEMO-relevant materials. Journal of Nuclear Materials, 2007. 367-370: p. 1271– 1280.

22. J. Knaster, A.M., T. Muroga, Materials research for fusion. Nature Physics, 2016. 5(2): p. 77-80. 23. M.R. Gilbert, S.L.D., D. Nguyen-Manh, S. Zheng, L.W. Packer, J.-Ch. Sublet, Neutron-induced dpa, transmutations, gas production, and heliumembrittlement of fusion materials. Journal of Nuclear Materials, 2013. 442: p. S755-S760. 24. V. Barabash, G.F., R. Matera, A.R. Raffray, ITER Home Teams, Armour materials for the ITER Plasma Facing Components. Physica Scripta, 1999. T81: p. 74-83. 25. M. Merola, F.E., R. Raffray, P. Chappuis, T. Hirai, A. Martin, Overview and status of ITER internal components. Fusion Engineering and Design, 2014. 26. T. Hirai, F.E., V. Barabash, A. Durocher, A. Fedosov, L. Ferrand, T. Jokinen, V. Komarov, M.Merola, S. Carpentier-Chouchana, N. Arkhipov et al., Status of technology R&D for the ITER tungsten divertor monoblock. Journal of Nuclear Materials, 2015. 463: p. 1248-1251. 27. S.J. Zinkle, N.M.G., Operating temperature windows for fusion reactor structural materials Fusion Engineering and Design, 2000. 51-52: p. 55-71. 28. W. Eckstein, J.L., Sputtering of tungsten and molybdenum. Journal of Nuclear Materials, 1991. 183: p. 19-24. 29. N. Noda, V.P., R. Neu, A review of recent experiments on W and high Z materials as

plasma-facing components in magnetic fusion devices. Journal of Nuclear Materials, 1997. 241-243: p. 227-243. 30. B. Lipschultz, D.A.P., B. LaBombard, J.E. Rice, D. Smith, S.J. Wukitch, A study of molybdenum influxes and transport in Alcator C-Mod. Nuclear fusion, 2001. 41(5): p. 585.

31. V. Philipps, T.T., Y. Uedaa, A. Pospieszczyk, M.Z. Tokar‘b, B. Unterberg, L. Konen, B. Schweer, U. Samm, P. Wienhold, Molybdenum test limiter experiments in TEXTOR. Nuclear Fusion, 1994. 34(11): p. 1 417–1429. 32. H. Bolt, V.B., G. Federici, J. Linke, A. Loarte, and K.S. J. Roth, Plasma facing and high heat flux materials – needs for ITER and beyond. Journal of Nuclear Materials, 2002. 307-311: p. 43-52.

33. H. Bolt, V.B., W. Krauss, J. Linke, R. Neu, S. Suzuki, N. Yoshida, Materials for the plasma-facing components of fusion reactors. Journal of Nuclear Materials, 2004. 329-333: p. 66-73. 34. Philipps, V., Tungsten as material for plasma-facing components in fusion devices. Journal of Nuclear Materials, 2011. 415: p. S2-S9. 35. J.N. Brooks, L.E.-G., A. Hassanein, T. Sizyuk, Plasma-facing material alternatives to tungsten Nuclear fusion, 2015. 55: p. 043002. 36. N. Baluc, K.A., J. L., V.M.C. Boutard, E. Diegele, S. Jitsukawa, A. Kimura, R. L. Klueh, A., and R.J.K. Kohyama, et al., Status of R&D activities on materials for fusion power reactors Nuclear fusion, 2007. 47(10): p. S696–S717. 37. T.J. Novakowski, J.K.T., A. Hassanein, Effect of high-flux, low-energy helium ion irradiation on Ta as a plasma-facing material. Scientific reports, 2016. 6: p. 39746. 38. V.A. Makhlaj, I.E.G., N.N. Aksenov, B. Bazylev, O.V. Byrka, I. Landman, J. Linke, S.V. Malykhin, A.T. Pugachov, M.J. Sadowski, E. Skladnik-Sadowska, M. Wirtz, Plasma exposure of different tungsten grades with plasma accelerators under ITER- relevant conditions. Physica Scripta, 2014. T161: p. 014040. 39. M. Victoria, N.B., P. Spätig., Structural materials for fusion reactors. Nuclear Fusion, 2001. 41(8): p. 1047-1053. 40. Gelles, D.S., On quantification of helium embrittlement in ferritic/martensitic steels. Journal of NuclearMaterials 2000. 283-287: p. 838–840. 41. T. Muroga, M.G., S. J. Zinkle, Overview of materials research for fusion reactors. Fusion engineering and design 2002. 61-62: p. 13-25. 42. C. Thomser, V., et al., Plasma facing materials for the JET ITER-like wall. Fusion science and technology, 2012. 62. 43. Was, G., Fundamentals of Radiation Materials Science, ed. Springer-Verlag. 2007.

44. Robinson, M.T., Basic physics of radiation damage production. Journal of Nuclear Materials, 1994. 216: p. 1-28.

45. Hoffelner, W., Materials for Nuclear Plants. 2013: Springer-Verlag. 46. S.J. Zinkle, J.T.B., Structural materials for fission and fusion energy. Materals today, 2009. 12(11): p. 12-19.

“Crystals are the same as people, imperfections make them more interesting.” Colin Humphreys (1941-present) 2 Material properties and degradation mechanisms of W, Ta and their alloys This chapter is dedicated to the description of basic properties of W, Ta and selected binary alloys attractive for nuclear fusion reactor applications. The chapter starts with an introduction to fundamental properties of these materials and is followed by their microstructural characteristics and relevant mechanical properties. At the end of the chapter, the relevant radiation damage phenomena reported in the literature for these materials are discussed. Furthermore, the consequences of the radiation-induced lattice damage for their properties under the ion/neutron fluxes expected in reactor service conditions are pointed out. 2.1 Fundamental properties 2.1.1 Crystallography Both tungsten and tantalum crystallize in a body-centered cubic structure (α-phase) at ambient pressure and in a wide temperature range up to their melting point. The β-phase has been found mostly in films [1, 2]. The structure of the tungsten β-phase is called A15 cubic. The β-phase of tantalum is hard and brittle and its crystal symmetry is tetragonal.

The γ-phase of tungsten monocarbides (γ-WC1−x) has a cubic structure of the B1 (NaCl) type [3]. Figure 2.1 shows the unit cells model of each crystal structure described.

Figure 2.1 Unit cell model of the different crystallographic phases of Ta and W: (a) Body- centred cubic, α-phase of Ta and W; (b) Tetragonal crystal structure of tantalum β-phase;

Bulk tantalum presents the stable α-phase which is chemically resistant to most acids and has high electrical conductivity and ductility. Tantalum has not been observed to undergo a transition from ductile to brittle behavior with decreasing temperature as low as -250°C

[5]. The lattice parameter of α-phase has a value of 3.3020 Å at room temperature [6] and its space group is Im-3m. The dependence of the lattice parameter on tungsten concentration in Ta-W alloy is given in Figure 2.2. Tantalum β-phase can be irreversibly transformed to the α-Ta phase when heated up to 775°C, and usually exists in the form of thin films. It can co-exist with the α-phase together at non-equilibrium synthesis conditions and/or stabilization by impurities such as nitrogen and oxygen. The beta phase of tantalum presents different mechanical properties as compared to the alpha-phase, it is hard and brittle, its tetragonal space group is P42/mnm, and the lattice parameters take the value of a = 10.194 Ǻ, c = 5.313 Ǻ at room temperature [7].

Figure 2.2 Variation of the lattice parameter with the composition of Ta-W alloys [8].

Tungsten exists in two major crystalline forms: α and β. The α-phase has a body-centered structure and is the more stable form. A15 cubic β phase is metastable and exists at ambient conditions owing to non-equilibrium synthesis. It is known that the long-term neutron irradiation of tungsten may lead to the transmutation of tungsten into rhenium and later into osmium forming brittle phases. The duplex α+σ field is expected to be entered in about 2.3 years of service and the pure σ-phase in round 5.7 years. Hence, the five-year service irradiation is expected to bring the W–Os–Re composition closer to the 100% σ- phase field as shown in Figure 2.3. The final composition of initially non-alloyed W is expected to be: 75.1W12.8Os11.9Re (at.%) [9].

The neutron irradiation of W-Re alloys showed the formation of WRe (σ-phase) and WRe3 (χ-phase) phase precipitates with the atom-probe field-ion microscope analysis of the W-10%Re [10] and W-25%Re alloys [11] respectively. A mechanism was suggested for the formation of the WRe3 precipitates in W-25%Re whereby self-interstitial atoms at the

45 edge of displacement cascades react with Re atoms in order to produce a mobile mixed dumbbell (see section 2.2.2). Two of the dumbbells react together, resulting in a di-Re cluster that is immobile, which will then itself reacts with a further dumbbell, producing a

Re3 cluster which will react with a W self-interstitial atom, resulting in a WRe3 formation.

Figure 2.3 Predicted compositional change with time due to 14.1 MeV neutron irradiation of tungsten with a wall heat loading of 2 MW m-2 [9].

Bright and dark-field images of these precipitates formed in W after neutron irradiation to 0.98 dpa at 800°C are shown in Figure 2.4.

Figure 2.4 TEM images of W after neutron irradiation to 0.98 dpa at 800°C: (a) bright- field image, (b) and (c) dark-field images of σ (WRe) and χ (WRe3) precipitates [12].

When alloying tantalum and tungsten forming binary Ta-W alloys, there are two crystal structures that may take place: ordered B2 phase of CsCl type when W atom will take the position of the central atom, and D03 - Fe3Al type (Ta3W and TaW3). The unit cells of these structures are presented in Figure 2.5. Considering the high melting point of the alloy, the B2 and D03 ordered phases of Ta-W alloys are stable at temperature, not exceeding ~720°C.

Figure 2.5 Models of unite cell: B2 type (left); D03 type (right).

2.1.2 Phase diagram The binary phase diagram of Ta-W system is shown in Figure 2.6. The two components are soluble in each between 3020°C and 3422°C, which are the melting points of Ta and W respectively. Solid phase exists in this system up to 3000°C. There is a strong tendency forming a CsCl-type B2 superstructure when the W content exceeds 10at.% up to 90at.% in Ta-W alloys below 800°C.

Figure 2.6 The calculated equilibrium phase diagram and melting temperatures of Ta–W alloys: cal-2 (cal-1) represent the phase boundaries between B2 and A2 phases, [8].

Figure 2.7 Experimentally determined W-Re phase diagram and Helmholtz energy difference ΔF at T=1225°C [13].

According to the equilibrium phase diagram (Figure 2.7), W-10%Re is subsaturated with respect to the solvus line of the primary solid solution (bcc). Concentration of 30at.% of Re in W is subsaturated with respect to the bcc+σ duplex – phase. The addition of Re above 25at.% leads to the formation of the brittle σ or χ phases. However, these phases have been observed at much lower Re concentrations following neutron irradiation [14]. 2.1.3 Diffusion Atoms in the lattice sites continuously oscillate around an equilibrium position. If the temperature rises, the average velocity of the atoms will increase. Hence, receiving a high amount of energy through thermal activation, some of the atoms may be able to migrate into the space between the occupied lattice sites by jumping from their atomic positions to adjacent locations. These atoms may change the positions with the neighboring atoms or most likely with nearby vacancies. This is the so-called substitutional atom diffusion. The migrating atom requires the thermal energy to be high enough to break all the bonds with the other neighboring atoms and locally distort the lattice as demanded for the jump. The result of all these processes is the continuous movement of atoms in the crystal lattice, which has a common name – diffusion. In case of a displacement of atoms in non-alloyed metals, the phenomenon is called self-diffusion. If the atoms of one element move into

48 another one, then the term hetero-diffusion is used, or inter-diffusion. The term ―diffusion couple‖ is used as well if two materials inter-diffuse at elevated temperature. The diffusion is described by the two Fick‘s laws, for the steady and non-steady state respectively. The first Fick‘s law says that the flux goes from regions of high concentration to regions of low concentration with a magnitude that is proportional to the concentration gradient.

(1) where J is a flux of particles which diffuse, D – the diffusion coefficient, c – the concentration of diffusing particles as a function of position x. The concentration gradient is assumed to be constant with time in this case. Fick's second law shows the change in the concentration of the diffusing substance as a function of time:

(2) where t – time. In this case, the concentration gradient varies with time. The diffusion coefficient is temperature dependent and can be expressed using an Arrhenius-type equation:

(3) where Q is an activation energy for diffusion process: the highest Q, the lowest diffusion,

R – is the gas constant (8.314J/mol·K) and D0 is diffusion constant or pre-exponential factor, and it contains such parameters as the vibrational frequency of the atom in the lattice and its coordination number. The pre-exponential factor can be calculated using Vineyard‘s formula [15]:

∏ ∏ (4)

Where the and are the set of vibrational frequencies of the system for the equilibrium and saddle point configurations, respectively. N represents normal frequencies of the entire system at the starting point of the transition to the N-1normal frequencies of the system constrained in the saddle point configuration.

Table 2.1 contains the data of activation energies for 4 bcc metals with the ration Qs/Tm which should be nearly constant for different metals.

Table 2.1 Diffusion activation energy (Qs) of bcc metals [16]

Metal Activation energy (Qs), Qs/Tm, (cal/mol/K) kcal/mole Ta 63.8±4 19.5 W 71.4±4 19.4 Mo 52.6 18.2 Nb 54.5 19.8

Self-diffusion in tungsten Self-diffusion in tungsten was measured in the temperature range of ~1400-3100°C in order to elucidate the key parameters for point-defect formation and motion [17]. The diffusion coefficient for tungsten was measured by conventional radiotracer thin-layer sectioning technique. In the beginning, a thin layer containing radioisotope is deposited over the sample surface. Afterward, the sample is annealed at a particular temperature for a certain period of time, and lastly, the sample is sectioned into thin layers either by mechanical or sputter grinding techniques. The diffusion profile, for example, the relative tracer concentration as a function of sample depth is determined from this experiment. A schematic of this method is shown in Figure 2.8.

Figure 2.8 Radiotracer diffusion demonstrating four basic steps.

The γ-emitting 187W was used as the radiotracer to be deposited. After, the thickness of 0.5 nm was spark-machined off the cylindrical sides of samples to be sectioned by grinding. Diffusion parameters received in this work for W are presented in Table 2.1. The Arrhenius plot for self-diffusion in tungsten is curved and is fitted to a sum of two exponentials with activation energies 5.4 and 6.9 eV. The lower activation energy corresponds to diffusion by single vacancies. The higher activation energy can represent motion either by divacancies or by interstitials [17].

Figure 2.9 shows the Arrhenius plot of the self-diffusion coefficient available from the literature on tungsten [17].

Figure 2.9 Arrhenius plots for self-diffusion in tungsten [17].

The presence of hydrogen (H) strongly affects most of the W properties, due to phenomena like vacancy formation and blistering [18]. Moreover, H is known to be trapped in impurities, vacancies, dislocations and grain boundaries affecting the microstructure evolution of the material [19]. Therefore, the blistering problem induced by H retention is now one of the most challenging issues for the plasma facing material design. In order to understand the underlying mechanism, the diffusion behavior of H in W requires extensive study. Recently the activation energy for diffusion of hydrogen atoms in tungsten was reported [20]. Hydrogen atoms in the tungsten lattice occupy tetrahedral sites, as shown in Figure 2.10 (a) (positions 0-4 and 6-8) as well as an octahedral site, position 5. The hydrogen atom can jump between two tetrahedral sites in two ways: either directly between two nearest tetrahedral sites (0-1 or 0-3, for example) or through an octahedral site (0-5-6, for example). The short distances between 0-1, 0-2, 0-3, etc. are ≈ a/√8 ≈ 1.13 Å and the long distances between 0-6 are ≈ a/2 ≈ 1.6 Å. A hydrogen atom in position 0 can make 4 short jumps into sites 1-4, and 2 long jumps in sites 6 and 6* (symmetrically below position 0). The temperature dependence of the diffusion coefficient calculated for the

51 short and long jumps is shown in Figure 2.10 (b). The diffusion coefficient calculated for long jumps in the range of 1100-1700°C is approximated by the diffusion coefficient: D(T) = 2.3×10-7exp (-0.29/kT) m-2s-1. The activation energy was found to be 0.39 eV.

Figure 2.10 Potential positions of interstitial H atoms in the bcc unit cell of W: (a) positions 1-4 and 6-8 are tetrapores, position 5 is an octapore; (b) Temperature dependence of the diffusion coefficient D for the short (1) and long (2) jumps [20].

Self-diffusion in tantalum The self-diffusion coefficient in tantalum (Figure 2.11) was measured by field electron microscopy between ~1300-2100oC [16]. The study of diffusion was restricted to

T<0.7Tmelt, where Tmelt. denotes melting temperature, when the diffusion occurs through the jump of adatoms to neighboring sites. The activation energy was measured to be 63.8 kcal/mole [16].

Figure 2.11 Measured surface self-diffusion coefficient of tantalum with comparison data for W and Mo [16].

Ta-W interdiffusion Interdiffusion in the substitutional solid solution of Ta-W system has been studied in the temperature range 1300-2100°C [21]. Below this range, the diffusion rate is expected to be extremely slow as this system exists in a solid form at temperatures up to 3000°C. The variation of the activation energy for inter-diffusion Q, with the composition of single- phase systems, lies in three regions: (I) 0-20 at.% Ta, Q decreases from approximately 140 kcal/mole to 130.5 kcal/mole; (2) 20-80 at. % Ta, Q is constant at approximately 130.5 kca1/mole; (3) 80-l00at. % Ta, Q decreases from130.5 kcal/mole to 120 kcal/mole. With a higher concentration of tantalum, the diffusivity is faster, as the activation energy decreases monotonically with increasing Ta content. An increase in tungsten concentration

53 from 60% to 90% leads to a gradual increase of the diffusivity from 0.8·10-11 cm2/s to 1.9·10-11 cm2/s. The interdiffusion coefficient is observed at the minimum in the region of tungsten concentration from 40 to 60wt. %, see Figure 2.12.

. Figure 2.12 The curve of the dependence of tungsten diffusion coefficient on its concentration in the W-Ta system [22].

2.1.4 Physical properties Tungsten is a primary candidate for plasma-facing components due to its desirable physical properties: high melting point (3422°), high thermal conductivity and high resistance to sputtering and erosion [23]. However, due to the main concern of relying on just one material, there are promising alternative materials being considered in order to broaden the choice for fusion reactor technology. There are concerns about He, D, T, and neutron-induced microstructural changes (‗tendrils‘, bubble formation, etc) and resulting surface integrity; fatigue, dust, and other issues. Table 2.2 presents selected physical properties to compare W and Ta. Tantalum has similar melting characteristics as tungsten but the lower thermal conductivity. But the heat of vaporization is slightly less, and erosion of tantalum from melt-layer splashing can be lower than tungsten since a thinner melt layer could be developed. Hence, thermal and mechanical properties may not degrade relatively quickly in extreme fusion environments.

Table 2.2 Thermo-physical properties of tungsten and tantalum as high-Z candidates for plasma facing components in nuclear fusion reactors [24] Material W Ta Atomic number Z 74 73 Melting point (°C) 3422 3107 Atomic density (1028atoms m-3) 6.28 5.48 Thermal conductivity (W/mK) at 900°C 110 61 Specific heat of vaporization (KJ/g) at 4.5 4.1 respective melting point

Sputtering is a process whereby particles are ejected from a solid target material due to the bombardment of the target by energetic particles [25]. Sputter erosion/redeposition affects the lifetime of plasma facing components, core plasma contamination, and tritium breeding zone due to co-deposition. As can be seen from the Figure 2.13, Ta and W have the highest incident energy threshold for self-sputtering coefﬁcient due to atomic density and surface binding energy differences (if enough energy is transferred, binding forces can be overcome). This translates into a higher temperature limit in the region between the plasma core and the vessel wall.

Figure 2.13 Self-sputtering yield of candidate high-Z materials as a function of ion energy [24].

It should be stressed here that due to high He+ fluxes and fluences together with elevated temperatures (up to 1500°C expected on the divertor tiles), tungsten surfaces may develop

55 significant surface changes including surface pores, bubbles, and nano-scale tendril-like structures known as ―fuzz‖ caused by He irradiation which are shown in Figure 2.14 and Figure 2.15 [26-28]. Fuzz is a result of the near-surface He trapping at defect sites due to nanometre-sized He bubbles which form in the W surface layer growing at an elevated temperature [29].

Figure 2.14 SEM images of various bursting features of blisters in the recrystallized tungsten exposed to a fluence of 1026 D m−2 at 250°C and then heated to over 800°C [28].

Figure 2.15 Cross-sectional SEM micrographs of W targets exposed at 850°C (left), and 1050°C (right) to pure He plasma for 1 h [30].

Recently, tantalum was proposed as an alternative plasma-facing material to tungsten due to its higher fluence threshold for He+ ion-induced surface nano-structuring, and therefore a lower risk of contaminating the reactor plasma [31]. Tantalum did not show any fiber- form structure formed over a wide range of temperature, 0.2-0.4Tm, Tm denotes the melting point [32]. Table 2.3 shows the possible relationship of nanostructure growth to the physical properties of W, Re, and Ta.

Table 2.3 Physical properties of materials and the relation to nanostructure growth, where ∆T is the temperature window in which fuzz formation can occur. The shear modulus given corresponds to room temperature, without any effect of He content [26] Meltin Shear Temperature Crystal Nanostructure Z Material g Point Modulus Range Structure Growth (°C) (GPa) (∆T/Tm) Well-developed 74 Tungsten (W) bcc 3422 161 0.27-0.5 fuzz Well-developed 75 Rhenium (Re) hcp 3186 178 0.36 fuzz Loops and/or 73 Tantalum (Ta) bcc 3017 69 loops with short ~0.29 protrusions

Tantalum has an initially lower thermal conductivity than tungsten. This is important to ensure that the material will withstand and efficiently transfer large heat loads. However, a higher fluence threshold for the surface nano-structuring would suggest that Ta surface‘s thermal conductivity is less likely to degrade over the lifetime of reactor operation. 2.2 Microstructure 2.2.1 Recovery, recrystallization and grain growth Tantalum The recrystallization temperature in this work is considered as the temperature at which the 50% of the cold-worked material recrystallizes (usually in one hour). Typically this temperature is in the range of 0.3-0.5 of the melting point. The recrystallization temperature depends on the previous microstructure and its stored energy in the form of dislocations. But once this temperature is reached, the material changes its microstructure. Dislocation rearrangement and annihilation happen during recovery, normally prior the recrystallization. Eventually, recrystallization implies the formation of new grains. Thereafter, hardness will decrease in comparison with the material before recrystallization due to the reduced dislocation density. However, due to the appearance of new grain boundaries hardness may potentially increase as well. It is essential to anneal as-received material to sufficiently high temperature beforehand. Recrystallization restores the ductility of the material and decreases tensile strength.

Influence of cold rolling reduction on Ta recrystallization The evolution of the deformed microstructure of pure tantalum in the temperature range from room temperature to 1100°C after 70% and 90% of thickness reduction is depicted in Figure 2.16.

0%rolled

70%rolled

0%rolled

70%rolled +1h.at1000°C

9 +1h.at1000°C

rolled

9 +1h.at1100°C

70%rolled +1h.at1100°C

Figure 2.16 Microstructure of tantalum after the following thermo-mechanical processes: a) 70% rolled; b) 70% rolled+1h. at 1000oC; c) 70% rolled+1h. at 1100oC; d) 90% rolled; e) 90% rolled+1h. at 1000oC; f) 90% rolled+1h. at 1100oC [33].

As it can be seen from the Figure 2.Figure 2.16 the microstructure of tantalum after 70% and 90% thickness reduction by cold rolling are both recovered below 1000°C and are fully recrystallized after 1 hour of annealing at 1100 °C. But the 70% cold-rolled sample has some residual stress. Deformation structure of 90% cold rolled sample was fully recovered after 1 hour at 1000 °C. Hence, initial deformation structure affects recrystallization behavior. But although in both samples, i.e. 70% and 90% rolled, the original deformed microstructure has been modified during annealing at 1100°C.

Influence of heating rate on Ta recrystallization The recrystallization temperature of cold-rolled tantalum was reported to depend on the heating rate. As the heating rate decreases, the recrystallization temperature increases (see Figure 2.17). The reason why recrystallization temperature may change due to the heating rate is due to the recovery which occurs before recrystallization. At high heating rates, the material is not able to complete the recovery process before recrystallization commences. Some of the stored energy which could be spent in recovery process becomes available to

58 drive recrystallization. But at low heating rate recovery of tantalum is essentially complete before the recrystallization temperatures are reached.

Figure 2.17 Recrystallization temperatures versus heating rate for cold-worked tantalum [34].

Tungsten Recrystallization studies of W with an initial grain size of 1.4 μm have shown that primary recrystallization starts at 1200oC for long annealing times ~43 hours, and for only an hour long annealing time at 1300oC. After annealing at 1300oC for an hour, a grain size of 17 µm was obtained. For a one hour treatment at 2000oC, a grain size of 27 µm was observed. Additionally, secondary recrystallization (or so-called extensive grain growth) was seen at 2000oC, complete grain growth was achieved after 1 hour at 2300°C as shown in Figure 2.18 [35]. As can be seen from this figure, during secondary recrystallization, grain size exceeds the average size of grains in tungsten.

Figure 2.18 Effect of annealing temperature on the W microstructure: 1h. at 2000°C: secondary growth at the top surface (left); 1 h. at 2300°C, complete secondary recrystallization (right), ×50.

An annealing treatment at 1400oC for 20 hours resulted in large grains without dislocations observed as shown in Figure 2.19 [36].

Figure 2.19 Ex-situ TEM micrograph of a 0.1 mm-thick tungsten foil: as-received (left). The arrows indicate a grain boundary and a sub-grain boundary respectively; dislocation network in as-received (middle); after annealing for 20 h at 1400oC (right).

2.2.2 Dislocation structure Atomic arrangements in real materials do not follow ideal crystalline templates. In the crystal lattice, real metals have various lattice defects (imperfections) that distort bonds between atoms and have an influence on properties of metals. There are following well-known structural defects:  Point or zero-dimensional defects – vacancies, interstitial atoms, substitutional or interstitial impurities. These defects have atomic dimensions. Figure 2.20 shows possible point defects which may occur in the crystal lattice.

Figure 2.20 Possible point defects in a crystal lattice.

 Linear defects – dislocations, dislocation lines, clusters and agglomerates of vacancies and interstitials.  Planar defects – grain boundaries, twin boundaries, stacking faults. Metals and alloys after manufacturing process normally have a polycrystalline structure. The grains are normally irregularly shaped and oriented differently in relation to other adjacent grains, whereby the properties of real metals are averaged taken into account the orientation distribution function of the material, and anisotropy phenomena are observed.  Volume defects –three-dimensional atomic defects: cracks, voids, inclusions. Hence, metals have a number of imperfections in their crystal structure. The crystal lattice defects described above significantly affect the plasticity and strength of the material. In order to break interatomic bonds, external mechanical stress should be applied. In ideal crystals, this stress is several orders of magnitude higher than for real crystals due to the lattice defects, especially dislocations. When a certain stress above a critical value is applied to a crystal system, dislocations move, and slip occurs. Slip in bcc metals happens along the plane with the vector which connects the atoms of closest approach. The Burgers vector b denotes the magnitude and direction of the lattice distortion caused by dislocations. There are no close-packed planes in the bcc crystal structure. Hence, heat is required for dislocations to slip. There are six slip planes of type {110}, each with two <111> directions (12 systems), 24 {123} and 12 {112} planes each with one <111> direction (36 systems), so in total there are 48 potential slip systems, shown below in Figure 2.21

Figure 2.21 Schematic representation of the slip systems {110}<111>, {112}<111>, {123}<111> in bcc metals.

Dislocation lines occur on the surface during the dislocation motion through the crystal as shown in Figure 2.22. The annealed microstructure of tantalum sample shown in this figure contains different regions: mostly free of dislocations, with lines and tangles. So some regions are free when the other ones contain a high density of dislocations predictably caused by dislocation slip, as the authors claim that the dislocation content in tantalum decreases slowly with annealing.

Figure 2.22 TEM images of annealed Ta samples showing regions of: low dislocation density (left); long straight dislocation lines (centre); dislocation tangles (right) [37].

Usually, dislocation tangles form if the metal is heavily deformed, which does not allow the dislocations to easily glide. That may lead to work hardening and the strengthening of a metal by plastic deformation. Interaction of dislocation lines and dislocation tangles with other lattice defects during plastic deformation leads to the material strengthening. During the first stages of plastic deformation, dislocations start moving along the slip system. This slip system the lowest critical resolved shear stress, but then other slip systems can become activated. Additionally, dislocations can move from one slip plane to another (cross-slip) until they meet grain boundaries or other defects such as other dislocations. Bcc metals cross slide extensively and it has been suggested that cross-slip of different (110) planes may integrate with slip-on higher order planes [38]. Due to different grain orientations, plastic deformation does not start simultaneously in all the grains. The Schmid factor describes the slip plane and the slip direction of a stressed material with respect to the direction of the applied stress (σ). This stress can resolve the most amount of shear stress and is expressed as: τ=σ·m (5) where m – Schmid factor and can be expressed by: m=cos(φ)cos(λ) (6)

62 where φ – the angle with the glide plane and λ –the angle with the glide direction of the applied stress, both shown in Figure 2.23.

Figure 2.23 Schematic representation of applying a tensile stress along the long axis of a cylindrical single crystal sample with cross-sectional area A.

The addition of alloying elements changes the mechanical properties of the material. Tantalum doped with a controlled amount of tungsten up to 10wt.% increases the yield strength and work hardening rate, Figure 2.30 [37]. The addition of tungsten causes the dislocation structure to change from one in which a cell structure dominated to one in which a tangled structure dominates. The samples showing this phenomenon after being deformed at room temperature up to 30%at a strain rate of 0.1s-1 are shown in Figure 2.24.

a Ta b Ta-2.5W c Ta-2.5W d Ta-10W

Figure 2.24 Dislocation structures formed in Ta-W alloys tested at room temperature at a strain rate of 0.1 s-1: (a) region of Ta showing dislocations cells; (b)Ta-2.5W showing a cellular structure; (c) Ta2.5W showing a region that does not contain cells; (d) dislocation structure observed in Ta-10W, the bands of high dislocation density [37].

Tungsten additions cause a change in the deformed structure from one that consists primarily of dislocation cells to one which consists primarily of dislocation tangles. With increasing tungsten concentration the cells that do form in these materials are smaller in diameter. The increase in work hardening rate can be correlated with an increase in the number dislocation density in the material and/or result from the effect of tungsten. However, the authors also claim that there were commercial materials used which contained ~103 at. ppm interstitial impurities which may have an accompanying or synergistic effect with the tungsten alloying.

Irradiation-induced loop formation and evolution When a metal is being exposed to irradiation with particles a pair of self-interstitial atom shifted from its positions in the lattice and a vacancy forms. In bcc metals, self-interstitial atoms share a single lattice site. There is a symmetrical shift of these two atoms respectively to the lattice site. The center of mass remains at the interstitial site. The most stable configuration in Ta and W is the <111> dumbbell. Other configurations can be found in the Figure 2.25.

Figure 2.25 Schematic representation of potential interstitial configurations in a bcc lattice: (a) octahedral, (b) tetrahedral, (c) crowdion, (d) <100> dumbbell, (e) <110> dumbbell, (f) <111> dumbbell.

Irradiation-induced dislocation structure SIA and vacancies diffuse through the crystal lattice. However, interstitial atoms are more mobile as the probability of an empty adjacent interstitial site is higher than for a vacancy adjacent to a host atom. Continuing their generation during the radiation cascade, interstitials and vacancies can end up being inserted between densely packed atomic planes of the lattice, and clustered together into disc-like geometries (Figure 2.26). This is how

64 interstitial or vacancy type of loops eventually form. In bcc such as tantalum and tungsten, unfaulted dislocation loops with a Burgers vector ½a <111> or a <100> dominate in the microstructure. The faulted ½a0 <110> loops are not favorable in bcc structures as they have high formation energy and become unstable at high temperatures [39].

Figure 2.26 Schematic representation of interstitial (left) and vacancy-type (right) dislocation loops [40].

Density-functional theory calculations showed that mostly in bcc metals (except α-Fe), the axially-symmetric <111> self-interstitial atom configuration has the lowest energy of formation [41] as shown in Figure 2.27. For α-iron, the <110> self-interstitial atom configuration has the lowest energy of formation. When vacancy loops cluster together, they form 3-dimensional voids a few nanometre in diameter, as explained below in section 2.4.3.

Figure 2.27 Relative formation energies of selected self-interstitial atom configurations for bcc transition metals [41].

The traditional mechanism of formation of both types of perfect dislocations is described by the shearing of ½a0 <110> partial dislocations at an early stage of growth, according to the classical dislocation reaction theory [42]:

(7)

̅ (8)

Dislocation loops of a <001> type are less mobile than ½a<111> as their Burgers vector is larger and, hence, it requires a higher threshold migration energy, which means a higher temperature is needed to activate the loop motion. Hence, a/2<111> dislocations, which are glissile, whereas sessile a<100> dislocations require more energy to form, but they are stable. At high temperatures of irradiation, a <001> loops can be considered as relatively immobile defects. The high mobility of 1/2 a<111> loops has a strong effect on the evolution of the microstructure under irradiation. In contrast, a<001> loops require higher formation energy and would not be expected at a low temperature of irradiation.

Interstitial loops in bcc metals have preferentially a Burgers vector b = ½a <111> [17] and therefore enclose a stacking fault. At an early stage in their growth, these loops shear in either the <001> or <110> direction to remove the stacking fault and produce loops with b

= or respectively. The a<001> loops have only been observed after irradiation at elevated temperatures as higher energy is required for <110> shear. So, both

and a <001> loops are initially rectilinear and lie on a (110) plane. The core energy of the dislocation affects the size and orientation of the loops.

2.3 Mechanical properties Tungsten has the brittle-to-ductile transition temperature (BDTT) of tungsten is relatively high and this is one of the worldwide concerns of using this refractory metal as a plasma- facing material, see Figure 2.28 Previously, the BDTT of W was generally reported between 200–400°C (the range is caused by the initial microstructure, the utilised strain rates and other variations in testing methods), and is predicted to increase with neutron irradiation up to 800–1000°C [43, 44], which denotes the high difficulty of using tungsten as a structural material. Thus, tungsten stays brittle at relatively high temperatures, and this can lead to premature failure during low-temperature operating conditions in the fusion reactor.

Figure 2.28 The dependence of the ductile-brittle transition temperature on strain rate ε in recrystallized tungsten with a mean grain diameter of 50 μm [45].

Therefore, it is of paramount importance to improve the ductility of W-based materials and to develop such W- based compositions with other refractory metals of similar properties, so that W-based components can be operated at a minimum temperature regime without a risk of fracture. In order to improve the ductility of W, three main methods are being investigated: the production of alloys, composite materials and nanostructuring [46]. It is necessary to know the mechanical properties of these materials and to describe how their microstructures evolve with increasing deformation. Recently the effect of tantalum addition to tungsten was studied by first-principles calculations. Elastic moduli (bulk modulus B, shear modulus G, Young‘s modulus E) were calculated and reported to decrease with increasing Ta concentration. It was found that ductility of W-Ta alloys increases with more addition of tantalum. Elastic moduli are presented in Figure 2.29. If the ratio of bulk modulus to shear modulus is larger, the better ductility the material exhibits.

Figure 2.29 Bulk modulus (B), shear modulus (G), and Young‘s modulus (E) of bcc W–Ta alloys as a function of Ta concentration [47].

The addition of tungsten to tantalum, i.e. Ta2.5W and Ta10W alloys, increases the ultimate tensile strength of the material at room temperature (Figure 2.30). The addition of 2.5wt.%W has a little effect on the yield strength of the alloy, but the addition of 10% tungsten almost doubles the yield strength.

Figure 2.30 Stress-strain curves for tantalum and Ta-W alloys tested at RT. In the figure lines denoted by 1 correspond to tests run at strain rates of 10-1s-1, denoted by 3 correspond to tests run at 10-3s-1 and denoted by 4 correspond to tests run At 10-4s-1 [48].

Ta2.5W alloy is the most useful in applications where low-temperature strength is important as well as high corrosion resistance and good formability. This alloy is also higher in strength than pure tantalum while maintaining good fabrication characteristics.

Ta10W alloy may be considered in high-temperature applications up to 2400°C and high strength in a corrosive environment is required. This alloy has roughly twice the tensile strength of pure tantalum, yet retains tantalum's corrosion resistance and most of the pure tantalum's ductility [5].

2.4 Radiation damage 2.4.1 Radiation-matter interaction Neutrons are subatomic particles without a net electric charge. The cross-section (probability that a nuclear reaction will happen) of the neutron is weak or ~0 in comparison to the cross-section of the proton with a similar mass. That is why neutrons penetrate deeper into the matter. Neutrons do not interact with the atomic electrons in the target material, only with the nuclei of the atoms and also with the magnetic moments of the atoms. This gives an additional contribution to scattering. The possible types of interaction between neutrons and nuclei in a given material are shown in Figure 2.31.

Figure 2.31 Schematics of neutron-matter interaction.

As can be seen in Figure 2.31, the principal types of neutron interaction with a target material are: scattering (elastic or inelastic), neutron absorption (radioactive capture and nuclear fission) and transfer reactions with neutron emission and charged particle ejection. Scattering can be classified as elastic and inelastic collisions. During irradiation, the incident neutron can shift an atom from its place only if it transfers an energy which is higher than threshold energy for the displacement. During an elastic collision, part of the neutron‘s kinetic energy transferred to the nucleus does not change. During inelastic scattering, the final kinetic energy is lower than before interaction. The energy loss

69 through the inelastic collisions depends on the excitation energy levels of the nucleus. Part of the energy of the incident neutron is being absorbed by the recoiling nucleus what leaves a nucleus in an excited state. Neutrons slow down more intensively at high energies and by heavy nuclei during inelastic scattering. During nuclear reactions, a neutron absorbed or captured by a nucleus can cause emission of several neutrons or charged particles (protons, deuterons or alpha particles). Following the nuclear reaction, the composition of the material may undergo modifications. For example, hydrogen and helium can be generated as a result of neutron absorption followed by transmutation reactions. Neutron capture process may lead to the formation of unstable activation products (radioactive nuclides). Such radioactive nuclei can exhibit half-lives ranging from small fractions of a second to many years. During nuclear fission reactions, the nucleus of an atom splits into lighter nuclei. The fission process produces ‗free neutrons‘ and -photons, and releases energy. In a fission reaction, the neutron is first absorbed by the nucleus, before it splits into two fragments as in typical chain fission reaction:

(9) During irradiation with charged particles, they can transfer energy by passing close to the atoms. Charged particles in general transfer energy by electronic excitations to the atom bringing up the atomic electrons to higher energy levels, and also by ionization when the charged particle has enough energy to remove an electron. Charged particles can also be scattered. The stopping power (dE/dx) indicates the change in energy E of the bombarding particle as a function of the penetration depth x in the material. To join the three processes: inelastic collisions (c), nuclear reactions (n) and electronic excitation/ionization (e), the stopping power can be expressed as follows:

(10)

During inelastic collisions, an ion with sufficient energy collides with a target atom, and the bombarded atom is displaced from its lattice, forming a primary knock-on atom (PKA) and hence, one vacancy is left behind. This PKA receives a significant amount of energy and starts colliding with further atoms, potentially causing the production of additional recoils which are known as secondary knock-on atoms. This process is shown schematically in Figure 2.32. Therefore, for each collision, two-point defects are

70 produced. The first one corresponds to the atoms shifted from their positions in the lattice (termed self-interstitial atoms, SIAs), and secondly, vacancies formed as the recoil atoms left their lattice sites empty. A SIA and a vacancy form a Frenkel pair. These defects restrict the motion of gliding dislocations and therefore increase the hardness of the material.

Figure 2.32 Schematic representation of the radiation cascade damage.

Information about the consequences of radiation damage and reactions causing the material degradation at longer time scales is collected in Table 2.4.

Table 2.4 Various kinds of radiation damage phenomena and the expected material degradation [49]

2.4.2 Quantification of cascade damage As it was mentioned in chapter 1.2, a certain amount of energy is needed to displace an atom from its site within a crystal lattice. This energy is denoted as Ed and called displacement threshold energy. Several models were developed, based on the initial work by Seitz, to approximately evaluate a number of atoms which are displaced by an energetic bombarding particle during the damage cascade. Firstly, in 1949 Seitz proposed that any atom receiving energy larger than Ed actually is permanently shifted to an interstitial position and estimated Ed for several solid materials to be 20 – 25eV. Seitz said that the knocked-on atom is in the range of velocity in which it dissipates its energy either by elastic encounters or by excitation of the valence electrons. However, the constituent particles do not possess sufficient momentum to result in a relative displacement of atoms; this displacement occurs only because the ions are made mobile by thermal fluctuations [50]. Fast particles such as protons or neutrons and fission fragments have sufficient momentum that they may cause displacements directly and hence induce effects at any temperatures. Thereafter, some mathematical explanations for determining the number of vacant lattice sites or interstitial atoms were outlined done by Snyder and Neufeld [51]. Since then, the mostly applied model was the one proposed by Kinchin and Pease which is a linear displacement model based on several assumptions [52]. They suggested that vacancy and the interstitial atom will be formed due to secondary collisions (energy T is transferred to the lattice atom and if T>Ed) between interstitial atoms that were moving and stationary atoms within the lattice of a material following fast neutron irradiation. Wherein, energy transfer in the collision is given by the hard sphere, isotropic scattering model. In their model, two moving atoms are created when a PKA first strikes a stationary atom. PKA with initial energy T after the collision has residual energy T − ε and the struck atom receives an energy ε − Ed. Hence, the number of atom displacements resulting from a primary recoil atom of energy T can be expressed as:

ν(T) = ν(T −ε) +ν(ε −Ed) (11) with displacement probability for T > Ed.

However, the energy transfer ε is unknown. Since the PKA and lattice atoms are identical, ε may lie anywhere between 0 and T. And the number of displacements induced by PKAs with energy, T, was defined by Kinchin and Pease as:

(12)

{

As it can be seen from Figure 2.33, between the displacement threshold energy and

, there is a probability that one atom will be displaced, at energies between and the cut-off energy, , atoms will be displaced and at energies higher than (where energy loss from electronic stopping occurs) the number of displacements remains constant, at .

Figure 2.33 Model by Kinchin-Pease showing the number of displacements ν as a function of PKA energy (T) [53]

Snyder and Neufeld presume that an energy Ed can be consumed in each collision. Comparing it with the Kinchin–Pease model, it may be expected that ν(T) would then reduce due to energy loss term which is added. However, since atoms are allowed to leave the collision with energy less than Ed, an increase in ν(T) will occur. The result was very similar:

ν(T) = 0.56(1+ ) for T > 4Ed (13)

Lindhard [54] proposed that the parameter ν(T) not to be interpreted solely as the number of displacements produced per PKA, but need to be taken as a part of the original PKA energy, which is passed to the atoms of the lattice in slowing down. According to Lindhard:

ν(T) = ξ(T) ( ) (14) where

ξ(T) = (15)

is reduced PKA energy:

(16)

In 1974, Robinson and Torrens developed the MARLOWE binary collision code to enable detailed computational simulations of high-energy displacement cascades [55]. Later on, in 1975 Norgett, Robinson and Torrens (NRT) developed the Kinchin-Pease model further [56]. In order to obtain the NRT dpa, the number of Frenkel pairs should be divided by the number of atoms in particular volume. Nowadays, a standard measure of irradiation damage is dpa, however, care must be taken when using it as it does not consider migration, coalescence, and recombination of defects. NRT model defines the number of Frenkel pairs, induced by a PKA, with initial energy, T as shown below:

̂ (17)

where ̂ is available energy for the generation of atomic displacements via collisions that are elastic and is the threshold displacement energy; 0.8 is so-called displacement efficiency which is independent on the PKA energy, material or its temperature. The NRT dpa provides an environment-independent radiation exposure parameter that in many cases can be successfully used as a radiation damage correlation parameter.

2.4.3 Correlating neutron and ion irradiation damage Performing neutron irradiation in order to understand the future behaviour of potential structure material for nuclear reactors is of primary interest due to its ability to match the dose, temperature and frequently primary recoil spectra. However, it is known to be expensive and special infrastructure is required to irradiate and examine radioactive samples afterward. Additionally, it is not easy to irradiate a large number of samples at a wide variety of irradiation conditions as well to study individual mechanisms, as cascades during neutron irradiation are being formed from mono-energetic ions. Ion irradiations which have been successfully used since 1960s-70s [57] are considered to be a promising

74 substitute for a neutron damage. Low cost, close control of dose, damage rate and temperature with almost negligible residual radioactivity for proton irradiations or its absence at all for heavy ion irradiations, which allow examining samples straight after irradiation, may help to largely contribute to the understanding of irradiation effects on various types of materials. A huge advantage is that damage rates similar to that during real reactor irradiation can be achieved in order of days, for instance, 1 dpa/day for protons and 100 dpa/day for self-ion irradiation. To compare, available reactors worldwide can only go up to 20 dpa/year. One of the potential difficulties with ion irradiation is a shallow penetration depth. The penetration depth of proton beam into the matter with an energy of a few MeV can achieve tens of micrometers in comparison with an increased neutron penetration depth at lower energy. Figure 2.34 demonstrates damage profiles in stainless steel depending on energy and type of irradiation [58].

Figure 2.34 Damage profiles for 1 MeV neutrons, 3.2 MeV protons, and 5 MeV Ni11 ions in stainless steel [58].

There are several parameters that need to be considered when comparing different types of irradiation experiments. The smaller proton mass and the smaller recoil energy than of heavy ions, result in a damage morphology characterizing by smaller, more widely spaced cascades than in the case with ions or neutrons which also results in a lower reachable dose rate. The number of displacements per proton is less than that for a heavy ion, requiring higher fluxes of protons to achieve the same damage level in the same amount of time. Discussion on the effect of irradiation dose rate and the required temperature shift will be given below.

Dose rate effect Dose rate during irradiation is usually expressed in the units of displacement per atom (dpa) per unit time (second) or as a density of elementary defects (number of interstitial atoms/vacancies) per atom per unit of time. So, dose rate affects the density of the defects produced in a unit volume per unit time, with a higher density generated in a unit volume per unit time in a high dose rate ion irradiation experiment, in comparison to a low dose rate neutron irradiation experiment. Within moving particles, the frequency by which interaction events occur is proportional to the square of the density of the particles. Therefore, in high dose rate experiments, there will be an increased number of interactions between defects induced by irradiation [59]. Point defect annihilation can occur via three different mechanisms: loss at surfaces, grain boundaries or other extended sinks; capture of vacancies or interstitials, leading to defect growth or shrinkage, and nucleation of the same type of defects and recombination of interstitials with vacancies. With increased dose rate or a decreased temperature, the recombination mechanism is more likely to occur, whereas, at low dose rates or high temperatures, the sink mechanism is more dominant [60].

Temperature shift When ions are being utilized to mimic neutron effects, there is a difference in the damage occurred. In general, for an increased dose rate an increased temperature is required to reach the same type of damage [58]. Proton irradiation at dose rates which are about 100 times of those for neutrons and at slightly elevated temperatures have proven to accurately reproduce the irradiated microstructure. Whereas, self-ion irradiation of alloys that are in the recombination-dominant regime call for widely different irradiation temperatures for different microstructure features. This analysis assumes that irradiation is in the recombination-dominant or sink-dominant mode. For a given change in dose rate, the shift in temperature can be found at a constant dose to keep Ns invariant, where Ns is the number of defects per unit volume that is lost to sinks. Mansur developed equations to deduce the required increase in temperature [61]. If

Ns, is constant (important in describing microstructure development involving total point defect flux to sinks), the temperature shift for a given dose rate ratio is:

( )

(18)

( )

76 where k is Boltzmann‘s constant, T is temperature, G is the point defect generation rate and

is the vacancy migration energy. For a certain modification in dose at the same dose rate, temperature shift has to be found to maintain Ns invariant for changing dose rate (constant dose). This is shown on Figure 2.35 for a range of three different vacancy migration energies.

Figure 2.35 The temperature shift from 473K (200°C) required at constant dose in order to maintain the same point defect absorption at sinks as a function of dose rate, normalized to initial dose rate.

In earlier work of Mansur [62], a different type of temperature shift was attained requiring, in this case, the net flux of vacancies over interstitials to a particular type of sink (cavities) to be invariant. This net flux is relevant to swelling rate. The temperature shift derived in the same way and for the recombination dominated steady-state case to keep swelling rate invariant is:

( )

(19)

( )

where denotes vacancy formation energy.

Figure 2.36 shows the temperature shift to keep Ns invariant for changing dose (constant dose rate).

Figure 2.36 The temperature shift from 473K (200°C) required at a constant dose rate in order to maintain the same point defect absorption at sinks as a function of dose, normalized to the initial dose.

But this theoretically predicted shift much depends on the type of microstructural defect (formation of dislocation loops, precipitates or voids) that is under analysis, hence, there is not a single temperature shift that will recreate all the same damages expected from a neutron irradiation at a certain temperature, to a higher dose ion irradiation. It should be noted that at increased temperatures there is an increased mobility of defects, and therefore increased defect losses at sinks, which could mean less damage, from one side [60]. However, if there is an increased dose rate for the same dose, the irradiation time is shorter, leading to less time for annealing of defects, and therefore increased damage. The temperature shift tends to balance this effect.

2.4.4 Radiation-induced hardening and embrittlement Hardness is not a fundamental material property, but rather is a measure of complex multi- axial elastic and plastic deformation phenomena. Irradiation-induced lattice defects become effective barriers that impede dislocation movement and therefore cause hardening of the material. There is a simplified correlation between the yield stress and the hardness [63]:

(20) where both yield stress and HV are expressed in Pa. Irradiation-induced hardening, or to be more correct an increase in yield stress ΔσYS, of the irradiated material occurs due to the

78 pinning of dislocations by the lattice defects produced under irradiation. The dispersed barrier hardening model (DBH) [64, 65] is usually applied to correlate the radiation- induced hardening with the structural damage observed by transmission electron microscopy. In this model the irradiation hardening is related to a number density of defects N of diameter d according to:

√ (21) where  is the barrier strength coefficient, M the Taylor factor,  is the shear modulus, b the Burgers vector of the dislocations, N the dislocation density and d the average dislocation size, considering dislocations as the radiation-induced defects. The barrier strength factor α highly depends on the defect type and its size as shown in Table 2.5.

Table 2.5 Barrier strength factor () as a function of defect type and size used in the dispersed barrier hardening model [66] Voids Dislocation Precipitat Defects 1-2 loops 2.3 nm 3-4 nm >4 nm es nm Barrier strength 0.15 0.25 0.3 0.35 0.4 0.6 factor (α)

Extensive investigations exist of W+ ion irradiation of W-Ta for nuclear fusion applications [67]. A 2 MeV W+ ion beam was used to achieve doses of 0.07, 1.2, 13 and 33 dpa and a damage layer of ~300 nm in thickness at a temperature of ~300°C. A summary of the hardness increase as a function of dose in W and W-5Ta is given in Table 2.6. As it can be seen in the table, saturation in hardening is observed at a high damage level in the W-5Ta alloy due to the solute atoms effect on dislocation loop motion. The loop density was reported higher in W-5Ta with average loop size smaller than in W, see Figure 2.37. With increasing irradiation temperature, the loop number density reduced. In W, at a constant dose of 1.2 dpa, this trend appeared to be linear; the loop density decreased with increasing temperature at a rate of ~1020 loops/m3/°C. The loop number density increased with increasing irradiation dose, In W-5Ta, higher densities of loops were observed than in W under identical irradiation conditions.

Table 2.6 Hardness data for W-irradiated W and W-5Ta materials [67] Damage level (dpa) Hardness (GPa) W W-5Ta 0 7.62 7.3 0.07 7.97 8.8 1.2 8.35 10.1 13 8.35 10.9 33 8.55 11.7

Figure 2.37 Loop number density in W and W-5Ta as a function of irradiation temperature for selected radiation damage levels [68].

Dislocation loops occurred in W-5Ta of higher number density and smaller mean size are shown in Figure 2.38.

Figure 2.38 Selected micrographs illustrating the radiation damage microstructure in W and W-5Ta alloy as a function of irradiation temperature and dose [68].

In the Ta-W system, the effect of alloying content on the mechanical behavior under neutron and neutron + proton irradiations have been reported [69]. Ta and Ta-1W and Ta-10W alloys were irradiated with neutrons up to a damage level of 25.2 dpa in the temperature region of 60-100°C. In Ta-1W no brittle fracture was observed at doses up to

7.5 dpa. At least 7% of total elongation was reported at 7.5 dpa. The strength of Ta-10W before irradiation was reported to be much higher in comparison with Ta and Ta-1W. Irradiation up to 7.5 dpa caused a 45% increase in yield stress and rapidly localized necking close to the yield point, see Figure 2.39. At 25.2 dpa, the ultimate tensile stress reached 1480MPa, which is 1.7 times of the yield stress before irradiation. As soon as this stress level was achieved, the sample was raptured with a very low plastic strain.

Figure 2.39 Stress-strain curve for Ta-1W (left) and Ta-10W (right) tested at room temperature after neutron + proton exposure [70].

2.4.5 Radiation-induced void swelling During the radiation cascade at longer times, self-interstitial atoms and vacancies form, due to the displacement of atoms from their equilibrium positions. The type of defect created is dependent on the material at longer time scales, type of bombarding particle, temperature, radiation damage level and dose. The evolving point defects can form clusters with the same type of defects or may be trapped and annihilated by the defects which have an opposite nature (recombination events). Denser zones of dislocation loops and tangles developed with increasing damage level cause increase in hardness. Whereas, diluted zones, related to void growth, lead to radiation swelling and eventually to material embrittlement. The effect of irradiation temperature on the defect number density, size or nature is connected with the point defect mobility. Hence, temperature affects the interaction of irradiation-induced defects with other point defects and/or accumulated clusters. In this section, the limited available information about the radiation-induced void formation in tungsten, tantalum and their binary alloys is summarized. Body-centered

82 cubic metals possess a higher resistance to swelling, as compared to face-centered cubic materials; however, the mechanism of void nucleation is still generally under discussion [71]. Voids nucleate as a result of vacancy agglomeration and were observed in tungsten and tantalum after proton, neutron and heavy ion irradiations at ≥400° C.

Void nucleation and growth in tungsten was reported to occur at Tirrad./Tmelting ~ 3 [72]. The summary of the void radius and number density after proton or neutron irradiation is presented in Table 2.7. Proton irradiation induces a larger size of voids and higher density as well, even though the irradiation temperature difference is 100°C.

Table 2.7 Proton and neutron-irradiation-induced voids in W Type of Damage level, Temperature, °C Void diameter, Void density, irradiation dpa nm 1022/m3 Proton 500 3.6 13.9 0.15 Neutron 600 2.6 6.4

Another report of neutron irradiation of tungsten shows void nucleation at 400°C and at 0.96 dpa. More than half amount of loops were of <2nm in diameter, see Figure 2.40 [73].

a b c

Figure 2.40 Voids observed in neutron-irradiated tungsten to a) 0.96 dpa at 540°C, b) 0.4 dpa at 740°C c) 1.54 dpa at 750°C [73].

The increase of damage less than 2 times was reported to cause the formation of the void lattice in tungsten as shown in Figure 2.41 and 2.42. Voids were ordered along [100] direction and void diameter reached 5 nm. The irradiation temperature was 750°C [74]. Smaller voids of 3nm in diameter ordered in a lattice were observed after neutron irradiation at 550°C and with fluence 1022n/cm2.

Figure 2.41 Incipient void lattice formation observed in neutron-irradiated tungsten up to a fluence of 12×1022 neutrons/cm2 and at a temperature of 750°C [74].

Figure 2.42 Ordered array of voids in tungsten irradiated at ~550°C to a fast neutron fluence of  1×1022 n/cm2 [75].

Table 2.8 Comparison of the reported irradiation conditions for void lattice formation in tungsten [73-75] Reported data of neutron Irradiation parameters irradiation

T (°C) 550 750 Damage level (dpa) n/a 1.54 Fluence (ions/cm2)/ (n/cm2) ~1022 ~1021 Energy (MeV) n/a >0.1 n/a = not applicable.

The effect of neutron irradiation up to a fluence of 2.5×1022 neutrons/cm2 and at 585°C on tantalum leads to void ordering into a bcc lattice along the [111] direction (Figure 2.43). The average void diameter was determined to be 6.1nm. Voids of smaller size have not been detected in tantalum at an irradiation temperature <600°C. However, void-induced swelling data available from immersion density measurements and experiment-oriented modeling revealed the occurrence of void swelling at temperatures of ~200°C or higher, with a peak swelling at 635°C [76]. Published data on irradiation conditions of tungsten and tantalum which led to void lattice formation are given in Table 2.8 and Table 2.9.

Table 2.9 Comparison of the reported irradiation conditions for void lattice formation in tantalum. Proton Neutron irradiation Ni+ irradiation [77] [78] (current thesis) T (°C) 345 585 >925 Damage level (dpa) 0.25 ≥35 ≥55 Damage rate (dpa/sec.) 3.3 10-6 n/a n/a Fluence (ions/cm2)/ ~1018 ~1022 n/a (n/cm2) Energy (MeV) 3 >0.1 3.2

Figure 2.43 Void ordering observed in tantalum after neutron irradiation up to 2.5×1022 neutrons/cm2 and at 585°C [77].

So, with increasing radiation dose, voids may start to become ordered from initially being randomly distributed within the matrix. Mostly these radiation-induced void lattices are isomorphic with the material crystal structure [79]. The proposed mechanism in the literature for the void lattice formation is based on the anisotropic energy transfer provided by long propagating discrete breathers. The existence of local excitations in complex systems has been known for a long time. The breather is state localized in space and a time-periodic. However, this state is structurally unstable due to the existence of an unlimited spectrum of small oscillations. In the discrete system, the distance between atoms is fixed, which leads to the presence of discrete translational invariance. A discrete system containing defects may have both, traveling waves and vibrational defect localized modes. If a periodic discrete system is nonlinear, it can support spatially localized vibrational modes as exact solutions even in the absence of defects. Because the atoms of the system are all in equal positions, only a special choice of the initial conditions allows selecting a group of atoms with this mode, called a discrete breather. Basically, discrete breathers are spatially localized large-amplitude vibrational modes in lattices that exhibit strong anharmonicity [80]. In a crystal lattice without defects, the dissipation of kinetic energy via discrete breathers motion is low. But defects distort the lattice symmetry and act as scattering centers for the discrete breathers. Scattering of discrete breathers on the lattice defects causes localized atomic excitations. Intensity and relaxation time of these excitations depend on the defect structure and kinetic energy of the breathers. A scattering of discrete breathers on the void surfaces excites the surface atoms, which increases the rate of the vacancy emission from voids. As the breather propagation range is larger than the void spacing, voids can be

86 shielded from the breather fluxes along close-packed directions. And this can serve as a driving force for the void ordering as illustrated in the Figure 2.44.

Figure 2.44 Dissolution of a void in the ‗interstitial‘ position due to the absorption of breathers coming from larger distances as compared to ‗regular‘ voids. The voids shield each other from the breather ﬂuxes along the close-packed directions [81].

Another mechanism of void ordering is based on the diffusion anisotropy of SIAs and small interstitial dislocation loops along closed packed directions or planes [82]; [83]; [84]. It was reported that at the onset of void ordering, lattices form on close-packed crystal planes that results in voids ordered along close-plane directions. In bcc metals a 1-D SIA diffusion mechanism may occur along the close-packed <111> directions Basically, voids start nucleating randomly and the voids which are aligned with one another along the direction or plane of SIA migration will be shadowed from anisotropically migrating SIAs. There is a term crowdion-supply cylinder which extends from the void surface along the close-packed direction [85]. There are 8 cylinders for a void in bcc lattice.The radius of the cylinder is equal to the void radius. Schematic of the crowdion-supply cylinder is illustrated in Figure 2.45.

Figure 2.45 Crowdion-supply cylinders (length Lc) of a void which run parallel to the close-packed directions of the host lattice [85].

When the separation of two voids along a close-packed direction is sufficiently small (l ∼ λ), their crowdion-supply cylinders overlap or a cylinder of the one void even protrudes into the other void, see Figure 2.46. Thus, a reduction of the on supplies to both voids occurs. As a result, the net vacancy fluxes to both voids increase, which leads to an enhanced growth of these voids.

Figure 2.46 The crowdion-supply overlap of the crowdion-supply cylinders [82].

For random void distributions, the only locations which may serve as sites for further nucleation are locations at which voids can have one or two overlapping ISCs. The latter ones are more favorable because in there voids are able to grow to larger sizes and are thus more resistant to shrinkage. 2.4.6 Nuclear transmutation reactions Neutron capture reactions can cause transmutations to different elements within the target material, which in its turn change the material‘s properties. The transmutation of nuclei usually happen through (n,γ), (n,p), (n,np) and (n,α) reactions [86], where n - neutron, p - proton,  - alpha particle, β - beta particle,  - gamma photon. Naturally-occurring tungsten have five stable isotopes: 180W(0.1%), 182W(26.3%), 183W(14.3%), 184W(30.7%), and 186W(28.6%). Most of the transmutation products in tungsten are generated via (n,γ) and (n,2n) reactions that mainly produce rhenium and osmium through the following paths [87]:

{ (22)

{ (23)

Figure 2.47 represents the sketch of the reactions. Neutron absorption in 180W generates traces of tantalum through β+ decay of 181W.

Figure 2.47 Pathways of rhenium and osmium generation from tungsten under neutron irradiation. The isotopes highlighted in blue are stable isotopes [88].

In ITER after 2 DD (deuterium-only plasma) and 12 DT (fusion of deuterium and tritium atoms) years of operation, W is predicted to give rise to W – 0.18%Re alloy, with the additional formation of 1 appm of He and 2 appm of H. The transmutation of tungsten into rhenium and later in osmium (W-Re-Os) induced by neutron bombardment during the reactor operation is predicted to place the alloy composition close to the phase field of brittle  precipitates [89]. However, the helium formed during the reaction may stabilize defect production [90]. Tantalum, which is considered as a potential first wall material substitute to tungsten, either as a through-thickness or coating material, after 5 years of 2 MW m-2 neutron load wall will lead to the alloy containing 22.9Ta, 76.5W, 0.16Os and 0.34Re (at.%), with traces of other elements [9].

Figure 2.48 Tantalum transmutation characteristics as a function of irradiation time [9].

Summary of the power-plant transmutation products for W and Ta after 1, 3 and 5 years is given in Table 2.10. In Figure 2.48 we can see that half of the Ta transforms into W after 2.5 years of service, and there is only a small production of Re and Os at that time. That is why the use of Ta is expected to help solve the problem of the brittle σ-phase formation in W. Table 2.11 summarizes the neutronic response of tantalum and tungsten, together with other candidate structural materials.

Table 2.10 Summary of the power-plant transmutation products for W and Ta. [89]

Table 2.11 Neutronic response of first-wall/blanket structural material candidates [91].

Tungsten alloyed with tantalum will experience the formation of a lower amount of transmutation products such as Re or Os at a specific time in reactor operation. [89]. Hence, with the right choice of the starting material composition, it is possible that the concentration of W could be kept relatively constant under fusion neutron-irradiation conditions, thereby preserving the optimal mechanical properties of the material during the expected lifetime of the W-based component in service.

2.5 Main research tasks, limitations found in literature and objectives of the project

2.5.1 Current status of the work in the literature

Tungsten is going to be used as an armour material for the divertor in ITER and DEMO. Therefore, the use of pure tungsten for divertors may experience several challenging problems: high levels of displacement damage due to fast neutrons, the production of helium by injection from the plasma and in (n, α) reactions; and the formation of transmutation products such as rhenium, osmium, and tantalum, which is known form brittle precipitates in tungsten. The mechanisms listed above will contribute to the increase in hardening and embrittlement of tungsten, which is extremely brittle material already prior irradiation. Summarizing attractive for fusion properties of tungsten, such as high thermal conductivity, low vapour pressure (10-7 Pa), radiation swelling resistance allowing to extend the lifetime of the divertor, and high energy threshold for sputtering, tungsten needs to improve its ductility as its brittle nature can be fatal and may cause failure during low-temperature reactor conditions. Alloying tungsten with a similar refractory element such as tantalum is considered to suppress void and dislocation formation which are responsible for the material hardening in tungsten and hence, will improve mechanical properties. Current data on W-Ta alloys in fusion-relevant radiation environments remains limited and present mostly information related to changes mechanical properties of W-Ta alloys under irradiations. Hence, this data requires a better understanding of the mechanisms of structural damage formation and microstructural evolution of this damage in W-Ta alloys. Additionally, alternative option as substitution tungsten with tantalum as a coating or full-thickness material needs more detailed research as the correlation of the hardening phenomenon with the characteristics of the irradiated structures of Ta-W alloys remains unknown and understanding and experimental data about the behaviour of W-Ta alloys under irradiation and its damage structure evolution at elevated temperatures require more detailed investigation. There also have been limited studies into the effect of proton irradiations of these materials as a technique to simulate effects of neutron irradiation. In terms of hardness, protons produce hardness results similar to neutron damage. This makes it even more important to continue investigations in proton irradiations as mimicking fusion neutron damage.

2.5.2 Limitations

Limited studies have been found into the effect of proton irradiations as a technique to emulate neutron irradiation, particularly if we compare it to the research devoted to neutron irradiations. The only major study comparing proton irradiations to neutron irradiations, specifically for fusion applications has been carried out by He et al [92]. However, these have only been performed at temperatures of 500-600°C. Hence, there is a scope to further investigations of using proton beams as a proxy for neutron damage, in particular, for fusion applications. In ITER the temperatures reached on the outer target of the divertor could vary between 200-1150°C [93]. It will, therefore, be beneficial to further study the impact of temperature on the proposed materials under irradiation by conducting experiments at temperatures close to the expected ones in order to look at its impact on irradiation-induced damage. In the literature, there was not much information observed on the radiation damage study of alloyed compositions within W-Ta system and the influence of the alloying component on irradiation-induced defect evolution. However, evidence on microstructural defect development which is responsible for the changes in mechanical properties is required. Tantalum is predicted to have a repulsive interaction with self-interstitial atom (SIA) defects such as <111> crowdion [94] and dumbbell [95] configurations, and potentially restrict the mobility of SIAs and interstitial loops in tungsten. Additionally, hardness increase at lower damage levels but earlier hardness saturation were observed in alloys exposed to 2 MeV W+ ion irradiation at 300oC [96] whereas the lower rate of hardness increase (at higher damage level) with a later stage of hardness saturation were reported for pure tungsten after 800 MeV proton irradiation [97]. Fragmented microstructural data on self-ion irradiation of W and W-Ta revealed smaller loop size but higher number density of those loops in W-Ta rather than in W. The correlation of the hardening phenomenon with the characteristics of the irradiated structures remains unknown and needs to be investigated. Additional interest is in the study of potential lattice defects which can migrate into a free surface that presents an effective sink for the defects. Here precise comparison is needed between ex-situ and in-situ analysis at certain damage levels. Dislocation loops caused by proton radiation applying energies between 0.5-8 keV in the temperature range of 20-800°C were reported to slip and escape into the sample free surface in both single crystal and polycrystalline tungsten material, during an in-situ TEM experiment [98].

Again the effect of alloying component will present a special interest in this part of the research as well.

2.5.3 Thesis objectives

The scientific and technological challenges which are going to be addressed in this thesis in order to consider compositions of the tungsten-tantalum system as well as alternative tantalum-based alloys as the material candidates for the plasma-facing components such as divertor in the future fusion reactors are the following:  Implementation of intense proton irradiation experiments in a range of temperatures and damage levels to observe the defects at very early stages of their nucleation;  Explore the ways of improving the mechanical performance of tungsten under heat- radiation loads of fusion environment considering two concepts, such as alloying tungsten with controlled amounts of tantalum (W5wt.%Ta) and studying tantalum- based compositions as a potential alternative to tungsten;  Understand and control mechanisms of the nucleation and development of irradiation-induced nano-scale structure defects in the alloys within the tungsten- tantalum system at different damage levels and in a range of temperatures;  Reveal the peculiar features of the alloying component in W-based as well as Ta- based compositions;  Improve the mechanical properties at high temperatures, in order to expand the upper limit of the operating temperature window of the divertor;  Improve the resistance to radiation hardness, in order to expand the lower limit of the temperature window.

2.6 References 1. M. Arita, I.N., Tungsten films with A15 structure. Journal of applied physics, 1993. 32: p. 1759-1764. 2. Y.G. Shen, Y.W.M., Structure and properties of stacking faultedA15 tungsten thin films. Journal of Materials science, 2001. 36: p. 93-98. 3. A. S. Kurlov, A.I.G., Tungsten Carbides. Structure, properties and application in hardmetals. Vol. 184. 2013: Springer. 4. http://www.geocities.jp/ohba_lab_ob_page/structure6.html. [cited 2017 15.09]. 5. T.E. Tietz, J.W.W., Behavior and properties of refractory metals. 1965: London : Edward Arnold. 6. Y.M. Smirnov, V.A.F., Crystal structure of tantalum, niobium, and vanadium at 110—400°K. SOVIET PHYSICS JETP, 1966. 22(4).

7. A. Jiang, A.Y., N.O. Nnolim, T.A. Tyson, L. Axe, S.L. Lee, P.Cote, Investigation of the structure of b-tantalum. Thin Solid Films, 2003. 437: p. 116-122. 8. K. Masuda-Jindo, V.V.H.N.T.H., P.E.A. Turchi, First principles calculations of thermodynamic and mechanical properties of high temperature bcc Ta–W and Mo– Ta alloys. Journal of Alloys and Compounds, 2008. 452: p. 127-132.

9. Cottrell, G.A., Sigma phase formation in irradiated tungsten, tantalum and molybdenum in a fusion power plant. Journal of Nuclear Materials, 2004. 334: p. 166–168.

10. R. Herschitz, D.N.S., An atomic resolution study of homogeneous radiation- induced precipitation in a neutron irradiated W-10at.% Re alloy. Acta metallurgica, 1984. 32(8): p. 1155-1171.

11. R. Herschitz, D.N.S., An atomic resolution study of homogeneous radiation- induced precipitation in a neutron irradiated W-25at.% Re alloy. Acta metallurgica, 1984. 32(8): p. 1155-1171.

12. M. Fukuda, K.Y., S. Nogami, Akira Hasegawa, T. Tanaka, Microstructural development of tungsten and tungsten–rhenium alloysdue to neutron irradiation in HFIR. Journal of Nuclear Materials, 2014. 455: p. 460-463. 13. M. Ekman, K.P., G. Grimvall, Phase diagram and lattice instability in tungsten±rheniumalloys. Journal of Nuclear Materials, 2000. 278: p. 273-276.

14. T. Tanno, A.H., M. and J.-C.H. Fujiwara, S. Nogami, M. Satou, T. Shishido, K. Abe, Precipitation of Solid Transmutation Elements in Irradiated Tungsten Alloys. Materials Transactions, 2008. 49(10): p. 2259-2264. 15. Vineyard, G.H., Frequency factors and isotope effects in solid state rate processes. Journal of Physics and Chemistry of Solids, 1957. 3: p. 121-127.

16. S. Hok, M.D. and A.l.o. the, A measurement of the surface self-diffusion of tantalum. Surface science letters, 1981. 107: p. L362-L366. 17. J. N. Mundy, S.J.R., N. Q. Lam, H. A. Hoff, L. J. Nowicki, Self-diffusion in tungsten. Physical review B, 1978. 18(12). 18. W. Wang , J.R., S. Lindig , C. Wu, Blister formation of tungsten due to ion bombardment. Journal of Nuclear Materials, 2001. 299: p. 124-131. 19. K. Heinola , T.A., K. Nordlund , J. Keinonen, Hydrogen interaction with point defects in tungsten. Physical Review B, 2010. 82 (9): p. 094102. 20. N. N. Degtyarenko, A.A.P., Simulation of diffusion of hydrogen atoms in the lattice of tungsten. Journal of Physics: Conference Series, 2016. 748: p. 012010. 21. A.D. Romig, M.J.C., Interdiffusion in the Ta-W system. Journal of Applied Physics, 1985. 58: p. 3425. 22. I. A. Tregubov, L.N.K., 0. S. Ivanov, THE INTERDIFFUSION OF TANTALUM AND TUNGSTEN. 1968: Academy of science of USSR.Reports. p. 423-424. 23. Y. Nemoto, A.H., M. Satou, K. Abe, Microstructural development of neutron irradiated W±Realloys. 2000. 283-287: p. 1144±1147. 24. J.N. Brooks, L.E.-G., A. Hassanein, T. Sizyuk, Plasma-facing material alternatives to tungsten Nuclear fusion, 2015. 55: p. 043002. 25. Murty, M.V.R., Sputtering: the material erosion tool. Surface Science, 2002. 500: p. 523-544. 26. S. Takamura, N.O., D. Nishijima, S. Kajita, Formation of nanostructured tungsten with arborescent shape due to helium plasma irradiation. Plasma and fusion research, 2006. 1: p. 051.

27. M.J. Baldwin, R.P.D., Helium induced nanoscopic morphology on tungsten under fusion relevant plasma conditions. Nuclear Fusion, 2008. 48: p. 035001. 28. W.M. Shu, E.W., T. Yamanishi, Blister bursting and deuterium bursting release from tungsten exposed to high fluences of high flux and low energy deuterium plasma. Nuclear Fusion, 2007. 47: p. 201-209.

29. S. Kajita, N.Y., R. Youshihara, N. Ohno, M. Yamagiwa, TEM observation of the growth process of helium nanobubbles on tungsten: Nanostructure formation mechanism Journal of Nuclear Materials, 2011. 418: p. 152-158. 30. M.J. Baldwin, R.P.D., Formation of helium induced nanostructure ‗fuzz‘ on various tungsten grades. Journal of Nuclear Materials, 2010. 404 p. 165-173. 31. T.J. Nowakowski, J.K.T., A. Hassanein, Effect of high-flux, low-energy He+ ion irradiation on Ta as a plasmafacing material. Scientific reports, 2016. 6: p. 39746.

32. S. Takamura, Y.U., Experimental identification for physical mechanism of fiber- form nanostructure growth on metal surfaces with helium plasma irradiation Applied Surface Science 356, 2015. 356: p. 888-897. 33. D. Raabe, G.S., H. Weisshaupt, K. Lücke, Textures of cold rolled and annealed tantalum. Materials Science Forum, 1994. 10: p. 299-305. 34. D. Beckenhauer, P.N., R. Pick, Effect of heating rate on the recrystallization temperature of tantalum. 1993. 12: p. 449-450. 35. K. Farrell, A.C.S., J. O. Stiegler, Recrystallization, grain growth and the ductile- brittle transition in tungsten sheet. Journal of the less common metals, 1967. 13: p. 141-155.

36. J. Reiser, M.R., B. and A.H. Dafferner, X.Yi, Tungsten foil laminate for structural divertor applications – Analysis and characterisation of tungsten foil. Journal of Nuclear Materials, 2012. 424: p. 197-203.

37. Briant, C.L. and D.H. Lassila, The Effect of Tungsten on the Mechanical Properties of Tantalum. Journal of Engineering Materials and Technology, 1999. 121(2): p. 172-177. 38. Walt, C.M.v.d., Slip in the bcc metals. Acta Metal., 1989. 17. 39. L.A Zepeda-Ruiz, J.M., B.D. Wirth, Philosophical magazine, 2005. 85. 40. J. Weertman, J.R.W., Elementary Dislocation Theory. 1993: Oxford University Press. 41. P. M. Derlet, D.N.-M., S. L. Dudarev, Multiscale modeling of crowdion and vacancy defects in body-centered-cubic transition metals. Physical review B, 2007. 76: p. 054107. 42. B. L. Eyre, R.B., On the formation of interstitial loops in b.c.c.metals. Philosophical Magazine, 1965. 12: p. 31-39.

43. H. Bolt, V.B., G. Federici, J. Linke, A. Loarte, and K.S. J. Roth, Plasma facing and high heat flux materials – needs for ITER and beyond. Journal of Nuclear Materials, 2002. 307-311: p. 43-52.

44. S.J. Zinkle, N.M.G., Operating temperature windows for fusion reactor structural materials Fusion Engineering and Design, 2000. 51-52: p. 55-71. 45. A. C. Chilton, A.S.W., The effects of strain rate and pressurization on the ductile- brittle transition temperature of polycrystalline sintered tungsten Jouwzal of the Less-Common Metals, 1969. 17: p. 447-450. 46. M. Rieth, S.L.D., S. M., J.A. Gonzalez de Vicente, T. Ahlgren, S. Antusch, D. E. J. Armstrong, M., and N.B.e.a. Balden, A brief summary of the progress on the EFDA tungsten materials program. Journal of Nuclear Materials, 2013. 442 p. 5173-5180.

47. D. Jiang, Q.W., W. Hu, Z. Wei, The effect of tantalum (Ta) doping on mechanical properties oftungsten (W): A first-principles study. Journal of material research, 2016. 31(21): p. 3401-3408. 48. Jung, P., A. Schwarz, and H.K. Sahu, An apparatus for applying tensile, compressive and cyclic stresses on foil specimens during light ion irradiation. Nuclear Inst. and Methods in Physics Research, A, 1985. 234(2): p. 331-334. 49. L.M. Giancarli, M.A., D.J. Campbell, V.A. Chuyanov, M.Y. Ahn, M. Enoeda, C. Pane,Y. Poitevin, E. Rajendr Kumar, I. Ricapito, Y. Strebkov, S. Suzuki, P.C. Wong, M. Zmitko, Fusion Engineering and Design, 2012. 87: p. 395-402. 50. Seitz, F., On the disordering of solids by action of fast massive particles. Discussion of the Faraday society, 1949. 5: p. 271–282. 51. W.S. Snyder, J.N., Disordering of Solids by Neutron Radiation. Physical Review, 1955. 97: p. 1636-1646. 52. G.H. Kinchin, R.S.P., The Displacement of Atoms in Solids by Radiation. Reports on Progress in Physics, 1955. 18: p. 1-51. 53. Was, G., Fundamentals of radiation materials science: metals and alloys. 2007. 54. J. Lindhard, M.S., H.E. Schiot, Mat. Fys. Medd. Dan. Vid. Selsk., 1963. 33: p. 1– 42 55. M.T. Robinson, I.M.T., Computer simulation of atomic-displacement cascades in solids in the binary-ccollisionapproximation. Physical Review B, 1974. 9: p. 5008- 5024.

56. M. J. Norgett, M.T.R., I. M. Torrens, A proposed method of calculating displacement dose rates. Nuclear Engineering and Design, 1975. 33: p. 50-54.

57. S.J. Zinkle , S.K., Helium-assisted cavity formation in ion-irradiated ceramics. Journal of Nuclear Materials, 1991. 179-181: p. 395-398. 58. Was, G., Challenges to the use of ion irradiation for emulating reactor irradiation. Journal of Materials Research, 2015. 30: p. 1-25. 59. S.L. Dudarev, M.R.G., K. and H.M. Arakawa, Z. Yao, M.L. Jenkins, P.M. Derlet, Langevin model for real-time Brownian dynamics of interacting nanodefects in irradiated metals. Physical Review B. Condensed Matter and Materials Physics, 2010. 81: p. 1-15.

60. Abromeit, C., Aspects of simulation of neutron damage by ion irradiation. Journal of Nuclear Materials, 1994. 216: p. 78-96.

61. Mansur, L.K., Theory of transitions in dose dependence of radiation effects in structural alloys. Journal of Nuclear Materials, 1993. 206: p. 306-323. 62. Mansur, L.K., Correlation of neutron and heavy-ion damage : II. The predicted temperature shift if swelling with changes in radiation dose rate. Journal of Nuclear Materials, 1978. 78: p. 156-160. 63. J.R. Cahoon, W.H.B., A.R. Kutzak, The Determination of Yield Strength from Hardness Measurements. Metallurgical and Materials Transactions B, 1971. 2: p. 1979. 64. Seeger, A.K. On the theory of radiation damage and radiation hardening. in Second United Nations International Conference on the Peaceful Uses of Atomic Energy. 1958. Geneva, Switzerland: United Nations, New York.

65. Taylor, G.I., The mechanism of plastic deformation of crystals. Proceedings of the Royal Society A, 1934. 145: p. 362. 66. X. Hu, T.K., M. Fukuda, N.A.P. Kiran Kumar, L.L. Snead, B.D. Wirth, Y. Katoh, Irradiation hardening of pure tungsten exposed to neutron irradiation. Journal of Nuclear Materials, 2016. 480: p. 235-243. 67. Roberts, D.E.J.A.a.A.J.W.a.S.G., Mechanical properties of ion-implanted tungsten– 5 wt% tantalum. Physica Scripta, 2011. 2011(T145): p. 014076.

68. X. Yi, M.L.J., K. Hattar, P. D. Edmondsone, S. G. Roberts, Characterisation of radiation damage in W and W-based alloys from 2 MeV self-ion near-bulk implantations. Acta Materialia, 2015. 92: p. 163-177.

69. T.S. Byun, S.A.M., Dose dependence of mechanical properties in tantalum and tantalum alloys after low temperature irradiation. Journal of Nuclear Materials, 2008. 377(1): p. 72-79. 70. Byun, T.S. and S.A. Maloy, Dose dependence of mechanical properties in tantalum and tantalum alloys after low temperature irradiation. Journal of Nuclear Materials, 2008. 377(1): p. 72-79. 71. B. Singth, J.E., Significant differences in defect accumulation behaviour between fcc and bcc crystals under cascade damage conditions Journal of Nuclear Materials, 1995. 226: p. 277-285. 72. J.C. He, G.Y.T., A. Hasegawa, K. Abe, Microstructural development and irradiation hardening of W and W–(3–26) wt%Re alloys after high-temperature neutron irradiation to 0.15 dpa. Nuclear Fusion, 2006. 46(11). 73. T. Tanno, M.F., S. Nogami, A. Hasegawa, Microstructure Development in Neutron Irradiated Tungsten Alloys. Materials Transactions, 2011. 52(7): p. 1447-1451.

74. T. Tanno, A.H., J.-C. He, M. Fujiwara, S. Nogami, M. Satou, T. Shishido, K. Abe, Effects of Transmutation Elements on Neutron Irradiation Hardening of Tungsten. Materials Transactions, 2007. 48(9): p. 2399-2402. 75. V. K. Sikka, J.M., Superlattice of voids in neutron-irradiated tungsten. Journal of Applied Physics, 1972. 43: p. 4942. 76. J. F. Bates, A.L.P., Dimensional Stability of Tantalum Control Rods in LMFBRs. Nuclear Technology, 1972. 16: p. 406-409. 77. Wiffen, F.W., The microstructure and swelling of neutron irradiated tantalum. Journal of Nuclear Materials, 1977. 67(1): p. 119-130. 78. Loomis, B.A. and S.B. Gerber, Ordered void arrays in ion irradiated Ta. Journal of Nuclear Materials, 1978. 71(2): p. 377-378. 79. Evans, J.H., Void and bubble lattice formation in molybdenum: A mechanism based on two-dimensional self-interstitial diffusion. Journal of Nuclear Materials, 1983. 119(2): p. 180-188. 80. D. A Terentyev, A.V.D., V. I. Dubinko, S. V. Dmitriev, E. E. Zhurkin, M. V. Sorokin, Interaction of discrete breathers with primary lattice defects in bcc Fe.

Modelling and Simulation in Materials Science and Engineering, 2015. 23: p. 085007. 81. V.I. Dubinko, A.G.G., E. Melnichenko, R. Vasilenko, Radiation-induced reduction in the void swelling. Journal of Nuclear Materials, 2009. 385: p. 228-230. 82. A.A. Semenov, C.H.W., W. Frank, Diffusion anisotropy and void development under cascade irradiation. Applied Physics A, 2008. 93: p. 365-377.

83. Jäger, W. and H. Trinkaus, Defect ordering in metals under irradiation. Journal of Nuclear Materials, 1993. 205: p. 394-410. 84. A.V. Barashev, S.I.G., On the onset of void ordering in metals under neutron or heavy-ion irradiation Philosophical Magazine, 2010. 90: p. 1787-1797. 85. C.H. Woo, W.F., A theory of void-lattice formation. Journal of Nuclear Materials, 1985. 137: p. 7-21. 86. J. Knaster, A.M., T. Muroga, Materials research for fusion. Nature Physics, 2016. 5(2): p. 77-80. 87. Greenwood, L.R., Neutron interactions and atomic recoil spectra. Journal of Nuclear Materials, 1994. 216: p. 29-44. 88. R.Pampin, Tungsten transmutation and resonance self-shielding in PPCS models for the study of sigma-phase formation. 2005.

89. M.R. Gilbert, J.-C.S., Neutron-induced transmutation effects in W and W-alloys in a fusion environment Nuclear fusion, 2011. 51(4): p. 043005. 90. Yoshida, N., Review of recent works in development and evaluation of high-Z plasma facing materials Journal of Nuclear Materials, 1999. 266-269: p. 197-206. 91. D.L. Smith, S.M., M. Billone, R. Mattas, Performance limits for fusion first-wall structural materials. Journal of Nuclear Materials, 2000. 283-287: p. 716-720. 92. J.C. He, A.H., K. Abe, Effects of transmutation elements on the defect structure development of W irradiated by protons and neutrons. Journal of Nuclear Materials, 2008. 377(2): p. 348-351.

93. S.J. Zinkle, L.L.S., Designing Radiation Resistance in Materials for Fusion Energy The Annual Review of Materials Research, 2014. 44: p. 241. 94. M. Muzyk, D.N.-M., K. J. Kurzydłowski, N. and S.L.D. L. Baluc, Phase stability, point defects, and elastic properties of W-V and W-Ta alloys. Physical Review B 2011. 84 p. 104115-1- 104115- 7.

100

95. W. Setyawan, G. and R.J.K. Nandipati, Ab initio study of interstitial cluster interaction with Re, Os, and Ta in W. Journal of Nuclear materials, 2017. 484: p. 30-41. 96. D.E.J. Armstrong, A.J.W., S.G. Roberts, Mechanical properties of ion-implanted tungsten–5 wt% tantalum. Physica Scripta, 2011. 2011(T145): p. 014076. 97. S.A. Maloy, M.R.J., W. Sommer, G.J. Willcutt, M. Lopez, T.J. Romero, The Effect of 800 MeV Proton Irradiation on the Mechanical Properties of Tungsten. Materials Transactions, 2002. 43: p. 633-637. 98. R. Sakamoto, T.M., N. Yoshida, Microstructural evolution induced by low energy hydrogen ion irradiation in tungsten. Journal of Nuclear Materials. 220-222: p. 819-822.

101

“An experiment is a question which science poses to Nature, and a measurement is the recording of Nature’s answer.” Max Planck (1858 – 1947) 3 Experimental methods In this section, the sample preparation procedures and main working principles of the experimental techniques applied during this research project are discussed, together with some limitations of the selected methodologies. The principal technique used for characterizing samples before and after irradiation was microscopy: optical microscopy to analyse the sample surface quality during its metallographic preparation, and electron (scanning and transmission) microscopy to analyse the microstructure of the studied samples before and after irradiation. 3.1 Sample preparation In order to prepare samples for microstructural analysis prior irradiation experiments it is necessary to achieve a good quality of the sample surface. This process comprised two main steps:  Grinding – to remove all coarse imperfections from the initial surface, e.g. holes, scratches, roughness, pits, cavities, etc.;  Polishing – to achieve a mirror-like final surface suitable for electron microscopic analysis. The cutting equipment was used to cut the samples initially to smaller pieces of ~1cm2. Following the cutting procedure, samples were annealed for recrystallization in high vacuum (~10-5 Torr) at 1400°C for <10h. The annealed samples were embedded into resin, using a hot mounting press, to be prepared for the next steps including mechanical grinding and polishing. The choice of the right abrasive, polishing cloth and lubricant highly depends on the hardness and ductility of the material. SiC abrasive papers with successively smaller grit size (from grit 220 up to 4000) were used for mechanical grinding, followed by polishing cloths made of woven wool or selected fabrics and applying polishing suspensions. After grinding during ≥1 minute on each SiC paper and using a force of 10-30N (depending on sample dimensions), the sample thickness decreased by ≤200μm. For tantalum-based materials after grinding, the process moved from an abrasive grain size of about 6μm in the 4000 grit SiC paper to a grain size of 1μm, then to a grain size of 0.3μm using OP-A alumina suspension and finally to 0.04 µm applying OP-S silica suspension. In case of tungsten-based

102

materials, diamond suspensions were applied for the polishing steps. Table 3.1 contains a summary of the grinding and polishing conditions.

Table 3.1 Sample preparation procedure adopted for tantalum- and tungsten-based materials.

Step I Grinding Step II Polishing Sample Material Time Force Material Time Force

1 min 10 N OP-A 1μm 5 min

220-800 SiC based based

OP-A 0.3μm 5 min 30 N 1200-4000 SiC 2 min 30-35 N materials 2 min OP-S 0.04μm 10 min

Tantalum 4 min

220-800 SiC 1 min 15 N DiaPro Dur 3 min

3μm

s 1200-2000 SiC based

2 min 35-40 N DiaPro Nap 3 min 35 N Largo/DiaPro

material 4 min 1μm Largo 9 μm

Tungsten 3 min 20 N OP-S 0.04μm 10 min

For a successful transmission electron microscope (TEM) analysis, the final specimen in a form of a thin foil should have a zone transparent for electrons. For this purpose the foils were prepared by the twin-jet electropolishing procedure until the foil perforation. After electropolishing the foil thickness must reach ~100 nm or less in the area of interest, depending on the electron acceleration voltage in the microscope. To obtain such a type of specimen, the initial material in a form of metallic sheet needs to be ground from both sides using SiC paper (220 grit – 4000 grit) to a final foil thickness ≤100µm. Experimentally it was found that the tungsten-based foil should be of 70 – 80 μm in thickness, because the material is brittle. A lower thickness of the specimen at this step may help to avoid cracking during 3mm disc punching. After 3mm-diameter discs have been punched, the process moves on to the electropolishing procedure using the Struers TenuPol-5 polishing unit together with a Julabo FP50 close-cycle cooling unit. Table 3.2 collects the temperature and the electrolytes used for different materials.

103

Table 3.2 Electrolytes and temperature used for electropolishing Ta- and W-based alloys Material Electrolyte Temperature, °C Ta-based alloys 15vol.% sulphuric acid (95%) and

W-5Ta 85vol.% methanol (3M H2SO4 in ~5°C methanol)

W aqueous solution of 0.5 wt.% Na2S

The final quality of the foil after electropolishing depends on several factors: current density, foil thickness, temperature of the electrolyte, electrolyte composition and time. The method of electropolishing implies applying an electric potential difference between the surface and a liquid solution which could be acid or basic, namely the etching solution. The sample disc is being thinned from both sides simultaneously until two concave shaped pits form. As the two pits join together and the jets touch each other, a hole with a diameter of the order of the micrometers in diameter forms. The border of that hole can be tens of nm thick, enough for electrons with energy of 100-300 keV to pass through and therefore allow the subsequent TEM analysis. In order to stop the process at the right time a laser is pointed perpendicularly from one side of the disk. As soon the hole is formed the laser light passes through it and is detected on the opposite side of the sample. Figure 3.1 shows a plot of the electrical current passing through the surface as a function of the electricpotential during electropolishing. Three main phenomena may occur with increasing current, namely etching, polishing and pitting. That is why it is important to find the suitable conditions for smooth electropolishing of a particular sample.

Figure 3.1 Electric current as a function of potential during electropolishing [1].

104

The electropolishing method for preparing irradiated samples is slightly different. After being exposed to proton irradiation, the 1mm-thick specimens require a relatively gentle preparation to remove certain amount of the material during electropolishing in order to reveal desirable irradiated layer. Firstly, the proton-irradiated sample should be ground from the non-irradiated side of the sample with SiC abrasive paper (500-4000grit) to a thickness of ~100 µm, and afterwards a 3 mm-diameter disc is punched from it. The next step is to mount the disc in a twin-jet for Tenupol-5 electropolishing unit. The same electrolytes as for the non-irradiated bulk material were used in this case (Table 3.2). The proton irradiation dose in displacements per atom (dpa) was calculated by the quick Kinchin-Pease calculation in SRIM [2]. The 3MeV proton beam penetrates ~32 µm in depth in Tantalum, so this depth of interest should not be damaged during subsequent TEM disc preparation by electropolishing. The procedure can be described in four main steps, see Figure 3.2. Prior to electropolishing, the non-irradiated side of the sample was covered with Elektron Technology’s acid-resistant Lacomit varnish [3]. The adequate amount of material needs to be removed in order to reveal the zone of interest at the target depth from the irradiated layer. Afterwards, the Lacomit varnish is used to protect the ‘just polished’ irradiated side from the aggressive acid present in the electrolyte and the sample is being perforated from the non-irradiated side. In order to remove this varnish easily at the end of the described procedure, acetone was used. The amount of material removed as a function of time was assessed using a Keyence VK-X200K 3D Laser Scanning Microscope.

Figure 3.2 A sketch of the procedure used for the preparation of irradiated samples for TEM analysis.

105

3.2 Ion irradiation experiments Displacement damage calculation with SRIM The displacement damage was calculated using SRIM for the proton energy of 3 MeV. The displacement damage calculated by SRIM is defined as the number of displacements produced per ion per unit length. These values are calculated as a function of depth into the material (Figure 3.3). The curve shows increase of damage level toward the end of range and achieves its maximum at the Bragg peak (31μm), shortly before the energy drops to zero. This so-called Bragg curve describes the energy loss of ionizing radiation during its travel through the material. 1.20 Total DPA 1.00 0.80

0.60

DPA 0.40 Damage 0.20 0.00 0.00 10.00 20.00 30.00 40.00 50.00 -0.20 Depth (μm)

Figure 3.3 SRIM calculated displacement damage profile as a function of depth for 3 MeV proton beam irradiation of pure Ta at current of 9.5 mA during 36 hours with accumulated charge of 1.39 Coulombs.

Ex-situ proton irradiation at the Dalton Cumbrian Facility (DCF) Samples in the form of metallic sheets were irradiated with a 3 MeV proton beam produced using a 5 MV Tandem Pelletron accelerator at the Dalton Cumbrian Facility of the University of Manchester [4]. The main components of the beam line used for irradiation are shown in Figure 3.4a. The samples were mounted on a custom developed sample stage which is a part of irradiation chamber, and includes all other necessary components for monitoring the sample temperature and charge deposited on the sample continuously. The target stage design allows to irradiate several samples simultaneously, in our case up to four samples at the same time.

Irradiation stage The irradiation stage, shown in Figure 3.4b, was designed to accomplish two tasks: securely mount four samples for irradiation studies and provide resistive heating to reach and maintain

106

the irradiation temperature of the samples. Aluminium plates were used to isolate the vacuum flange and other metal components from the high temperatures generated by the heater, located on the Nimonic block underneath the sample, and also by the proton beam itself. A tantalum shield is placed between the ceramic heater and the samples, and serves two purposes: to prevent the protons from damaging the thin layer of PBN (Pyrolytic Boron Nitride) that covers the heating element, and to allow for the charge induced by the protons to be grounded to the vacuum chamber and measured for radiation dose determination. The minimum thickness of tantalum necessary to stop the protons is 43μm, but a thicker shield is preferred for mechanical stability and heat transfer considerations.

Figure 3.4 Platform for the irradiation experiments in the Dalton Cumbrian Facility: (a) Ion accelerator beam line used for the proton irradiation experiments of Ta-W alloys [4]; (b) main components of the target stage [5].

107

Irradiation area determination A slit system of four tantalum plates was located in front of the sample, and allowed to define the desirable irradiated region of the sample, and also to check the uniformity of the irradiation over the desired area, by measuring the charge on the four plates simultaneously and independently (Figure 3.5).

Figure 3.5 Slit system used to optimise the beam position on the sample and to determine the area of the sample to be irradiated uniformly by the ion beam.

Even slight modifications in the current collected on the aperture indicate a drift in the proton beam over time. This can be manually fixed by steering the beam back onto the initial sample position. Figure 3.6 shows schematically the position and main dimensions of the tantalum plates in the slit system, and also the irradiated area in the sample.

Figure 3.6 Schematics of the slit system. The assembly is comprised of four electrically isolated tantalum plates.

108

At the end of the irradiation experiment, a Faraday cup is inserted to stop the irradiation. The system is cooled back to room temperature, the vacuum chamber is vented, the stage is removed from the vacuum chamber, and the samples are removed from the stage. The sample activation was checked after each experiment by measuring the potential alpha, beta and gamma radiation emission using a Berthold dosimeter and a gamma spectrometer.

In-situ proton irradiation at MIAMI system Irradiation facility MIAMI-1 (Microscope and Ion Accelerator for Materials Investigations) located at the University of Huddersfield is composed of a 100 keV electrostatic accelerator coupled with a JEOL JEM-2000FX transmission electron microscope (Figure 3.7) operating at an accelerating voltage of 200 kV [6]. The ion beam enters the microscope at 30° with respect to the electron beam direction, see Figure 3.8. There are several sample holders available: single tilt, double tilt, heating and cooling double tilt holders. This gives an access to a wide range of sample orientations and sample temperatures from -170 to 100°C or RT to 1300°. Beam current at the sample position was measured using current metering rod which can be inserted to the TEM as the sample holder. An entrance aperture is located to be exactly at the TEM sample position and the transmitted ion beam current is measured on a collector plate. The electron beam of the TEM is detected on the current metering rod which is centred using the x and y specimen shift controls of the microscope.

Figure 3.7 JEOL JEM-2000FX TEM placed at the end of the ion beam line of the MIAMI-1 ion irradiation facility at the University of Huddersfield.

109

Figure 3.8 Sketch of ion beam production and transport systems at the MIAMI-1 ion irradiation facility [6].

3.3 Characterisation methods 3.3.1 Optical microscopy Optical microscopy is a standard method to quickly assess surface quality of metallic samples. This tool has been used routinely during the sample preparation prior to ion irradiation. This instrument uses visible light and a set of magnifying lenses to produce a magnified image of the sample. An optical microscope creates a magnified image of specimen with an objective lens (providing different magnifications) and magnifies the image further more with an eyepiece. The resolution of optical microscopy is  0.2 microns.

Figure 3.9 Optical microscope Olympus GX71 [7]

The Olympus BX41M reflected light microscope and Olympus GX71 (Figure 3.9) – advanced research metallurgical microscope was used to obtain images of metallic bulk samples.

110

3.3.2 Electron-matter interaction Before describing the capabilities of electron microscopy, it is useful to explain how the electron-material interaction is used to generate the signal, and how to obtain the desired information from the area of study. The detected signals depend on the electron-atom interaction process as shown in Figure 3.10.

Figure 3.10 The generation of various detectable signals, arising from the interaction of an electron with the atoms in a solid.

If the crystalline sample is thin and hence transparent enough for electrons to pass through, then electrons of the transmitted beam can be direct (i.e. no change in direction) or scattered. If the electron beam is scattered, electrons change their direction due to its interaction with the sample. The elastic scattering of electrons occurs due to the interaction of the electron wave front with the periodic potential of the crystalline atomic structure. The beam of electrons does not lose kinetic energy in this case, but their path is deviated by a given angle. Atomic number influences on how much the electron will be deviated from the original path. Larger atoms with a higher number of electrons have stronger electric fields that exert a stronger influence on the paths of the electron beam. During inelastic scattering the incident electrons may transfer sufficient energy to an electron from the sample, such that electron is displaced from its orbital, ionizing the atom and creating a hole in core electronic states. An outer electron may fall into the hole and release excess energy through the emission of an X-ray. This X-ray energy is equal to the difference between the energy of the two electron states. As such, the X-ray energy is characteristic of this specific transition and so may be used to identify the type of atom from which it

111

originated. If energy of the incident electron is high enough, electron which was collided may escape and come out as a secondary electron. The secondary electrons provide information about morphology and topography on sample. Back-scattered type electrons (BSE) come from deeper layers of the sample, where the incident electrons are reflected or back-scattered out of the sample interaction volume by elastic scattering interactions with the atoms. Atomic number or Z contrast images are based on the relationship between backscattered electron emission characteristics and sample composition: as the size of the atomic nucleus increases, hence, the number of BSE increases and contrast changes; bigger atoms have stronger electric fields that perturb more the beam electron paths. Hence, a "brighter" BSE intensity correlates with higher Z number (heavy elements), and "darker" areas have lower Z (light elements). Once a specimen is being examined, the electron beam generates the following types of signals: secondary electrons, back scatter electrons, Auger electrons, X-rays and photons, each one being emitted from different depths of the sample. A typical interaction volume of the electrons with the sample can be represented by a pear- or a teardrop-like shape (Figure 3.11).

Figure 3.11 Signals emitted from different parts of the electron-sample interaction volume.

112

3.3.3 Scanning electron microscopy (SEM) SEM was broadly used in this project for microstructural and surface characterisation of both non-irradiated samples and samples after irradiation exposure. The main functions of SEM used in this project are described below. The SEMs used for this thesis were (SEM) FEI Quanta 250 field emission gun FEG (E)SEM and FEI Quanta 650 FEG (E)SEM. The main difference between these two microscopes is the size of the motorized stage inside the vacuum chamber (50mm for the Quanta 250 and 150mm for the Quanta 650) and the motorized z-range (25mm and 65mm, respectively). The Quanta 650 FEG is designed with a rather spacious chamber, allowing to analyse relatively large specimens and also to install sample environments for in- situ thermo-mechanical studies. The scanning electron microscope is an instrument which allows analysing extremely small objects at nanometer resolution (3-6 nm) and achieving magnifications of up to 1000 000x providing high resolution imaging in a digital format. The SEM uses an electron beam of relatively low energies of tens of keV to scan the sample surface in a high vacuum chamber. The principle of the electron microscope operation is based on the interaction of the accelerated charged particles (electrons) with the sample and the use of an electromagnet to scan the beam over the region of interest of the sample. The motion of the electrons can be described as a wave process with a wavelength λ = 0.02Å (for a 200keV electron beam). In contrast, the light wavelength in the optical microscope is λ = 0.4 ...0.8µm. The main components of an SEM are the following (see Figure 3.12):  Column for electron beam generation and acceleration;  Specimen chamber;  Vacuum system (high-vacuum mode up to 10-5Pa);  Monitor with control panels.

113

Figure 3.12 Diagram of the main components of a typical SEM.

Calculation of the electron penetration depth into the matter The penetration depth (R) was calculated according to the Kanaya-Okayama formula [8]:

, µm (1) where A is the atomic weight (g/mol), E is the electron beam accelerating voltage (kV), Z is 3 atomic number and ρ is the density of material (g/cm ). We used an accelerating voltage of 20kV. The penetration depth for tantalum was around 1μm.

Imaging Both FEI SEMs were used for low magnification imaging of the scratch-free surface of the samples after the last polishing step, in order to estimate their adequacy for ion irradiation experiments. In the first case, the electron beam interacts with the atoms of the specimen. Those atoms can absorb the energy and give off their own electrons. There is a detector

114

which has a positive charge on it (~300 V) to pick up secondary electrons. Once these electrons come in to the Faraday cage which is positively charged, they reach the detector which uses the information to form the image on a computer screen. This type of electrons is useful to examine the surface quality. The second type of electrons is called back scatter electrons which don’t come from atoms of the sample, but correspond to the incoming electrons that are scattered back from the sample surface and come from a deeper level of the specimen. To detect them a specific back scatter electron detector is being used. The position of both, EDX and EBSD detectors in the SEM is shown in Figure 3.13.

Figure 3.13 Position of the EDX and EBSD detectors in the SEM.

Energy-Dispersive X-ray spectroscopy (EDX or EDS) The EDX or EDS technique is conducted by exciting the sample with a focused electron beam. This excitation causes the emission of X-rays from atoms in the sample. These individual X-rays are detected by an X-ray detector and converted into proportional electrical voltages. In more detail, when the incident electron beam with enough energy hits the sample it ejects an electron from the inner shell close to the atomic nucleus. This atomic state is highly unstable, so an electron from a higher energy level drops down in order to fill the space in the inner electronic shell. In order for this electron from a higher energy level to transfer to a lower energy level, it needs to release a quantum with the energy difference between those electronic states, and a characteristic X-ray is emitted from the sample, see Figure 3.14.

115

Figure 3.14 Example of X-ray emission in sodium atomic nucleus (left); electron shells and main electronic transitions (right).

Due to the fact that the emitted X-rays are characteristic of the atom from which it originated, they can be used to identify elements present in a sample. The amount of energy released is not only dependent on the element that the electron beam hits but also on the energy between the electron shells involved into the process. The adopted nomenclature for the electron shells in the atom begin with the internal shell called K followed by L, M, N, O, P and Q shells (Figure 3.14). The EDS analysis is mostly concerned with the K, L and M shells. If the electrons come to fill an electron-deficient shell from the closest outer shell, then the transition is denoted as an α-signal. However, if the electrons come from the second shell it is known as a β-signal.

Electron Backscattered Diffraction (EBSD) In order to perform EBSD analysis, the primary electron beam is incident on a sample whose surface is tilted 70° with respect to the direction of the incoming electron beam using a working distance of 5-10 mm. EBSD can be used to determine the local crystallographic orientation or to determine the spatial phase distribution. The phosphor screen of the EBSD detector placed several millimetres away from the sample surface will convert the incoming electron into a photon, thus transforming the backscattered signal into a visible light signal. When the diffracted part of the backscattered electrons interacts with the phosphor screen it creates a pattern, i.e. Kikuchi pattern, which depends on the crystal structure and grain orientation of the sample under investigation. Kikuchi patterns form as a result of a two-step process. First, the electrons incident on the sample are inelastically scattered in all directions. Second, many of these scattered electrons will be at the Bragg angle with respect to various

116

atomic planes in the sample, and so these electrons diffract according to Bragg’s law (eq.2). There is a highly sensitive CCD camera located behind the phosphor screen. This camera captures the Kikuchi pattern and transfers it to a computer for a detailed data analysis. A sketch of the sample/detector geometry and the generated EBSD signal from the sample is presented in Figure 3.15.

Figure 3.15 Schematic setup of an EBSD detector and sample orientation in the SEM.

3.3.4 Transmission electron microscopy In the transmission electron microscope a beam of electrons is transmitted through a relatively thin area of a sample, with a maximum thickness not exceeding a few hundreds of nanometres. The TEM image contrast of crystalline specimens has three main origins:  Mass-thickness contrast – absorption or Z contrast, which stems from the proportionality of the scattering power with the atomic number Z;  Phase contrast, stemming from the interference between the transmitted and the diffracted beams. Phase-contrast images are formed by removing the objective aperture entirely or by using a very large objective aperture. This helps to make sure that not only the transmitted beam, but also the diffracted ones are allowed to contribute to the image;  Diffraction contrast means a change of intensity in the image that is formed when the diffraction condition is changed with different areas of the specimen. In the bright-field image (formed by the transmitted wave), the area where diffraction takes place loses its image intensity, hence, becomes dark. In the dark-field image (formed by the diffracted wave), the corresponding area gains image intensity, and becomes brighter. Therefore, images are being observed with either the transmitted or diffracted beam. The interaction of the electron beam with the material gives rise to a contrast that depends on the type of crystal lattice, defect structure and phases present in the material and the applied imaging conditions. In this work post mortem characterization of the radiation induced microstructure was performed using a (TEM) FEI Tecnai G2 20. A schematic view of the

117

microscope and the main components are shown in Figure 3.16. The microscope, equipped with a LaB6 electron emission gun, is operated at an accelerating voltage of 200 kV.

Figure 3.16 Schematic representation of the main components of TEM (left); TEM FEI Tecnai G2 20 with its main components labelled (right).

When high-velocity electrons are being transmitted through the sample, they interact with the regular atomic arrangement of the atoms within the sample. They are scattered coherently and since the crystalline sample has a regular distribution of the atoms, electrons scatter in a ‘coordinated manner’. That leads to electron diffraction, which allows us to study the crystal structure and lattice defects of the sample. The electrons diffract when interacting with the atomic planes of the crystalline sample, according to Bragg’s law:

nλ = 2d sinθ (2) where n is an integer (order of the reflection), λ is the wavelength of the incident electrons, d is the distance between the atomic planes and θ is the scattering angle between the incident beam and the scattering planes. Those angles particularly provide information about the interplanar spacing and the atomic arrangements within the crystal. For every hkl family of

118

crystal planes satisfying Bragg’s law, there will be a different scattered beam, inclined with respect to the direct beam at an angle two times the Bragg angle of that set of planes with spacing dhkl:

(3) √ for a cubic system, where a is the lattice parameter. Electrons coming from the condenser system of the TEM, may be scattered by the sample, which is placed in the focal plane of the objective lens. The electrons scattered in the same direction are focused in the back focal plane, and, as a result, a diffraction pattern is formed. Electrons coming from the same point of the object are focused in the image plane. In the TEM, the first intermediate image is magnified by further lenses (projective system). Diffraction contrast imaging is being formed by obtaining contrast from particular regions within the samples that are diffracting differently than the bulk of the sample. In the case of a dislocation there is a local strain field, so that the diffracted signal coming of the area close to the dislocation is different from the bulk sample. Complementary to standard electron diffraction is convergent beam electron diffraction, termed CBED. In this case instead of collecting the diffraction signal from a broad area of the sample, electrons are being focused down onto a small sample area ≤ 50nm [9]. The CBED technique helps to obtain additional crystallographic information about the sample. For example, the symmetry of the CBED pattern can be used to describe the space group of the crystal structure, to set two-beam condition to examine structural defects, and also to measure the foil thickness [10], which is described in detail in section 3.3.4.4. 3.3.4.1 Diffraction contrast mechanisms For each hkl family of planes complying with Bragg’s law, there will be a different scattered beam, tilted with respect to the direct beam at an angle of two times the Bragg angle of that set of planes with spacing dhkl. The construction of the Ewald sphere represented in Figure 3.17a helps to understand the relationship between the orientation of a crystal and the direction of the diffracted beam coming out of the crystal. In diffraction, the reciprocal lattice vector g, with its magnitude g =1/d , is referred as the diffraction vector, connecting the transmitted beam (k1) with diffracted spots (kD), as shown in Figure 3.17 and as represented by the formula:

k = kD – k1 = ghkl + sg (4)

119

The diffraction condition is often characterized by the excitation error, sg, which is the measure of deviation from the Bragg condition, see Figure 3.17b. At the Bragg angle s = 0 diffraction occurs. However, diffraction occurs when s ≠ 0 but at lower intensity,

Figure 3.17 (a) The Ewald sphere is shown intersecting an array of reciprocal-lattice point; (b) Schematic representation of the Ewald sphere intersecting the row of systematic reﬂections at point ng in the reciprocal space, where n is not necessarily an integer. g is the diffraction vector and sg is the deviation parameter from the exact Bragg conditions, sg=0 for the dynamical two-beam imaging condition. sg is parallel to the foil normal. The radius of the

Ewald sphere kI is drawn signiﬁcantly smaller than it should be for better representation. In reality k>>g.

As seen in Figure 3.17a, the origin of k1 coincides with the centre of the Ewald Sphere with radius = 1/λ. Any hkl point of the reciprocal lattice which is intersecting the sphere surface can be linked from the centre of it by the k of the diffracted beam (kD). Diffraction occurs only if a reciprocal lattice point lies on the surface of the Ewald sphere. Electrons are coming parallel to the optic axis of the microscope which is in the middle of the viewing screen, and there are particular directions to which the electrons are scattered constructively depending on the sample structure and orientation. The electrons coming of the planes parallel to each other and the objective lens brings them into focus in the back focal plane to a single point. In this case a bright field image (BF) is formed, see Figure 3.18a. If the objective aperture is removed of the direct beam to choose scattered electrons (e.g. a specific diffracted beam), a dark field image (DF) is formed, see Figure 3.18b. If the sample is bombarded by the incident beam at an angle equal and opposite to the scattering angle (by tilting the incident beam), then by this way the scattered electrons will now come down the optic axis of the microscope, as shown in Figure 3.18c.

120

Figure 3.18 Formation of (a) a bright field, (b) displaced aperture dark field and (c) centred aperture dark field image [11].

In order to obtain the diffraction pattern only from a selected area of the microstructure, for example a fine precipitate, an aperture that filters the transmitted beam is used, and the diffraction pattern is called SADP (Selected-Area Diffraction Pattern). By inserting SAD aperture in the image plane of the objective lens, a virtual aperture is created in the plane of the sample (shown in Figure 3.19). Only electrons falling inside the dimensions of the virtual aperture at the entrance surface of the specimen will be allowed through into the imaging system. All other electrons will strike against the SAD diaphragm. The available apertures on the Tecnai 20 TEM have diameters of the order of micrometers; in case the selected area has to be smaller, of the order of nanometers, the nano-diffraction technique can be utilized.

121

Figure 3.19 Selected are diffraction pattern (SADP) formation [11].

Additional information about the crystal structure can also be obtained from high resolution atomic images by performing a fast Fourier transformation (FFT). This mathematical technique recognizes in the image the presence of periodicities of the intensity signal due to the ordered disposition of the atoms. The FFT translates these periodicities into a diffraction pattern. A high resolution TEM image at the <111> zone axis is given in Figure 3.20a with the FFT of that image on Figure 3.20b, showing the cubic arrangement of the Ta atoms when observed along the <111> matrix orientation. a b

Figure 3.20 (a) HR STEM image along the <111> zone axis in Tantalum, (b) the FFT of that image shows the cubic symmetry along the <111> zone axis.

122

3.3.4.2 Contrast of dislocation loops and voids The distortion of the crystal lattice of a material due to the presence of dislocations or dislocation loops is at the origin of the diffraction contrast. In a sample oriented to fulfil the kinematical two-beam condition, where only one set of lattice planes is close to the Bragg diffraction condition, some of the planes are distorted by the local strain field of a dislocation towards the exact Bragg condition. This implies that locally the intensity of the diffracted beam is higher than away from the defect, thus detecting more intensity from the transmitted beam. As the bright field image is formed only with the transmitted electron beam and the diffracted beams are being cut off by the objective aperture, there is a loss of image intensity due to the presence of the defects (i.e. dislocations). Therefore by sequential choosing different orientations or zone axes we can study different lattice defects present in a given family of planes. As shown in Figure 3.21 image, tilted to different zone axis, shows that some dislocations will not be seen at particular reflections.

Figure 3.21 TEM analysis of the as-received Ta-10W material showing the different contrast of dislocations along different zone axis.

In the case of three-dimensional voids or cavities, the TEM technique does not utilize diffraction contrast but uses through-focal images of them, which implies a series of void images taken under- or over-focus. An illustrative example of the through focal imaging of radiation-induced voids is represented on Figure 3.22. The total amount of defocus should

123

not exceed 1 μm in case of voids with a diameter of 1nm. Image defocusing causes Fresnel fringes due to the ‘drastic’ change of the refractive index between the void and the surrounding matrix. The phase-contrast component arises from the change in average inner potential between the void and the matrix, which causes a phase shift between electrons which go through the void and those which pass through the surrounding crystal. Thus, small voids appear as bright discs encircled by a dark fringe in under-focused condition, and as dark discs with white fringe around under over-focused condition. Voids of diameter ≥1 nm can be observed applying this approach.

Under-focused condition Over-focused condition a b

Fig. 3.22 Bright field images of voids in tantalum proton-irradiated at 350oC and at 1.55 dpa, taken in TEM using the out-of-focus imaging technique [5].

3.3.4.3 Burgers vector analysis For the standard Burgers vector b determination, the Kikuchi maps are commonly used to facilitate systematic tilting from one zone axis to another. The Burgers vector b of dislocation lines or dislocation loops characterises the magnitude and direction of the local lattice deformation induced by these defects. The Burgers vector can be determined using TEM, taking into account that dislocations are invisible when b and the diffraction vector g are mutually perpendicular, that is when b lies in the diffracting plane of the crystal. Defining b is important in the description and understanding of the microstructure evolution during irradiation or plastic deformation. True invisibility is achieved when both g·b = 0 and g·b×u = 0 are fulfilled, where u is specified in terms of direct lattice vectors [uvw]. Since the dislocations are invisible when the dot product g·b = 0, tilting the sample in the TEM using a special double-tilt sample holder to several different diffraction conditions, i.e. different zones axis and g vectors, and acquiring images of the studied crystal defects helps to

124

determine their Burgers vector. An illustrative example of such image series used for dislocation analysis of a tantalum sample is presented in Figure 3.23. This method is widely applied in the investigation of dislocations and dislocation loops [12].

Figure 3.23 Bright-field TEM imaging of interstitial-type dislocation loops in tantalum proton irradiated at: top: 180(2)°C; bottom – 590(5)°C, shown for different two-beam conditions. These TEM data was used for the dislocation analysis using the g.b=0 invisibility criterion, where g denotes the scattering vector and b the Burgers vector, and revealed the a/2<111> (top – 180(2)C) and a<001> (bottom – 590(5)C) Burgers vector [5].

The nature of dislocation loops can be of interstitial type if they are represented in the lattice as an inserted atomic plane, or of vacancy type if the atomic plane is missing as explained in section 2.2.2. The nature of the radiation-induced dislocation loops was determined in this work using the outside-inside contrast method, where the dark line of the loop can appear inside the loop, or outside. This method is based on theory that under kinematical diffraction conditions contrast of the dislocation loop lies inside or outside the actual loop positions. This depends on loop orientation in respect to the electron beam, diffraction vector g, excitation error s sign (+/-). Inside contrast arises when (g · b) s < 0 and outside when (g · b) s>0. For example, if we have determined b=a/2<111>, and we set the electron beam at particular two-beam condition, ̅ ̅, as shown in Figure 3.24 with s to be positive, therefore the Kikuchi line is on the opposite side of the diffraction spot to the 000 direct beam. The majority of dislocation loops show outside contrast at this reflection.

125

Therefore, ̅ ̅, b=+/-[111], s>0, thus the inclination of the loop must be negative, so b= –[111]. The zone axis z in this case was [011]. When b·z >0 the loop is of vacancy type, when b·z <0 is of interstitial type [13]. g=-21-1

Figure 3.24 TEM data of dislocation loops showing outside contrast in a proton-irradiated Ta-10W sample.

3.3.4.4 Convergent beam electron diffraction technique (CBED) The experimental determination of the foil thickness in this work was conducted in the TEM. These measurements are of paramount importance to estimate the volume density of the crystal defects (such as dislocations) under study. One of the commonly used methods for the foil thickness determination is based on the intensity distribution recorded in the CBED patterns. In CBED patterns, Bragg spots appear as discs whose diameters depend upon the convergent angle 2α (Figure 3.25).

126

Figure 3.25 The formation of selected-area electron diffraction patterns (left) and convergent-beam electron diffraction (CBED) patterns (right) [9].

Thickness measurements are carried out when the sample is tilted to the two-beam condition, with only one hkl-reflection, in order to detect the transmitted and diffracted beam. The deviation parameter from the Bragg condition can be determined according to the formula:

Si =λ (5)

where ∆Qi- the angle for a fringe spacing i, (an integer: 1, 2, 3); λ – the wavelength of incident electrons;

ϴB – the Bragg angle for the diffracting (hkl) plane; d – the (hkl) interplanar spacing.

127

Figure 3.26 Intensity profile (Ig) as a function of s (deviation parameter) and 2ϴB (the separation of the (000) and (hkl) discs) (left); CBED image of 000 and 200 reflections with the parallel Kossel-Mollenstedt fringes in tantalum (right).

From the position of the Kossel-Möllenstedt fringes, the thickness t can be measured together with ξg which is an extinction distance. When electrons go through a crystalline specimen causing one Bragg reflection (two-beam condition), and an incident wave reaches a certain depth (t), its amplitude becomes zero and the reflected wave becomes maximum. If the incident wave reaches a depth twice as deep as the previous depth (2t), the amplitude of the reflected wave becomes zero again and the amplitude of the incident wave becomes maximum. Hence, the amplitudes of the incident and reflected waves exhibit beats with the depth. The distance of one periodicity of the beats is called "extinction distance." The extinction distance is a function of the deviation from the Bragg angle. According to Figure 3.26 the central bright fringe corresponds to the Bragg condition. The fringe-spacing corresponds to angles Δϴi. The variation in si causes an intensity oscillation across the disc. The foil thickness t can be derived using eq.(6) if the extinction distance ξg is known for the studied material using particular reflection at an electron accelerating voltage of 200 kV, based on the expression:

(6)

128

st where, nk – an integer related to the fringe’s index number, starting with nk=1 for the 1 fringe, so that the thickness t can be calculated. If the extinction distance is unknown, it is 2 2 possible to use a graphical method based on plotting (si/nk) vs (1/nk) according to eq.(5). The thickness of the specimen is given by the intercept of the resulting straight line with the 2 (si/nk) – axis, see Error! Reference source not found. 3.27. The relative error in thickness measurements by the graphical method is estimated to be ±10% [14].

2 2 Figure 3.27 Graphical representation of (si/nk) vs (1/nk) to obtain the foil thickness [10]

3.3.5 Hardness measurements It is of paramount importance in this work to link the changes in hardness with microstructural parameters measurable via electron microscopy, so that the mechanisms and consequences of radiation-induced damage in the studied materials can be elucidated. In order to achieve this goal, the hardness of the bulk materials after proton irradiation was assessed by nanoindentation using an MTS Nanoindenter equipment (Figure 3.28) fitted with an XP Berkovich type tip. The obtained hardness values at different damage levels along the sample cross section were compared to the hardness measured in the case of non-irradiated samples of the same batch. Hardness tests differ in the indenter shape used, load applied, methodology to calculate the hardness value, time of loading, dimensions of the sample or the irradiated layer thickness. Depending on these factors, the hardness may characterize the elastic or elastic–plastic properties, resistance to high and low deformations and to a rupture. A commonality of the

129

hardness methods is the application of a load during the contact of the indenter and the tested material. Ideally, the indenter has to be not deformable, and is normally made of diamond. The constant depth method was chosen to obtain the nanohardness data [15]. The hardness of a given material is the property to resist the penetration of another more rigid body (indenter) with no residual deformation. The Berkovich tip represents a 3-sided truncated pyramid that is suitable for analysis of the samples produced in this project work. Figure 3.29 shows the schematic of indenter-surface contact during and after stress applied.

Figure 3.28 Nanoindenter system from NanoIndenter XP System, MTS (left); schematic of the indentation and scratch instrument (right).

The hardness value (H) can be determined using the equation [16]:

H=Pmax/A (5) where A (mm2) represents the projected pyramidal area of indentation at the maximum load

Pmax (kgf).

In this work, nanoindentation tests have been performed over the cross-sectional area of each sample from the top of irradiated layer down, at an angle of 5o with respect to the top surface in order to reach 50-55 µm from the irradiated surface. The aim was to observe radiation- induced changes in hardness as a function of damage level, which could be related to an increasing density of crystal defects in the irradiated microstructure. These measurements also allowed us to determine the position of the Bragg peak which is related to the stopping of the proton beam in the sample during irradiation experiments, and to compare the results with the estimation of the beam penetration depth in the sample from SRIM calculations. The indentation depth was chosen to be 200 nm with a number of 100-120 indentations.

130

Figure 3.29 Scheme of the indenter – surface contact: under force F , where α is the angle between indenter surface and sample surface, Ac is the projected area of contact, radius a, which occurs at a distance hc from the indenter tip, hmax – the maximum indentation depth and hs is the bowing of the surface (top); After force removal where h0 is the residue indentation depth [17].

131

3.4 References 1. J. Wuk Park, K.Y.S., D. Kwan Chung, C. Nam Chu, Fabrication of micro-lenticular patterns using WEDM-grooving and electrolytic polishing. Journal of Micromechanics and Microengineering, 2013. 23: p. 125034. 2. E521-96, A., Standard Practice for Neutron Radiation Damage Simulation by Charged-particle Irradiation. 2009. 3. Z. Yao, S.Xu, M.L. Jenkins, M.A. Kirk, Preparation of TEM samples of ferritic alloys. Journal of Electron Microscopy, 2008. 57(3): p. 91-94. 4. P.T., Wady, et al. Accelerated radiation damage test facility using a 5 MV tandem ion accelerator. 2016. A 806, 109-116. 5. I. Ipatova , P.T.Wady, S.M. Shubeita , C. Barcellini , A. Impagnatiello , E. Jimenez- Melero, Radiation-induced void formation and ordering in Ta-W alloys. Journal of Nuclear Materials, 2017. in press. 6. J.A. Hinks, J.A.v.d.B., S.E. Donnelly, MIAMI: Microscope and ion accelerator for materials investigations. Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 2011. 29: p. 21003. 7. https://www.olympus-ims.com/en/microscope/gx71/. Olympus. [cited 2017 07.06]. 8. K. Kanaya, S.Okayama, Penetration and energy-loss theory of electrons in solid targets. Journal of Physics D: Applied Physics, 1972. 5(1): p. 43-58. 9. P.E. Champness, , Convergent beam electron diffraction. Mineralogical Magazine, 1987. 51: p. 33-48. 10. Delille, D., R. Pantel, and E. Van Cappellen, Crystal thickness and extinction distance determination using energy filtered CBED pattern intensity measurement and dynamical diffraction theory fitting. Ultramicroscopy, 2001. 87(1-2): p. 5-18. 11. D.B. Williams, C.B.Carter, Transmission Electron Microscopy. 2009, Springer. 12. A.Howie, M.J. Whelan, Diffraction contrast of electronmicroscope images of crystal lattice defects. iii. Results and experimental confirmation of the dynamical theory of dislocation image contrast Proceedings of the Royal Society A, 1962. 267: p. 206. 13. M.L. Jenkins, M.A.Kirk, Characterization of Radiation Damage by Transmission Electron Microscopy. 2001: IOP Publishing Ltd. 14. A. Harte, D.Jadernas, M. Topping, P. Frankel, C.P. Race, J. Romero, L. Hallstadius, E.C. Darby, M. Preuss, The effect of matrix chemistry on dislocation evolution in an irradiated Zr alloy. Acta Materialia, 2017. 130: p. 69-82.

132

15. J.T. M. De Hosson, C.A.Brebia, Surface Effects and Contact Mechanics XI: Computational Methods and Experiments. 2013: WIT Press. 16. R.A. Potyrailo, E.J. Amis, High-Throughput Analysis: A Tool for Combinatorial Materials Science. 2003: Springer. 17. T. Chudoba, N.M. Jennett, Higher accuracy analysis of instrumented indentation data obtained with pointed indenters. Journal of Physics D Applied Physics, 2008. 41: p. 215407.

133

4. Structural defect accumulation in tungsten and tungsten-5wt.% tantalum under incremental proton damage

I. Ipatovaa,b,*, R.W. Harrisonc, P.T. Wadyb, S.M. Shubeitab, D. Terentyevd, S.E. Donnellyc, E. Jimenez-Meleroa

a School of Materials, The University of Manchester, Manchester M13 9PL, UK bDalton Cumbrian Facility, The University of Manchester, Moor Row CA24 3HA, UK cSchool of Computing and Engineering, University of Huddersfield, Huddersfield, HD1 3DH, UK dSCK·CEN, Nuclear Materials Science Institute, Boeretang 200, Mol, 2400, Belgium

Published in the Journal of Nuclear Materials

DOI 10.1016/j.jnucmat.2017.11.030

Corresponding author (*): University of Manchester School of Materials Oxford Road Manchester M13 9PL United Kingdom Tel.: +44 7849290480 Email: [email protected]

134

Abstract We have performed proton irradiation of W and W-5wt.%Ta materials at 350 C with a step-wise damage level increase up to 0.7 dpa and using two beam energies, namely 40 keV and 3 MeV, in order to probe the accumulation of radiation-induced lattice damage in these materials. Interstitial-type a/2 <111> dislocation loops are formed under irradiation, and their size increases in W-5Ta up to a loop width of 21±4 nm at 0.3 dpa, where loop saturation takes place. In contrast, the loop length in W increases progressively up to 183±50 nm at 0.7 dpa, whereas the loop width remains relatively constant at 29±7 nm at >0.3 dpa, giving rise to dislocation strings. The dislocation loops and tangles are observed in both materials examined after a 3 MeV proton irradiation at 350 C. Ta doping delays the evolution of radiation-induced dislocation structures in W, and can consequently impact the hydrogen isotope retention under plasma exposure. Keywords: refractory metals, binary alloys, dislocation analysis, proton irradiation, transmission electron microscopy

Introduction Tungsten (W) and its alloys are leading material candidates for plasma-facing components of magnetically-confined fusion reactors, especially in high-heat-flux regions such as the divertor [1-3]. The attractiveness of W-based materials lies in their high resistance to plasma- induced sputtering, erosion and neutron-induced swelling, their thermal conductivity and high-temperature strength [4-7]. These materials will be exposed during service to high-heat loads between 0.1 and 20 MW/m2 [8, 9]. During plasma disruptions and edge-localised mode events, heat loads can even reach up to the GW/m2 range [10]. Unfortunately, the ductile-to- brittle transition temperature (DBTT) of tungsten is relatively high, namely 200–400 °C depending on its processing route and resultant grain structure [11, 12]. The DBTT is also reported to increase with neutron irradiation up to 800–1000 °C [13, 14]. The brittle nature of tungsten can lead to premature failure during low-temperature reactor operations [15]. Moreover, the neutron-induced sequential transmutation of tungsten into rhenium and later in osmium is predicted to place the alloy composition close to the 100% phase field of brittle  precipitates [16-18]. Therefore, it is of paramount importance to design the alloy composition and microstructure so as to enhance its ductility, and therefore to extend the temperature window of W-based alloys for safe reactor operation. One promising route to increase ductility is to alloy tungsten with tantalum (Ta), since the latter also improves resistance to water corrosion [19] and its neutronic performance is

135 relatively similar to that of tungsten [6, 16, 20, 21]. Ta also delays the compositional shift into the  phase field [16]. However, the existing knowledge about the behaviour of W-Ta alloys in fusion-relevant radiation environments remains limited. Therefore, it is not yet possible to predict reliably the lifetime performance of the W-Ta material and its potential failure in reactor operation conditions. Thermal desorption spectroscopy studies on W and W-Ta alloys exposed to deuterium plasmas at the plasma generator Pilot-PSI indicated that alloying W with Ta does not either introduce additional kinds of traps for deuterium or significantly influence the concentration of trapping sites [22]. Additional experimental evidence from literature [23-25] revealed that W-Ta alloys have a higher deuterium retention than W at low flux (1020m-2 s-1), whereas W features higher retention than W–Ta at high exposure flux (1024m-2 s-1). Very recently dual beam and sequential irradiation experiments have shown the suppression of the fuzz-like damage surface morphology in W and W-Ta alloys at 950C as the ratio of D+/He+ ions increases, probably due to a significant D detrapping at that temperature [26]. Past results about W doped with 5 wt.% of Ta and subjected to 2 MeV W+ ion irradiation at 300 oC revealed a radiation-induced hardness increase at relatively low damage levels, e.g. from 7.3±0.2 GPa in the unimplanted condition to 8.8±0.2 GPa at 0.07 dpa, attaining a saturation in hardness at 13 dpa with a value of 10.9±0.3 GPa [27]. An extensive pile-up around the indentation was observed in the unimplanted condition, whereas the extent and height of pile-up were reduced already at damage levels of 0.07 and 1.2 dpa. The authors concluded that the very first onset of plasticity occurred at a lower stress in irradiated material. The mechanism of this change in pile-up behaviour was unfortunately unknown at that time. A linear correlation between the Vickers microhardness and yield strength has been proposed for neutron-irradiated W [28]. Moreover, an increase in the yield stress and maximum tensile strength due to neutron irradiation has been reported in W, coupled with a loss in ductility and an increase in the ductile-to-brittle transition temperature [29]. Microstructural analysis showed that the dislocation loop number density in W-Ta alloys is greater than in W at the same damage level, but the average loop size is reduced [30]. Furthermore, implantation of 3000 appm helium in W-Ta alloys at 300 °C induces a hardening effect larger than in the case of self-ion implantation [31]. In addition, a rapid increase in hardness was observed after 800 MeV proton irradiation of tungsten up to a damage level of 0.8 dpa, followed by a lower rate of hardness increase at higher damage levels up to 23 dpa [32]. Despite these results, the mechanistic understanding of the structural

136 damage formation and evolution in W-Ta alloys leading to the observed hardness increase is very fragmented and requires additional experimental evidence. A suitable approach to monitor the formation and evolution of radiation-induced lattice defects in metallic materials is to perform in-situ experiments using a transmission electron microscope (TEM) coupled to an ion accelerator [33, 34]. However, the acceleration voltage in the electron gun of the microscope is normally not high enough to study regions in the TEM foil thicker than 50-200 nm. This implies that potential lattice defects, such as dislocation loops induced by radiation, could migrate to the sample surface, due to image forces, that would act as an effective defect sink. The consequence is a potential underestimation of the produced defect density, and/or changes in the morphology and predominant nature of the lattice defects [35, 36]. Radiation-induced dislocation loops were reported to migrate to the sample free surface in both single crystal and polycrystalline tungsten material, during an in-situ TEM experiment using proton energies between 0.5-8 keV in the temperature range of 20–800 °C [37]. This study is devoted to the in-situ observation of radiation-induced lattice defects in W-5Ta alloy at 350oC, and how they evolve with an increasing damage level at that temperature. The potential impact of the free surface in the in-situ proton-irradiated samples was assessed by performing equivalent irradiations using a higher proton beam energy and analysing the lattice damage ex-situ using TEM. Additional irradiation experiments using equivalent conditions were performed on tungsten, and the results used as a base line to understand the effect of tantalum on the evolution of the structural damage induced by proton irradiation.

Experimental The W-5wt.%Ta (W-5Ta) alloy was produced by powder metallurgy and provided by Plansee AG. The W-5Ta material was double forged and then annealed at 1600 C for 1 hour [38]. After delivery, the material was annealed for 1 h at 1000C for degassing [24]. Afterwards, small samples were machined and annealed at 1400 C for 2 hours to remove the defects from machining.The starting W material was provided by Goodfellow Cambridge Ltd. in the form 1mm-thick sheet. The as-received W material was initially annealed in vacuum at 1400 °C for 2 hours for recrystallization. The average grain size, as derived from Electron Backscattered Diffraction maps shown in Fig. 4.1, was 2.3±0.7 m (W-5Ta) and 3.9±0.8 m (W), respectively.

137

Figure 4.1 EBSD map of the W and W-5Ta microstructures prior to proton irradiation.

3mm-diameter TEM discs of both materials were prepared by mechanical pre-thinning, followed by electropolishing at a temperature of ~ -5 C using a Struers Tenupol-5 unit and an electrolyte comprising an aqueous solution of 0.5 wt.% Na2S for electropolishing W, or a mixture of 15 vol.% sulphuric acid (95%) and 85 vol.% methanol in the case of W-5Ta. In- situ proton irradiation experiments were performed at the Microscope and Ion Accelerator for Materials Investigations (MIAMI-1 facility) located at the University of Huddersfield [39]. The TEM discs were placed in a high-temperature Gatan 652 double-tilt specimen holder, and then mounted in a JEOL JEM-2000FX TEM operating at an accelerating voltage of 200 kV. The TEM is installed at the end of a beam line connected to an electrostatic accelerator that produces ion beams with energies up to 100 keV. The ion beam enters the microscope at 30o with respect to the electron beam direction. We performed step-wise sample irradiations using a 40 keV proton beam at a temperature of 350±2 °C. At the end of the experiment, we achieved a proton fluence of 9.5×1017 ions/cm2 for W and 9.7×1017 ions/cm2 for W-5Ta. Additionally, we performed an ex-situ proton irradiation experiment, attaining a damage level of 2 dpa at the Bragg peak position. For this experiment, we mounted both W and W-5Ta samples simultaneously. The samples were irradiated using a 3 MeV proton beam produced by a 5 MV tandem ion accelerator installed at the University of Manchester [40]. The two samples were kept at a temperature of 350±4 °C. Thermocouples were spot welded on the samples outside the irradiated area. Liquid indium was used to improve the thermal contact between the samples and the NIMONIC75® alloy block of the sample stage. We mapped the

138 sample temperature during heating/ion irradiation using an IR camera located at 30 with respect to the sample surface. The IR camera measures the black body temperature, which is related to the temperature of the sample by its thermal emissivity. Before switching on the beam, the sample is heated up to the relevant temperature in order to obtain the sample emissivity as a function of temperature. This was done by normalising the temperature measured by the IR camera without the beam to the value obtained from the thermocouples spot welded on the sample surface [40]. The temperature distribution measured in the two samples by the IR camera during irradiation was characterised by an average value of 350±4 C. A summary of the irradiation conditions of the experiments performed at both irradiation facilities is shown in Table 4.1.

Table 4.1 Main parameters used during the proton irradiation of W and W-Ta samples. The damage level of the in-situ samples corresponds to the average value over the disc thickness of 100 nm studied by TEM (see inset of Fig. 4.2). In this case, the samples were studied in- situ by TEM at selected incremental damage levels up to 0.7 dpa. The fluence given in the table refers to the total value achieved at the end of the in-situ irradiation experiment, corresponding to the damage level of 0.7 dpa. In contrast, the damage levels mentioned for the ex-situ irradiation corresponds to the value from where the foils where extracted for TEM analysis, using the samples irradiated to a fluence of 1.2×1019 protons/cm2. The damage is spread along a larger depth in the ex-situ 3 MeV irradiation, as compared to the higher slope seen in the damage profile from the in situ 40 keV irradiation (see Fig. 4.2).

In-situ Ex-situ

Material W W-5Ta W W-5Ta

Temperature (°C) 350±2 350±4

Proton energy 40 keV 3 MeV

Current (nA) 0.5 0.45 12 000

2 17 17 19 Fluence (ions/cm ) 9.5×10 9.7×10 1.2×10

2 14 14 17 Flux (ions/cm /s) 2.4×10 2.2×10 4.8×10 -4 -6 -6 Damage rate (dpa/s) ~1 ×10 ~3.5×10 ~4.6×10

Damage level (dpa) 0.7 0.3 0.4

139

We prepared TEM discs from the samples irradiated ex-situ using the electropolishing conditions described above. The damaged structures were characterised using an FEI Tecnai

T20 Transmission Electron microscope with LaB6 crystal, equipped with a double-tilt specimen holder and operating at 200 kV. The determination of the Burgers vector b of the observed dislocations made use of the g.b=0 invisibility criterion, where g denotes the scattering vector. The foil thickness was derived from the fringes spacing of the convergent beam electron diffraction pattern (CBED), and was used to obtain the dislocation density at each selected radiation dose [41]. The foil thickness was determined to be in the range of ~100-110nm. The error in thickness measurements using CBED patterns from the exact Bragg condition is assumed to be ±10% [42].The interstitial or vacancy nature of the dislocation loops was assessed by applying the inside-outside contrast method [41]. Furthermore, the damage profile was simulated using the SRIM software with the quick Kinchin–Pease approach [43,44]. We used a value for the displacement energy of 90 eV and default values for other software settings [45], together with the total charge deposited on the sample during the irradiation experiment. The simulated damage profiles for both in-situ and ex-situ experiments are displayed in Fig. 4.2, and the penetration depth from which TEM foils were prepared for ex-situ analysis is also indicated in this figure. Once implanted, the protons progressively lose energy by interaction with atoms in the target material. At high ion energies (>1 MeV/u), the stopping power is provided by the relativistic Bethe theory and the energy loss is due to conduction electrons and to the not quite free d electrons. In contrast, at low energies the loss is only due to free (s, p) conduction electrons, and the shell, Barkas– Andersen and Bloch corrections are introduced to the Bethe-Bloch formula [46, 47]. At low ion energies, the particle may capture electrons from the target and partially neutralize its nuclear charge. However, the Bethe–Bloch equation requires a constant particle charge [43]. In fact, H+ dominates at high energies in metals, whereas the negative ion H- presents the highest equilibrium charge state fraction at low energies. In the intermediate energy regime, the three hydrogen species, namely H+, H and H- are expected to co-exist [48]. A change of the slope in the energy loss versus ion velocity curve occurs at a characteristic threshold velocity for a given target material, with the higher slope being characteristic of higher ion energies [49].

140

Figure 4.2 Simulated damage profile using the SRIM software with the quick Kinchin – Pease approach and the total current deposited on the sample during the ex-situ or in-situ (inset) proton irradiation experiment. The asterisks indicate the regions of the sample from where TEM foils were extracted for ex-situ analysis. The damage level for the sample irradiated in-situ with 40 keV protons was estimated as the average value in the foil thickness of 100 nm. According to SRIM calculations, 97% of the incoming protons were transmitted through the foil. The Bragg peak for 40 keV protons in tungsten would be at 130 nm.

Results The evolution of the structural damage in W-5Ta and W up to a damage level of 0.7 dpa at a temperature of 350±2 °C, as observed by in-situ TEM during proton irradiation, is shown in Fig. 4.3 and 4.4, respectively.

141

Figure 4.3. In-situ observation of dislocation loop formation and evolution in W-5Ta at 350 C, under increasing damage levels induced by a 40 keV proton beam, close to the [111] zone axis. The ex-situ data corresponds to a TEM foil extracted from the W-5Ta sample irradiated with 3MeV protons to a damage level of 2 dpa at the Bragg peak position (see text and Fig. 4.2).

Fig. 4.4 In-situ observation of dislocation loop formation and evolution in W at 350 C, under increasing damage levels induced by a 40 keV proton beam, close to the [111] zone axis. The ex-situ data corresponds to a TEM foil extracted from the W sample irradiated with 3MeV protons to a damage level of 2 dpa at the Bragg peak position (see text and Fig. 4.2). Dislocation tangles are visible in the microstructure irradiated ex-situ.

In both materials dislocation loops are already visible at a damage level of 0.1 dpa, homogeneously distributed in the matrix, and their size appears to increase with radiation

142 damage. The dependence of the average loop size and number density with radiation damage level for both materials is shown in Fig. 4.5.

Figure 4.5 Average dislocation loop size and number density in W and W-5Ta as a function of damage level, as derived from the in-situ irradiation data.

The average loop length and width in both W-5Ta and W increases progressively up to a damage level of 0.3 dpa. At higher damage levels, the average loop width in both materials reaches saturation, with a somewhat higher value for W (29±7 nm) than for W-5Ta (21±4 nm) at 0.7 dpa. Moreover, the loop length in W-Ta also saturates beyond 0.3 dpa, at a value of 33±10 nm. However, the loop length in the case of W continues to increase, and does not reach saturation at 0.7 dpa, where its loop length takes a value of 183±50 nm. The increase in loop length in W with damage level occurs simultaneously with a continuous decrease in the loop number density. This deviates from the behaviour observed in W-5Ta, where the number density of the loops also seems to gradually level off and saturate at damage levels higher than 0.3 dpa. At the highest damage level of 0.7 dpa in this study, the

143 number density takes a value of 2.1 ×1021 m-3 (W-5Ta) and 0.8 ×1021 m-3 (W), respectively. Post-mortem TEM analysis of both in-situ irradiated materials up to 0.7 dpa showed that the loops are of interstitial nature and with Burgers vector of a/2 <111>. Bright field TEM images of W-5Ta and W samples irradiated ex-situ at selected damage levels are added for comparison in Fig. 4.3 and 4.4 respectively. TEM foils for ex-situ analysis were taken from regions away from the Bragg peak position, where the majority of hydrogen atoms are stopped and trapped by the high density of lattice defects. Ex-situ TEM samples show the presence of dislocation tangles, together with a number of dislocation loops similar to those observed in the in-situ inspected samples. The coexistence of dislocation loops and tangles is especially noticeable in the ex-situ W sample at a damage level of 0.3 dpa.

Discussion Molecular dynamic simulations have revealed that the damaged structure in bcc metals is initially characterized by a relatively large fraction of individual vacancy and self-interstitial atom (SIA) defects, together with a number of small mobile SIAs clusters [50]. The most stable interstitial cluster configuration in W is predicted to be a/2 <111> dislocation loops [51] and this coincides with our TEM results for both W and W-5Ta alloys. However, vacancy dislocation loops are metastable with respect to the transformation into spherical voids [51], and were not observed in the W and W-5Ta samples proton irradiated at 350 C. The vacancy migration enthalpy in tungsten is reported to take the value of 1.78 eV [52, 53]. In fact, the irradiation temperature of 350 C is below the critical temperature of 470 C for the recovery in W due to long-range vacancy migration [53]. Therefore, the radiation damage evolution in this study is mainly governed by the nucleation, diffusion and growth of a/2 <111> interstitial dislocation loops. The average loop size in both materials increases, whereas the number density reduces, with the damage level up to 0.3 dpa. Beyond this level, the loop length in W continues to increases, but the loop width gradually levels off. This trend gives rise to the appearance of the dislocation loop strings in W at higher damage level. Such evolution of the microstructure is in contrast with the loop behaviour in W-5Ta at higher damage levels, where both the loop width and length saturate. The formation of the elongated loops and their ordering in strings could be explained by their mutual elastic interaction and minimization of the dislocation strain energy. The fact that the same process is not observed in W-Ta suggests that Ta might have an impact on the mobility of dislocation loops (and single SIAs).

144

The free surface of the sample is expected to act as an effective non-biased sink for all mobile defects. However, the 1D migration of SIAs, attaining <111> crowdion configuration, should lead to their preferred capture at the surface, thus generating a super-saturation of vacancies. Consequently, 3D-migrating vacancies may form clusters in the near-surface regions of the sample [54]. In this study, we have not observed the appearance of vacancy clusters in the thin foil samples irradiated in-situ. The presence of a relatively high dislocation density, as observed in both the W and W-5Ta samples, would impose high cross section for the SIA- loop interaction and minimise the surface effect [55]. Furthermore, the stress field around dislocation loops is altered in the vicinity of the free surface, so that an attractive image force is generated that would drive the dislocation loops towards the surface. The image force is in competition with the glide and climb forces, that depend on the loop spacing and size, respectively [56]. Our ex-situ TEM analysis reveals the coexistence of dislocation tangles and loops in both materials, as opposed to the lack of dislocation tangles in the in-situ specimens, see Fig. 4.3 and 4.4. This difference points out to the microstructure observed ex-situ by TEM as corresponding to a later stage in the radiation damage accumulation, i.e. to a higher damage level, as compared to the in-situ samples where a fraction of dislocation loops may have escaped to the free surface of the sample. However, we have observed neither changes in the nature (i.e. interstitial to vacancy type) of the dislocation loops and their Burgers vector, nor the presence of a significant number of vacancy clusters, though we do observe the appearance of dislocation tangles [36]. The depth distribution of damage produced is directly proportional to the energy deposited by the proton beam (see Fig. 4.2). At low energies, protons produce damage predominantly with primary knock-on energies around the displacement threshold energy, and mainly create single Frenkel defects and small defect clusters. However, at high energies proton bombardment produces displacement cascades with larger defect clusters, similar to those produced under heavy ion and neutron irradiation. This difference is manifested in the damage function, and the number of clusters produced increases with the primary recoil energy [57, 58]. The string-like features or dislocation strings observed in W samples irradiated in-situ at damage levels above 0.3 dpa resemble rafting observed on (11-1) and (111) planes in neutron-irradiated W samples up to a fluence of 1 ×1022 n cm-2, assuming a Burgers vector of the loops to be a/2 <111> [59]. A density of rafts was estimated as 6.1×1014 rafts cm-2 at 430 C, and the rafts were likely to be associated with dislocation lines. Rafts were reported to comprise black spot clusters and/or the loops of the pure edge type which lie on habit

145 planes having the same indices as the respective raft planes [59]. Moreover, agglomeration of small interstitial loops into rafts has been reported for neutron irradiated molybdenum [60] and in the Mo-based alloy known as TZM [61]. These neutron-induced rafts, and most likely the string-like features observed in the proton-irradiated W samples in this study, may result from two simultaneously operating mechanisms [59]: (1) the preferential drift of SIAs to dislocations loops and (2) the glide and self-climb of those loops. The addition of 5 wt.%Ta seems to hinder the occurrence of string-like features in proton irradiated W materials. The thermal vacancy-mediated diffusion of Ta in W is relative slow, a -4 with an activation energy of E = 6.20 eV and a pre-exponential factor of D0 = 6.20×10 m2/s [21]. An Arrhenius-type dependence of the diffusion coefficient (D) on temperature (T) yields a value of D = 4.3 × 10-54 m2/s at a temperature of 350 C. We have not observed Ta segregation in W-5Ta material by electron energy loss spectroscopy, or the formation of Ta hydrides [62] by electron diffraction. In fact, Ta segregation and clustering has also not been observed in W-4.5Ta (at.%) under 2 MeV W+ irradiation at 300 and 500 C up to a damage level of 33 dpa [21]. However, recent Density Functional Theory calculations revealed the strong repulsive interaction between Ta and interstitial clusters such as crowdion [63] or dumbbell [64] configurations. The presence of Ta retards the mobility of SIAs and interstitial dislocation loops, and consequently the loop growth and coalescence [30], eventually transiting into dislocation strings. H can become trapped at lattice defects or impurities [65]. Both radiation-induced dislocation loops and dislocations formed by cold work are reported to act as effective traps for implanted deuterium in W [66]. Recent Density Functional Theory and continuum rate theory model evaluated the nucleation rate of H bubbles on a dislocation network as a function of depth [67]. Our TEM results can therefore have an impact in the mechanistic understanding of hydrogen isotope retention in these materials under plasma exposure. The H retention mechanisms in these damage microstructures will be subject of a future scope of work that will include not only the influence of the dislocation structure, but also the grain boundary structure and the local mechanical properties of both materials.

Conclusions The combined approach to irradiate W and W-5Ta samples using independently proton beams of two different energies, i.e. 40 keV and 3 MeV, coupled with transmission electron microscopy to assess in-situ/ex-situ the radiation induced damage, allows one to probe the defect structure evolution in these materials at different stages in the damage process,

146 i.e. different damage levels. In both materials, a/2 <111> interstitial dislocation loops form under proton irradiation. The loop size increases whereas the loop density reduces with increasing damage levels up to 0.3 dpa. Loop saturation takes place in W-5Ta at higher damage levels than in W. In contrast, the loop length in W continues to increase with damage level and gives rise to the formation of dislocation strings. Later stages in the damage accumulation sequence, i.e. higher damage levels, are characterised by the co-existence of dislocation loops and tangles in both materials. Ta solid solution delays the evolution of the dislocation structures with increasing proton-induced damage level, as compared to non- doped W, and can consequently have an impact on the hydrogen isotope retention and bubble formation under plasma exposure.

Acknowledgements The work described was supported by the Dalton Cumbrian Facility Project, a joint facility of the University of Manchester and the Nuclear Decommissioning Authority. We thank A.D. Smith and N. Mason for their assistance during the proton irradiation experiment. We acknowledge the support of EPSRC for the development of the MIAMI-1 Facility (EP/E017266/1).

References 1. M. Merola, F. Escourbia, R. Raffray, P. Chappuis, T. Hirai, A. Martin, Fusion Eng. Des. 89 (2014) 890. 2. J.H. You, Nucl. Fusion 55 (2015) 113026. 3. D.M. Yao, G.N. Luo, Z.B. Zhou, L. Cao, Q. Li, W.J. Wang, et al., Phys. Scr. T167 (2016) 014003. 4. E. Lassner, W. S. Schubert, Tungsten: Properties, Chemistry, Technology of the Element, Alloys and Chemical Compounds, 1999, Springer, New York, NY. 5. D.M. Duffy, Phil. Trans. R. Soc. A 2010 368, 3315. 6. M. Rieth, S.L. Dudarev, S.M. Gonzalez de Vicente, J. Aktaa, T. Ahlgren, S. Antusch, et al., J. Nucl. Mater. 432 (2013) 482. 7. R.G. Abernethy, Mater. Sci. Technol. 33 (2017) 388. 8. L. Giancarli, J.P. Bonal, A. Li Puma, B. Michel, P. Sardain, J. F. Salavy, Fusion Eng. Des. 75–79 (2005) 383. 9. N. Baluc, Phys. Scr. T138 (2009) 014004. 10. B. Bazyleva, G. Janeschitz, I. Landman, S. Pestchanyi, A. Loarte, G. Federici, et al., Fusion Eng. Des. 83 (2008) 1077.

147

11. T. Shen, Y. Dai, Y. Lee, J. Nucl. Mater. 468 (2016) 348. 12. V. Philipps, J. Nucl. Mater. 415 (2011) 52. 13. H. Bolt, V. Barabash, G. Federici, J. Linke, A. Loarte, J. Roth, K. Sato, J. Nucl. Mater. 307-311 (2002) 43. 14. S.J. Zinkle, N. M. Ghoniem, Fusion Eng. Des. 51–52 (2000) 55. 15. Ph. Mertens, V. Thompson, G.F. Matthews, D. Nicolai, G. Pintsuk, V. Riccardo, et al., J. Nucl. Mater. 438 (2013) S401. 16. G.A. Cottrell, J. Nucl. Mater. 334 (2004) 166. 17. T. Tanno, A. Hasegawa, J.-C. He, M. Fujiwara, S. Nogami, M. Satou, T. Shishido, K. Abe, Mater. Trans. 48 (2007) 2399. 18. M. R. Gilbert, J.-C. Sublet, Nucl. Fusion, 51 (2011) 043005. 19. T.E. Tietz, J.W. Wilson, Behavior and Properties of Refractory Metals, Stanford University Press, 1965. 20. D. Jiang, Q. Wang, W. Hu, Z. Wei, J. Tong, H. Wan, J. Mater. Res. 31 (2016) 3401. 21. Xu, D.E.J. Armstrong, C. Beck, M.P. Moody, G.D.W. Smith, P.A.J. Bagot, et al., Acta Mater. 124 (2017) 71. 22. Y. Zayachuk, M.H.J. ’t Hoen, P.A. Zeijlmans van Emmichoven, I. Uytdenhouwen, G. van Oost, Nucl. Fusion 52 (2012) 103021. 23. K. Schmid, V. Rieger, A. Manhard, J. Nucl. Mater. 426 (2012) 247. 24. Y. Zayachuk, M.H.J. ’t Hoen, P.A. Zeijlmans van Emmichoven, D. Terentyev, I. Uytdenhouwen, G. van Oost, Nucl. Fusion 53 (2013) 013013. 25. Y. Zayachuk, A. Manhard, M.H.J. ‘t Hoen, W. Jacob, P.A. Zeijlmans van Emmichoven, G. Van Oost, J. Nucl. Mater. 464 (2015) 69. 26. S. Gonderman, J.K. Tripathi, T. Sizyuk, A. Hassanein, Nucl. Fusion 57 (2017) 086001. 27. D. E. J. Armstrong, A. J. Wilkinson, S. G. Roberts, Phys. Scripta, T145 (2011) 014076. 28. X. Hu, T. Koyanagi, M. Fukuda, N.A.P. Kiran Kumar, L.L. Snead, B.D. Wirth, Y. Katoh, J. Nucl. Mater. 480 (2016) 235. 29. Y. Ueda, K. Schmid, M. Balden, J.W. Coenen, Th. Loewenhoff, A. Ito, A. Hasegawa, C. Hardie, M. Porton, M. Gilbert, Nucl. Fusion 57 (2017) 092006. 30. X. Yi, M. L. Jenkins, K. Hattar, P. D. Edmondson, S. G. Roberts, Acta Mater. 92 (2015) 163.

148

31. C.E. Beck, S.G. Roberts, P.D. Edmondson, D.E.J. Armstrong, MRS Proc. 1514 (2013) 99. 32. S.A. Maloy, M.R. James, W. Sommer, G.J. Willcutt, M. Lopez, T.J. Romero, Mater. Trans. 43 (2002) 633. 33. J.A. Hinks, Nucl. Instr. Meth. Phys. Res. A 267 (2009) 3652. 34. S. Ishino, Mater. Trans. 55 (2014) 396. 35. T. Muroga, N. Yoshida, N. Tsukuda, K. Kitajima, M. Eguchi, Mater. Sci. Forum 15- 18 (1987) 1087. 36. A. Prokhodtseva, B. Décamps, R. Schäublin, J. Nucl. Mater. 442 (2013) S786. 37. R. Sakamoto, T. Muroga, and N. Yoshida, J. Nucl. Mater. 222 (1995) 819. 38. J. Linke, T. Loewenhoff, V. Massaut, G. Pintsuk, G. Ritz, M. Rödig, A. Schmidt, C.Thomser, I. Uytdenhouwen, V. Vasechko, M.Wirtz, Nucl. Fusion 51 (2011) 073017. 39. J.A. Hinks, J.A. van den Berg, S.E. Donnelly, J. Vac. Sci. Technol. A 29 (2011) 21003. 40. P.T. Wady, A. Draude, S.M. Shubeita, A.D. Smith, N. Mason, S.M. Pimblott, E. Jimenez-Melero, Nucl. Instr. Meth. Phys. Res. A 806 (2016) 109. 41. M.L. Jenkins, M.A. Kirk, Characterization of Radiation Damage by Transmission Electron Microscopy, IOP Publishing Ltd, 2001. 42. Harte, D. J ̈dern ̈s, M. Topping, P. Frankel, C.P. Race, J. Romero, L. Hallstadius, E.C. Darby, M. Preuss, Acta Mater. 130 (2017) 69. 43. J.F. Ziegler, J. Appl. Phys. 85 (1999) 1249. 44. J.F. Ziegler, M.D. Ziegler, J.P. Biersack, Nucl. Instr. Meth. Phys. Res. B 268 (2010) 1818. 45. ASTM E521-96, Standard Practice for Neutron Radiation Damage Simulation by Charged-particle Irradiation, 2009. 46. H. Paul, Nucl. Instr. Meth. Phys. Res. B 268 (2010) 3421. 47. H. Paul, Daniel Sánchez-Parcerisa, Nucl. Instr. Meth. Phys. Res. B 312 (2013) 110. 48. M. Peñalba, A. Arnau, P.M. Echenique, F. Flores, R.H. Ritchie, Europhys. Lett., 19 (1992) 45. 49. E.D. Cantero, G.H. Lantschner, J.C. Eckardt, N.R. Arista, Phys. Rev. A 80 (2009) 032904. 50. B.N. Singh, J.H. Evans, J. Nucl. Mater. 226 (1995) 277.

149

51. M.R. Gilbert, S.L. Dudarev, P.M. Derlet, D.G. Pettifor, J. Phys.: Condens. Matter 20 (2008) 345214. 52. D. Nguyen-Manh, A. P. Horsfield, S. L. Dudarev, Phys Rev. B 73 (2006) 020101. 53. D.N. Seidman, Scripta Metall. 13 (1979) 251. 54. M. Kiritani, Mater. Sci. Forum 15-18 (1987) 1023. 55. D.I.R. Norris, Radiat. Eff. 15 (1972) 1. 56. F. Ferroni, E. Tarleton, S. Fitzgerald, Modelling Simul. Mater. Sci. Eng. 22 (2014) 045009. 57. K.L. Merkle, W.E. King, A.C. Baily, K. Haga, M. Meshi, J. Nucl. Mater. 117(1983) 4. 58. C. Abromeit, J. Nucl. Mater. 216 (1994) 78. 59. V.K. Sikka, J. Moteff, J. Nucl. Mater. 46 (1973) 217. 60. J.L. Brimhall, B. Mastel, Radiat. Eff. 3(1970) 203. 61. B.N. Singh, J.H. Evans , A. Horsewell, P. Toft, D.J. Edwards J. Nucl. Mater. 223 (1995) 95. 62. K.-T. Park, J.-H. Park, J.-H. Yoon, J.-E. Lee., I.-K. Park, Int. J. Refract. Met. Hard Mat. 65 (2017) 83. 63. M. Muzyk, D. Nguyen-Manh, K. J. Kurzydłowski, N. L. Baluc, S. L. Dudarev, Phys. Rev. B 84 (2011) 104115. 64. W. Setyawan, G. Nandipati, R.J. Kurtz, J. Nucl. Mater. 484 (2017) 30. 65. T. Tanabe, Phys. Scr. T159 (2014) 014044. 66. R. Sakamoto, T. Muroga, N. Yoshida, J. Nucl. Mater. 2.3-237 (1996) 776. 67. D. Terentyev, V. Dubinko, A. Bakaev, Y. Zayachuk, W. Van Renterghem, P. Grigorev. Nucl. Fusion 54 (2014) 042004.

150

Appendix A

The dislocation loops and networks observed in both materials at later stages act as effective hydrogen trapping sites, so as to generate hydrogen bubbles and surface blisters. Figure 4.6 displays the secondary electron images of both materials irradiated ex-situ up to two different damage levels at the Bragg peak position, namely 1 and 2 dpa. The surface of W presents a significant number of dome-like blisters, whose size increases with damage level. This is in stark contrast with the surface characteristics observed for the W-5Ta alloy, where both the number and size of blisters are significantly reduced as compared to those observed in W at the same damage level. In fact, in W-5Ta blisters start to be visible at the damage level of 2 dpa at the Bragg peak position, whereas in the case of W the blisters are already formed and grown to a few hundreds of microns in size at this damage level. However, the H retention and blister formation are rather complex processes that involve not only the ion-damaged structure but also the grain boundaries, local stresses and the mechanical properties of the material itself. We have decided to eliminate the discussion about H retention and blistering. These phenomena in W-Ta will be subject to a full scope of work. 1dpa 2 dpa

500 µm 300 µm 500 µm 300 µm

W-5Ta

500 µm 300 µm 500 µm 300 µm Figure 4.6 Secondary electron images of ex-situ W and W-5Ta samples irradiated at 350(4)°C. The damage level corresponds to the value at the Bragg peak position in each case.

151

5. Thermal evolution of the proton irradiated structure in tungsten-5wt.%

tantalum

I. Ipatovaa,b,*, R.W. Harrisonc, D. Terentyevd, S.E. Donnellyc, E. Jimenez-Meleroa

a School of Materials, The University of Manchester, Manchester M13 9PL, UK

bDalton Cumbrian Facility, The University of Manchester, Moor Row CA24 3HA, UK

cSchool of Computing and Engineering, University of Huddersfield,

Huddersfield, HD1 3DH, UK

dSCK·CEN, Nuclear Materials Science Institute, Boeretang 200, Mol, 2400, Belgium

Published in Journal of Fusion Energy

DOI: 10.1007/s10894-017-0145-y

Corresponding author (*):

University of Manchester

School of Materials

Oxford Road

Manchester

M13 9PL

United Kingdom

Tel.: +44 7849290480

Email: [email protected]

152

Abstract We have monitored the thermal evolution of the proton irradiated structure of W-5wt.%Ta alloy by in-situ annealing in a transmission electron microscope at fusion reactor temperatures of 500-1300oC. The interstitial-type a/2<111> dislocation loops emit self- interstitial atoms and glide to the free sample surface during the early stages of annealing. The resultant vacancy excess in the matrix originates vacancy-type a/2<111> dislocation loops that grow by loop and vacancy absorption in the temperature range of 600-900oC. Voids form at 1000oC, by either vacancy absorption or loop collapse, and grow progressively up to 1300oC. Tantalum delays void formation by a vacancy-solute trapping mechanism. Keywords: refractory metals, radiation damage, lattice defects, annealing, transmission electron microscopy

Introduction Tungsten and its alloys are the prime structural and armour material candidates for high-heat load components facing the hydrogen plasma in nuclear fusion reactors [1-3], due to their unique thermal conductivity and high-temperature strength, in combination with low surface sputtering yields and resistance to radiation-induced swelling [4-6]. The thermal loads on plasma-facing components are predicted to be 0.1-20 MWm-2 during steady-state reactor operation [7, 8], and even values close to GW/m2 range during off-normal events such as plasma disruptions and edge localised events [9]. The expected lowest shield temperatures for W-based armour materials range from 500C in the blanket‟s first wall to >800-900C in helium-cooled divertor designs and even >1700C in the divertor armour surface [1]. Unfortunately, tungsten manifests an intrinsic brittleness at low temperatures, with reported values of the ductile-to-brittle transition temperature (DBTT) in the range of 250-350C, depending on its microstructure and processing route [10, 11]. Neutron irradiation is reported to increase the DBTT up to 800–1000°C [12, 13]. In plasma-facing components of fusion reactors, the high neutron fluxes are predicted to induce radiation damage rates of 3- 30dpa/year [8]. Besides that, neutron bombardment will also cause the transmutation of W into Re and subsequently Os, and therefore trigger the formation of brittle  (ReW) and 

(WRe3) precipitates [14, 15]. The radiation-induced embrittlement stemming from lattice defects and brittle phases will continuously reduce the recommended temperature window for safe operation of W-based components [13], and potentially cause premature failure during low-temperature reactor operations [16].

153

In this work, we have assessed the radiation-induced lattice damage in W-5wt.%Ta (W-5Ta) alloy at the relatively low temperature of 350C, and how the damaged structure evolves during annealing up to 1300C. Tantalum additions to tungsten, either in solid solution or as short fibres, can potentially increase its ductility [1, 17] and also delay the formation of brittle phases stemming from the neutron-induced transmutation sequence [14]. Recent reports on self-ion irradiations of W-Ta alloys at temperatures 500C and up to a damage level of 33dpa confirm the absence of radiation-induced Ta clustering. Moreover, the presence of Ta hinders the formation of Re clusters under irradiation [18]. Interstitial-type dislocation loops form at those low irradiation temperatures in W-Ta alloys. The loop size is smaller and the loop number density is higher than in tungsten under equivalent irradiation conditions [19]. The low-temperature irradiation-hardening due to dislocation loops saturates at a damage level of 13dpa at 300C [20]. Ta is predicted to have a repulsive interaction with self- interstitial atom (SIA) defects such as <111> crowdion [21] and dumbbell [22] configurations, and potentially restrict the mobility of SIAs and interstitial loops in W [19].

Experimental details The as-received W-5Ta material was initially annealed in vacuum (~10-5 mbar) at1400°C for 2 hours for recrystallization. A 3mm-diameter TEM disc was prepared by mechanical pre- thinning, followed by electropolishing at a temperature of ~ -5C, using a Struers Tenupol-5 unit and an electrolyte comprising 15vol.% sulphuric acid (95%) and 85vol.% methanol. The TEM disc was proton irradiated at the Microscope and Ion Accelerator for Materials Investigations (MIAMI-1 facility) located at the University of Huddersfield [23]. The sample was placed in a high-temperature Gatan 652 double-tilt specimen holder, and then mounted in a JEOL JEM-2000FX TEM operating at an accelerating voltage of 200 kV. This microscope is installed at the end of a beam line connected to an electrostatic ion accelerator. The 40keV proton irradiation was performed at a temperature of 350(2)oC, up to a damage level of 0.7dpa and a fluence of 9.7×1017 ions/cm2. The proton irradiated structure was characterised at room temperature using an FEI Tecnai T20 Transmission Electron microscope with LaB6 crystal, equipped with a double-tilt specimen holder and operating at 200kV. The determination of the Burgers vector b of the observed dislocations made use of the g.b=0 invisibility criterion, where g denotes the scattering vector. The nature of the dislocation loops was determined using the inside-outside contrast method. The foil thickness was

154 derived from the fringes spacing of the convergent beam electron diffraction pattern, and was used to obtain the dislocation density applying graphical method [24]. Afterwards, the evolution of the proton irradiated structure was monitored in-situ during annealing using the same 200 kV JEOL JEM-2000FX TEM and the heating Gatan 652 double-tilt specimen holder. The disc was first heated from room temperature up to 500C at a rate of 100°C/min, and thereafter in steps of 100C up to 1300C at a lower rate of 50°C/min. After each temperature step, the disc was maintained at a constant temperature for a period of 20min, and TEM images were taken close to the [111] zone axis of the bcc structure of W-5Ta. Void analysis of the bright-field images was based on the “out-of-focus” imaging technique, where voids appear as white dots surrounded by black Fresnel fringes when recorded in an under-focused condition, and as dark dots with bright fringes in an over- focused condition [24]. Once the maximum temperature of 1300C was reached, the disc was cooled to room temperature, and the annealed structure analysed using the FEI Tecnai T20 Transmission electron microscope.

Results and discussion Dislocation structure The structure of the proton irradiated W-5Ta alloy is shown in Figure 5.1, together with its evolution during post irradiation annealing up to 1300C.

As irradiated 600oC 900oC

200 nm 200 nm 200 nm 1000oC 1100oC 1300oC

200 nm 200 nm 200 nm Figure 5.1 Dislocation loop evolution in W-5Ta alloy during annealing, after proton irradiation at a temperature of 350C and a damage level of 0.7dpa.

155

The as-irradiated microstructure is characterized by the presence of interstitial dislocation loops with a Burgers vector a/2<111>. The average loop diameter is 33(2) nm and the number density takes a value of 2.2×1021m-3. In contrast, the microstructure annealed at 1300C presents a lower dislocation density of 0.3×1021m-3 and the average loop diameter is 26(1) nm. The Burgers vector of the loops after annealing is still a/2<111>, however, the ellipsoidal shaped loops were determined to be of vacancy-type. The irradiated structure in bcc metals is characterized by a relatively large fraction of individual vacancy and self- interstitial atom (SIA) defects, together with a number of small mobile SIAs clusters, according to molecular dynamic simulations [25]. The enthalpy of migration of SIAs in tungsten equals to 85meV [26, 27], and SIAs are mobile at temperatures as low as -253C corresponding to annealing stage I [28, 29]. Both the free surface of the sample and the interstitial dislocation loops act as biased sinks for mobile SIAs [30]. Furthermore, an attractive image force is generated that drives the dislocation loops towards the surface [31]. Small interstitial a/2<111> loops have recently been observed to hop in one-dimension during in-situ isochronal annealing of a TEM tungsten disc, previously irradiated with a 2MeV W+ beam to a ion fluence of 1014 W+/cm2. The loop hopping is significant at temperatures in the range of 300-700C, with increasing hopping frequency with temperature, and leads to loop loss to the sample surface [32]. Besides that, detrapping of SIAs from dislocations occurs in tungsten in the annealing stage II at -170-430C [28]. Therefore, the interstitial-type a/2<111> loops observed in the „as-irradiated‟ structure of W-5Ta are not stable below 500C, and either glide towards the sample surface and/or lose SIAs, who may also migrate and become trapped at the free surface. The migration of SIAs and interstitial dislocation loops towards the surface generates an excess of vacancies in near-surface regions of the sample [33]. Vacancy migration takes place in tungsten during the annealing stage III, corresponding to the temperature range of 430- 650C [28, 34]. This range is shifted towards lower temperatures with increasing neutron fluence [35]. The activation enthalpy of long-range vacancy migration in tungsten is reported to be 1.7-2.0eV [36, 37]. In fact, significant migration of single vacancies to small vacancy clusters was observed in tungsten at 400C using field ion microscopy [38]. Therefore, we can expect that a significant amount of the initial interstitial loops have disappeared during heating the W-5Ta sample up to 500oC, and the structure is characterized by the presence of additional vacancy-type a/2<111> dislocation loops. At that temperature, the loop density amounts to 1.5×1021m-3 and the average loop size to 28(2) nm. A step-wise increase in

156 annealing temperature above 500oC causes a progressive reduction in loop density, which levels off at temperatures close to 900C, see Figure 5.2. Vacancy-type defects become mobile in this temperature regime, which corresponds to annealing stage IV reported to occur at 650-1000C [28]. This trend in loop characteristics evidences the growth of vacancy-type loops by loop and vacancy absorption and, to a lesser degree, by loop coalescence [39]. At 900C the loop density equals to 0.8×1021m-3 and the average loop size to 26(2) nm.

Figure 5.2 Variation in the number density of dislocation loops in W-5Ta alloy during post irradiation annealing up to 1300C. The coloured region of the graph represents the temperature range where voids are present in the microstructure during annealing. Inset TEM images show evidence of dislocation loop absorption with increasing temperature.

Void formation Annealing the sample at temperatures above 900C reveals a second progressive decrease in loop density, see Figure 5.2. This overlaps with the annealing stage V, whose onset is reported to occur at 1000C in tungsten [28]. This stage is characterised by the thermal dissociation of vacancies from lattice defects such as dislocation loops. Furthermore, we have also observed the occurrence of a significant number of voids in the microstructure at a temperature of 1000C, see Figure 5.3.

157

900°C 1000°C

200 nm No voids 200 nm

11001200°C°C 1300°C

200 nm 200 nm

Figure 5.3 Occurrence and evolution of voids in proton irradiated W-5Ta alloy during annealing.

At that temperature, the void number density amounts to 5.8×1021m-3. An increase in temperature beyond 1000C causes a decrease in the number density of voids, and concomitantly to an increase in the average void diameter, see Figure 5.4. The average number of vacancies per void (n), assuming a spherical shape of the voids, can be estimated using the expression [40]:

(1)

where denotes the average void radius and the atomic volume. The values obtained for n as a function of the annealing temperature are shown in the inset of Figure 5.4b.

158

Figure 5.4 Number density and average diameter of voids formed in irradiated W-5Ta alloy as a function of annealing temperature. The inset in (b) shows the dependence of the average number of vacancies per void on the annealing temperature (see text).

At the highest annealing temperature of 1300C in this study, the average number of vacancies per void is n 18×103, and the void number density is 1.2×1021m-3. Voids are formed either by the clustering of free vacancies, or by the shrinkage and collapse of vacancy-type dislocation loops via vacancy emission [41]. Once voids reach a critical size, they are stable and increase their size by absorbing additional mobile vacancies from the matrix. In fact, mesoscopic vacancy-type dislocation loops are predicted to be metastable with respect to the transformation into spherical voids [42]. Radiation-induced void swelling takes place in neutron irradiated tungsten in the temperature range of 600-900C at a damage level of 9.5dpa [43]. Voids are reported to form in tungsten at an early stage of irradiation, and a void lattice appears at a damage level of 1dpa at temperatures of 600-800C. Moreover, once a void lattice is formed in tungsten, it is expected to remain stable at higher dpa level [44]. We have observed the appearance of voids visible by TEM in W-5Ta alloy at a higher temperature, i.e. 1000C, as compared to tungsten. Tantalum constitutes an oversized atom in tungsten, and can reduce the free vacancy concentration in the matrix via a

159 solute-trapping mechanism [45], therefore delaying or reducing the void formation and the related swelling in W-5Ta alloy as compared to tungsten.

Conclusions To summarize, the microstructure of W-5Ta alloy after 40 keV proton irradiation at 350°C and a damage level of 0.7 dpa is characterized by the presence of interstitial-type a/2<111> dislocation loops. The in-situ post-irradiation annealing of W-5Ta in a TEM causes those interstitial loops to emit SIAs and also to glide to the free surface of the sample. The resultant excess of vacancies in the matrix evolves into vacancy-type a/2<111> dislocation loops, which grow preferentially by loop and vacancy absorption in the temperature range of 600- 900C. At a temperature of 1000C a significant number of voids appear in the microstructure, whereas the loop number density decreases. The void number density reduces gradually up to a temperature of 1300C, and the average void diameter increases simultaneously. Tantalum delays the onset of void formation in W-5Ta alloy, as compared to tungsten, via a solute-vacancy trapping mechanism.

Acknowledgments The work described was supported by the Dalton Cumbrian Facility Project, a joint facility of The University of Manchester and the Nuclear Decommissioning Authority. The authors are also acknowledge access to the MIAMI facility through the EPSRC funded UK national ion beam centre.

References 1. M. Rieth, S.L. Dudarev, S.M. Gonzalez de Vicente, J. Aktaa, T. Ahlgren, S. Antusch, et al. J. Nucl. Mater. 442 (2013) S173. 2. S. Wurster, N. Baluc, M. Battabyal, T. Crosby, J. Du, C. García-Rosales, et a., J. Nucl. Mater. 442 (2013) S181. 3. R. Neu, J. Riesch, J.W. Coenen, J. Brinkmann, A. Calvo, S. Elgeti, et al., Fusion Eng. Des. 109–111 (2016) 1046. 4. D.M. Duffy, Phil. Trans. R. Soc. A 368 (2010) 3315. 5. T. Shen, Y. Dai, Y. Lee, J. Nucl. Mater. 468 (2016) 348. 6. M. Wirtz, I. Uytdenhouwen, V. Barabash, F. Escourbiac, T. Hirai, J. Linke, et al., Nucl. Fusion 57 (2017) 066018.

160

7. L. Giancarli, J.P. Bonal, A. Li Puma, B. Michel, P. Sardain, J. F. Salavy, Fusion Eng. Des. 75–79 (2005) 383. 8. N. Baluc, Phys. Scr. T138 (2009) 014004. 9. B. Bazyleva, G. Janeschitz, I. Landman, S. Pestchanyi, A. Loarte, G. Federici, et al., Fusion Eng. Des. 83 (2008) 1077. 10. T. Shen, Y. Dai, Y. Lee, J. Nucl. Mater. 468 (2016) 348. 11. V. Philipps, J. Nucl. Mater. 415 (2011) 52. 12. H. Bolt, V. Barabash, G. Federici, J. Linke, A. Loarte, J. Roth, K. Sato, J. Nucl. Mater. 307-311 (2002) 43. 13. S.J. Zinkle, N. M. Ghoniem, Fusion Eng. Des. 51–52 (2000) 55. 14. G.A. Cottrell, J. Nucl. Mater. 334 (2004) 166. 15. A. Hasegawa, M. Fukuda, T. Tanno, S. Nogami, Mater. Trans. 54 (2013) 466. 16. Ph. Mertens, V. Thompson, G.F. Matthews, D. Nicolai, G. Pintsuk, V. Riccardo, et al., J. Nucl. Mater. 438 (2013) S401. 17. D. Jiang, Q. Wang, W. Hu, Z. Wei, J. Tong, H. Wan, J. Mater. Res. 31 (2016) 3401. 18. A. Xu, D.E.J. Armstrong, C. Beck, M.P. Moody, G.D.W. Smith, P.A.J. Bagot, et al., Acta Mater. 124 (2017) 71. 19. X. Yi, M. L. Jenkins, K. Hattar, P. D. Edmondson, S. G. Roberts, Acta Mater. 92 (2015) 163. 20. D. E. J. Armstrong, A. J. Wilkinson, S. G. Roberts, Phys. Scripta, T145 (2011) 014076. 21. M. Muzyk, D. Nguyen-Manh, K. J. Kurzydłowski, N. L. Baluc, S. L. Dudarev, Phys Rev. B 84 (2011) 104115. 22. W. Setyawan, G. Nandipati, R.J. Kurtz, J. Nucl. Mater. 484 (2017) 30. 23. J.A. Hinks, J.A. van den Berg, S.E. Donnelly, J. Vac. Sci. Technol. A 29 (2011) 21003. 24. M.L. Jenkins, M.A. Kirk, Characterization of Radiation Damage by Transmission Electron Microscopy, IOP Publishing Ltd, 2001. 25. B.N. Singh, J.H. Evans, J. Nucl. Mater. 226 (1995) 277. 26. R. M. Scanlan , D. L. Styris, D. N. Seidman, Philos. Mag. 23 (1971) 1459. 27. F. Dausinger, H. Schultz, Phys. Rev. Lett. 35 (1975) 1773. 28. L.K. Keys, J. Moteff, J. Nucl. Mat. 34 (1970) 260. 29. R. M. Scanlan , D. L. Styris, D. N. Seidman, Philos. Mag. 23 (1971) 1439. 30. D.I.R. Norris, Radiat. Eff. 15 (1972) 1.

161

31. F. Ferroni, E. Tarleton, S. Fitzgerald, Modelling Simul. Mater. Sci. Eng. 22 (2014) 045009. 32. F. Ferroni, X. Yi, K. Arakawa, S.P. Fitzgerald, P.D. Edmondson, S.G. Roberts, Acta Mater. 90 (2015) 380-393. 33. M. Kiritani, Mater. Sci. Forum 15-18 (1987) 1023. 34. D.N. Seidman, Scripta Metall. 13 (1979) 251. 35. L. K. Keys, J. P. Smith, J. Moteff, Scripta Metall. 1 (1967) 71. 36. M.W. Thompson, Phil. Mag. 5 (1960) 278. 37. H. Schultz, Scripta Metall. 8 (1974) 721. 38. K.M. Bowkett, B. Ralph, Proc. Roy. Soc. A 312 (1969) 51. 39. X. Yi, M. L. Jenkins, M. Briceno, S. G. Roberts, Z. Zhou, M. Kirk, Philos. Mag. 93 (2013) 1715. 40. G. Was. Fundamentals of Radiation Materials Science. Springer-Verlag, 2007. 41. B.L Eyre, D.M. Maher, Phil. Mag., 24 (1971) 767. 42. M.R. Gilbert, S.L. Dudarev, P.M. Derlet, D.G. Pettifor, J. Phys.: Condens. Matter 20 (2008) 345214. 43. J. Matolich, H. Nahm, J. Moteff, Scripta Metall. 8 (1974) 837. 44. A. Hasegawa, M. Fukuda, S. Nogami, K. Yabuuchi, Fusion Eng. Des. 89 (2014) 1568. 45. F. A. Smidt, J. A. Sprague, Scripta Metall. 7 (1973) 495.

162

6. Characterization of lattice damage formation in tantalum irradiated at variable temperatures

I. Ipatovaa,b,*, R.W. Harrisonc, P.T. Wadyb, S.M. Shubeitab, D. Terentyevd, S.E. Donnellyc, E. Jimenez-Meleroa

Published in Journal of Microscopy

DOI 10.1111/jmi.12662

Corresponding author (*): University of Manchester School of Materials Oxford Road Manchester M13 9PL United Kingdom Tel.: +44 7849290480 Email: [email protected]

163

Summary The formation of radiation-induced dislocation loops and voids in tantalum at 180(2), 345(3) and 590(5)C was assessed by 3MeV proton irradiation experiments and subsequent damage characterization using transmission electron microscopy. Voids formed at 345(3)C and were arranged into a body centred cubic lattice at a damage level of 0.55dpa. The low vacancy mobility at 180(2)C impedes enough vacancy clustering and therefore the formation of voids visible by TEM. At 590(5)°C the Burgers vector of the interstitial-type dislocation loops is a<100>, instead of the a/2 <111> Burgers vector characteristic of the loops at 180(2) and 345(3)C. The lower mobility of a<100> loops hinders the formation of voids at 590(5)C up to a damage level of 0.55dpa.

Lay description High-temperature metallic materials for demanding technological applications in radiation environments, such as future nuclear reactor systems and enhanced-output targets for spallation sources, will be subject to the continuous bombardment of energetic particles at elevated temperatures. Tantalum constitutes an advanced candidate material for those applications due to its high radiation tolerance, ductility and water corrosion resistance. We have characterized the damage caused by proton bombardment to tantalum at variable temperatures using transmission electron microscopy. The results revealed significant differences in structural damage as a function of temperature and damage level, and therefore help to understand the formation, or otherwise, of voids induced by radiation. These results constitute unique evidence of the occurrence of structural damage and voids in tantalum, which will pave the way to reliable predictions of radiation-induced swelling of tantalum- based components in future applications. Keywords: Nuclear materials, Tantalum, Electron microscopy, Radiation damage, Dislocation analysis, Void formation.

164

Introduction The continuous exposure of structural materials to intense radiation fields causes atomic displacement cascades that generate a high density of vacancies and self-interstitial atoms (SIAs) in equal proportions. Those lattice point defects can evolve into a range of defect structures such as dislocation loops, nano-clusters, stacking fault tetrahedra or voids (Was, 2007; Odette et al., 2008). The long-term lattice damage brings along phenomena such as radiation-induced hardening or void swelling that compromises the integrity of key structural components in applications such as nuclear reactor cores or targets in neutron spallation sources (Yvon et al., 2009; Azevedo, 2011; Zinkle et al., 2013). Radiation-induced void swelling was initially reported in face-centred cubic metals and alloys, such as austenitic stainless steel or nickel (Cawthrone et al, 1967; Norris, 1971). Material candidates for future nuclear reactor systems and higher-output targets for spallation sources are mainly based on body-centred cubic (bcc) metals due to their enhanced radiation tolerance and, in particular, to their higher void swelling resistance (Singh et al., 1995; Stork et al., 2014). Tantalum currently stands out as an advanced high-temperature material candidate due to its additional benefits of a high fluence threshold for He+ ion-induced surface nano-structuring, high water corrosion resistance, good workability, and ductility retention at relatively high radiation damage levels (Chen et al., 2003; Nelson et al., 2012; Novakowski et al., 2016). Void formation and swelling in bcc metals is interpreted in terms of the biased absorption of SIAs at defect sinks such as dislocation loops, and the evolution of the resultant vacancy excess in the matrix into vacancy clusters and eventually voids (Evan et al., 1985; Trinkaus et al., 1993). This overall process depends on the mobility of SIAs, vacancies and defect sinks, and therefore on temperature. However, the existing experimental evidence of the formation and evolution of SIA and vacancy arrangements at variable temperatures in tantalum is still fragmented. Earlier radiation damage studies in tantalum bombarded with neutrons up to a fluence of 2.5×1022 neutron/cm2 at a temperature of 585C reported the appearance of voids ordered on a bcc superlattice. Those voids randomise at higher irradiation temperatures up to 1050C and are absent at 425C (Wiffen, 1977). Heavy ion bombardment of tantalum induces the formation of vacancy clusters that evolve into randomly distributed voids at temperatures higher than 400C (Yasunaga et al., 2000). In this work we aimed to assess the formation of radiation-induced dislocation structures in tantalum at variable temperatures and their impact on the occurrence, or otherwise, of void formation. We used a 3MeV proton beam produced by an electrostatic particle accelerator as a surrogate of neutron bombardment

165

in void formation studies, since it generates a uniform radiation damage profile through a sample thickness of 30 μm. Ion irradiation experiments have been performed successfully since 1970s on metallic materials to mimic void swelling induced by neutron damage (Nelson et al., 1970; Mazey, 1990).

Experimental Proton irradiation The starting Ta material was provided by Goodfellow Cambridge Ltd. in the form of 1mm- thick sheet. The as-received material was annealed at 1400°C during 2 hours for recrystallization. Irradiation experiments were performed using the 3 MeV proton beam generated by a 5 MV tandem ion accelerator and a high-current TORVIS source installed at the Dalton Cumbrian Facility of the University of Manchester (Wady et al., 2016). Equivalent samples were irradiated at either 180(2), 345(3) or 590(5)C. During irradiation experiment the temperature was monitored by a non-contact IR camera placed at an angle of 30° with respect to the sample surface. Additionally, thermocouples were spot welded onto the samples close to the irradiated area. Liquid indium was used in order to improve the thermal contact between the samples and the NIMONIC75® alloy block of the sample stage, the latter containing the heaters and the water cooling loop. The average temperature values correspond to the irradiation period when the temperature was held relatively constant, i.e. excluding the heating to reach the target temperature and the final cooling down to room temperature, and the error in the irradiation temperature corresponds to the standard deviation of the temperature distribution during that period. The charge accumulated on the sample was measured during the irradiation experiment using a picoammeter. The accumulated charge was used to calculate the proton fluence. The irradiation conditions for the studied samples are summarised in Table 6.1. The damage rate was 1×10-5dpa/s in all irradiations.

166

Table 6.1. Summary of irradiation conditions used for tantalum at selected irradiation temperatures and radiation damage levels.

Irradiation Damage Damage Sampling Fluence Flux temperature level rate depth (1018 protons/cm2) (1018protons/cm2/h) (oC) (dpa) (10-6 dpa/s) (µm) 0.15 2.2 20 180(2) 4.4 0.2 0.55 9.3 29 0.9 0.9 0.15 8.9 28 345(3) 3.7 0.8 0.55 5.9 29 0.15 2.2 20 590(5) 4.4 0.2 0.55 9.3 29

The thickness of the irradiated layer of the samples was determined by nano-indentation measurements using a NanoIndenter-XP (MTS Systems Corp.) after sample cross sectioning, and supported by simulations using the SRIM software with the quick Kinchin–Pease approach and default values for other software settings (Ziegler et al., 2010; ASTM E521-96, 2009), see Figure 6.1.

Figure 6.1 Simulated damage profile using the SRIM software, together with the nano- hardness values measured along the cross section of the tantalum sample irradiated at 180(2)°C. The asterisks indicate the regions of the sample from where TEM foils were extracted for analysis.

167

Characterization of radiation damage Three millimetre diameter discs for transmission electron microscopy (TEM) were prepared by electro-polishing using the Struers TenuPol-5 and an electrolyte containing 15vol.% sulphuric acid (95%) – 85vol.% methanol at a temperature of 5C. Regions of the irradiated layer at two damage levels, namely 0.15 and 0.55 dpa, were selected for further analysis. In order to achieve the targeted level of damage, prior to electropolishing the non-irradiated side of the sample was covered with Elektron Technology’s acid-resistant Lacomit varnish which was simply removed thereafter by acetone. The sample was electro-polished from the irradiated side to a preselected depth below the surface. We assessed the amount of material removed by electro-polishing using a Keyence VK-X200K 3D Laser Scanning Microscope. A sample thickness of 20μm and 28μm was removed in order to obtain a TEM thin foil at the damage level of 0.15 dpa and 0.55 dpa respectively. Afterwards, the Lacomit varnish was removed and applied to the irradiated side of the sample, and finally the sample was back thinned for electron transparency by electropolishing (Z. Yao et al., 2008). The structural analysis was performed using the 200kV FEI Tecnai T20 Transmission Electron microscope equipped with a double-tilt specimen holder. The determination of the Burgers vector b of the dislocations was based on the g.b=0 invisibility criterion, where g denotes the scattering vector. This determination was based on estimating all g.b possible values for the relevant reflections from the TEM analysis made by tilting the sample to specific two-beam conditions. Prismatic dislocations in bcc metals are characterised by the Burgers vector a/2<111>, a<001> or a/2<110> (Okamoto et al., 1966). The predicted invisibility, or otherwise, of the loops was confronted with the experimental evidence from the acquired TEM images. The foil thickness was derived from the fringes spacing of the convergent beam electron diffraction (CBED) pattern. The error in thickness measurements using CBED patterns from the exact Bragg condition is assumed to be ±10% (Harte et al., 2017). Bright- field imaging of small radiation-induced voids was based on the “out-of-focus” imaging technique (Jenkins et al., 2001).

Results and discussion Dislocation structure The lattice damage in tantalum is evidenced by the presence of a relatively high density of ellipsoidal dislocation loops at the three studied irradiation temperatures and a damage level of 0.15dpa (Figure 6.2).

168

Figure 6.2 Evolution of dislocation structure in tantalum with damage level and temperature. TEM data taken along the [111], [113], [0 ̅2] zone axis.

The TEM images of dislocation loops presented in Figure 6.2 have been taken along the [111], [113] or [0 ̅2] zone axis. These loops are formed in the debris of the damage cascade by the migration and clustering of SIAs. At that damage level, the number density of dislocation loops observed at 180(2)°C is the highest (~110×1021m-3), and corresponds to approximately three times the loop density at 345(3)°C (~40×1021m-3) and eight times the value at 590(5)°C (~13×1021m-3). The reduction in loop density with increasing temperature occurs simultaneously with an increase in the average loop dimensions (Figure 6.3).

169

Figure 6.3 Average dislocation loop size and number density in tantalum as a function of temperature at 0.15dpa and 0.55dpa.

These results manifest the predominance of loop growth over nucleation as the irradiation temperature increases. The dislocation loops are determined to be of interstitial nature based on the inside-outside technique (Figure 6.4). Interstitial loop nucleation is easier than vacancy loop nucleation, since interstitial loop nucleation is much less sensitive to vacancy involvement than vacancy loop nucleation is to interstitial involvement (Russell et al., 1973). The loop growth rate increases with temperature in irradiated, annealed metals, i.e. at a higher temperature there are fewer loops but they are larger (Norris, 1972). The escape probability for an interstitial close to a vacancy may increase with increasing temperature (Urban et al., 1971), and small loops would therefore be unstable at higher temperatures. Moreover, the attractive elastic interaction with dislocations is larger for interstitials than for vacancies (Urban, 1970), so that larger loops would be able to further increase in size by absorbing more interstitials at higher temperatures. Besides that, recent in situ TEM studies revealed the small loop-loop interaction and coalescence in bcc W-based materials at higher temperatures (Yi et al., 2015). In both cases, small loops disappear at high temperatures in bcc metals in favour of larger loops.

170

Figure 6.4 Determination of the nature of the dislocation loops by the inside-outside technique in proton-irradiated tantalum at 590(5)°C. The loop show inside contrast in the g= ̅21 reflection and outside contrast in the g=12 ̅ which belong to the [0 ̅2] zone axis. The Burgers vector of the loops was determined to be a<001> (see text).

The loops are characterised by the a/2 <111> Burgers vector at 180(2) and 345(3)C. However, the dislocation analysis revealed that the loops present at 590(5)C correspond to the a<100> Burgers vector (Figure 6.5). Since the elastic energy associated with the dislocations is proportional to b2, where b denotes the length of the Burgers vector, a/2<111> type loops would be favoured (Hull et al., 2011). Traditionally the formation of both types of perfect dislocations is described by the shearing of a/2 <110> partial dislocations at an early stage of growth, according to the classical dislocation reaction theory (Eyre et al., 1965):

〈 〉 〈 〉 〈 〉

〈 〉 〈 ̅ 〉 〈 〉

The latter higher-energy shear only occurs at elevated temperatures. In this work we have not observed the formation of a/2 <110> partial dislocations at an early stage of radiation- induced growth at any of the studied irradiation temperatures. An in-situ TEM study from literature on the structure evolution of bcc Fe during electron irradiation and heating revealed the direct transformation of a/2<111> loops into a<100> loops, without the coalescence of the a/2<111> loop with an external loop. The process was proposed to involve the nucleation and propagation of a proper shear loop, triggered either by thermal fluctuations at high temperature or by the presence in the vicinity of an external loop acting as a source of a

171

significant shear stress (Arawaka et al., 2006). Recently molecular dynamics simulations pointed at the formation of a dumbbell configuration of SIAs, that evolves into a configuration of parallel crowdions on nearest neighbour <111> lines when the number of SIAs is 7, and subsequently into a/2<111> loops (Chen et al., 2013). Those a/2<111> loops can directly evolve into <100> loops at high temperatures by the proposed mechanism of gliding of <111> crowdions and jumping between different <111> directions, rotating into <100> orientation, and gliding of segments of {100} loops along <100>.The estimated energy barrier of a four SIA cluster jump from a/2[111](211) to a <100>{100}loop configuration was estimated to be 1.2 eV (Chen et al., 2013). Furthermore, in the less likely event that mobile a/2<111> loops with similar size and shape encounter themselves at high temperature, their interaction can lead to the formation of a<100> dislocation loops, as predicted by kinetic Monte Carlo simulations for bcc metals under irradiation (Xu et al., 2013). The resultant a<100> type dislocations would be glissile or sessile depending on the slip planes of the reacting dislocations (Spitzig et al., 1966), but the shear stress for gliding a<100> type dislocations is larger than for a/2<111> type dislocations (Hull et al., 2011). Recent molecular dynamic simulations on bcc metals yielded a value of 2eV for the one- dimensional migration energy of a<100> loops when the number of SIAs is 24 (Chen et al., 2013). Furthermore, an increase in damage level from 0.15 to 0.55dpa causes a reduction in the loop number density at the three irradiation temperatures, coupled with an increase in the average loop dimensions; see Figure 6.3 and Table 6.2. At 345C and 0.55dpa there is a relatively high density of dislocation loops presented simultaneously with the developed dislocation tangles (Figure 6.2). As the proton irradiation proceeds, the dislocation loops are in continuous movement (Norris, 1972) scavenging additional SIAs from the tantalum matrix, and some of those growing loops also coalesce.

172

Figure 6.5 Bright-field TEM imaging of interstitial-type dislocation loops in tantalum proton irradiated at: left – 180(2)°C; right – 590(5)°C, shown for different two-beam conditions. These TEM data was used for the dislocation analysis using the g.b=0 invisibility criterion, where g denotes the scattering vector and b the Burgers vector, and revealed the a/2<111> (top – 180(2)C) and a<001> (bottom – 590(5)C) Burgers vector.

Table 6.2 Principal characteristics of the radiation-induced dislocation loops in tantalum.

Number Temperature Damage Burgers Length Width o density ( C) level (dpa) vector (nm) (nm) 21 -3 (10 m ) 0.15 a/2 4.3(0.5) 3.1(0.4) 110 180(2) 0.55 <111> 8.6(0.4) 6.5(0.7) 61

0.15 a/2 9.2(1.0) 6.4(1.3) 40 345(3) 0.55 <111> 16.4(1.7) 11.7(1.8) 18 0.15 18.8(1.4) 14.8(1.4) 13 590(5) a <001> 3 0.55 37.8(3.2) 23.3(2.4)

Void formation Figure 6.6 shows the bright-field imaging of radiation-induced voids. The microstructure at 345(3)C and a damage level of 0.15dpa contains a high density of radiation-induced voids that are randomly distributed in the matrix. However, at 0.55dpa voids are arranged in a bcc lattice, oriented parallel to the underlying (111) bcc lattice plane of the Ta matrix and with an average void distance of 7.3(2)nm. The average void diameter increases from 1.2(1)nm

173

(0.15dpa) to 2.2(1)nm (0.55dpa), and the number density from 2.3×1023m-3 (0.15dpa) to 3.1×1023m-3 (0.55dpa).

Figure 6.6 TEM BF image showing the presence of radiation-induced voids in tantalum at 345(3)°C, and their absence at 180(2)°C and 590(5)°C.

The SIA trapping at mobile a/2<111> dislocation loops leaves a vacancy excess in the matrix that evolves into voids. A void disorderorder transition occurs with increasing damage level at 345(3)C. Existing models propose that non-aligned voids shrink and collapse, and consequently void ordering develops. This process is based on the anisotropic diffusion of SIAs or small interstitial dislocation loops along closed packed directions or planes (Jäger et al., 1993; Semenov et al., 2006; Semenov et al., 2008; Barashev et al., 2010) or on the anisotropic energy transfer provided by long propagating discrete breeders (Dubinko et al., 2009; Dubinko et al., 2011; Murzaev et al., 2015). In contrast, the TEM data for tantalum irradiated at 180(2)C reveal the absence of voids in the microstructure. At this lower temperature, interstitial dislocation loops are still formed and generate a vacancy excess in the matrix. However, the vacancy mobility is significantly reduced (Johnson , 1960; Satta et al., 1999) and, together with a number of vacancy-SIA recombination events, does not lead to enough vacancy clustering so that voids, if present, cannot be observed by TEM. Voids are also not observed in the sample irradiated at 590(5)C. At this temperature a<100>

174

dislocation loops with a lower mobility predominate in the microstructure. Dislocation mobility is needed to induce a local vacancy excess, and void nucleation would take place in the vicinity of dislocations (Kitajima et al., 1979). The upper temperature limit for void formation, which depends on particle fluence, is related to the reduction in the supersaturation of vacancies due to an increased thermal-equilibrium concentration of vacancies (Norris, 1972) and also, as observed in this study, to a change in the type of predominant dislocation loops that act as scavengers for self-interstitial atoms in the matrix.

The change in yield stress ( of the material caused by irradiation-induced defects, such as dislocation loops and voids, can be derived using the dispersed barrier model (Taylor, 1934; Seeger, 1958) :

√ (1) where  is the barrier strength coefficient, M the Taylor factor (2.7),  the shear modulus of tantalum (69GPa), b the Burgers vector of the dislocations (2.76Å), N the density and d the average size of the obstacles (Seeger, 1958; Rosenberg (1971)). The value of  depends on the defect type and size for a given temperature (Hu et al., 2016). A value of  = 0.2 has been reported for dislocations in tantalum (Yasunaga et al., 2000), whereas it takes a value of  = 0.25 in the case of voids with a diameter of 1-2nm (Hu et al., 2016). The change in hardness ( corresponds to:

(2) where the correlation factor K  3 and both and are given in Pa (Cahoon et al., 1971; Busby et al., 2005). Using the values for N and d for the a/2 ⟨111⟩ interstitial dislocation loops and for the voids observed at damage level of 0.15 dpa and at 350°C, we obtain a value of (loops) and (voids) respectively.

Conclusions Proton irradiation of tantalum causes the formation of interstitial-type dislocation loops whose size increases with temperature and damage level, and concomitantly the loop number density reduces. At 345(3)C the presence of mobile a/2<111> dislocation loops induced by radiation results in a vacancy excess in the matrix that evolves into randomly distributed voids at 0.15dpa. At the higher damage level of 0.55dpa, those voids are ordered into a bcc lattice with an average void distance of 7.3(2) nm along the (111) plane. Voids are not observed at either 180(2) or 590(5)C due to the low mobility of vacancies or a<100> dislocation loops, respectively.

175

Acknowledgments The work described was supported by the Dalton Cumbrian Facility Project, a joint facility of The University of Manchester and the Nuclear Decommissioning Authority.

References Arawaka, K., Hatanaka, M., Kuramoto, E., Ono, K., Mori H. (2006) Changes in the Burgers Vector of Perfect Dislocation Loops without Contact with the External Dislocations. Phys. Rev. Lett. 96,125506.

ASTM E521-96 (2009) Standard Practice for Neutron Radiation Damage Simulation by Charged-particle Irradiation.

Azevedo, C.R.F. (2011) A review on neutron-irradiation-induced hardening of metallic components. Eng. Fail. Anal. 18, 1921-1942.

Barashev, A.V., Golubov, S.I. (2010) On the onset of void ordering in metals under neutron or heavy-ion irradiation. Phil. Mag. 90, 1787-1797.

Busby, J.T., Hash, M.C., Was, G.S. (2005) The relationship between hardness and yield stress in irradiated austenitic and ferritic steels. J. Nucl. Mater. 336, 267-278.

Cahoon, J.R., Broughton, W.H., Kutzak, A.R. (1971) The determination of yield strength from hardness measurements. Metall. Trans. 2, 1979-1983.

Cawthrone, C., Fulton, E.J. (1967) Voids in Irradiated Stainless Steel. Nature 216, 575-576.

Chen, J., Bauer, G.S., Broome, T., Carsughi, F., Dai, Y., Maloy, S.A., Roedig, M., Sommer, W.F., et al. (2003) Summary of the results from post-irradiation examination of spent targets at the FZ-Juelich. J. Nucl. Mater. 318, 56-69.

Chen J., Gao N., Jung P., Sauvage T. (2013) A new mechanism of loop formation and transformation in bcc iron without dislocation reaction. J. Nucl. Mater. 441, 216-221.

Dubinko, V.I., Guglya, A.G., Melnichenko, E., Vasilenko, R. (2009) Radiation-induced reduction in the void swelling. J. Nucl. Mater. 385, 228-230.

176

Dubinko, V.I., Guglya, A.G., Donnelly, S.E. (2011) Radiation-induced formation, annealing and ordering of voids in crystals: Theory and experiment. Nucl. Instrum. Methods Phys. Res. B 269, 1634-1639.

Evans, J.H., Foreman, A.J.E., (1985) Some implications of anisotropic self-interstitials diffusion on void swelling in metals. J. Nucl. Mater. 137, 1-6.

Eyre, B. L., Bullough, R. (1965) On the formation of interstitial loops in b.c.c. metals. Phil. Mag. 12, 31-39.

Hull, D., Bacon, D.J. (2011) Introduction to dislocations. Butterworth-Heinemann.

Harte, A., J ̈dern ̈s, D., Topping M., Frankel P., Race C.P., Romero, J., Hallstadius, L., Darby, E.C., Preuss, M. (2017) The effect of matrix chemistry on dislocation evolution in an irradiated Zr alloy. Acta Mater. 130, 69-82.

Hu, X., Koyanagi, T., Fukuda, M., Kiran Kumar, N.A.P., Snead, L.L., Wirth, B.D., Katoh, Y. (2016) Irradiation hardening of pure tungsten exposed to neutron irradiation. J. Nucl. Mater. 480, 235-243.

Jäger, W., Trinkaus, H. (1993) Defect ordering in metals under irradiation. J. Nucl. Mater. 205, 394-410.

Jenkins, M.L., Kirk, M.A. (2001) Characterization of Radiation Damage by Transmission Electron Microscopy, IOP Publishing Ltd.

Johnson, A.A. (1960) Deformation, fracture and radiation damage, in body-centred cubic transition metals. J. Less-Common Metals 2, 241-252.

Kitajima, K., Kutagami, K., Kuramoto, E. (1979) Nucleation of voids in bcc metals. J. Nucl. Mater. 85-86, 725-729.

Mazey, D.J. (1990) Fundamental aspects of high-energy ion-beam simulation techniques and their relevance to fusion materials studies. J. Nucl. Mater. 174, 196-209.

Murzaev, R.T., Kistanov, A.A., Dubinko, V.I., Terentyev, D.A., Dmitriev, S.V. (2015) Moving discrete breathers in bcc metals V, Fe and W. Comput. Mater. Sci. 98, 88-92.

Nelson, R.S., Mazey, D.J., Hudson, J.A. (1970) The use of ion accelerators to simulate fast neutron-induced voidage in metals. J. Nucl. Mater. 37, 1-12.

177

Nelson, A.T., O’Toole, J.A., Vacilenti, R.A., Maloy, S.A. (2012) Fabrication of a tantalum- clad tungsten target for LANSCE. J. Nucl. Mater. 431, 172-184.

Norris, D.I.R. (1971) The use of the high-voltage electron microscope to simulate fast neutron-induced void swelling in metals. J. Nucl. Mater. 40, 66-76.

Norris, D.I.R. (1972) Voids in irradiated metals (Part I). Radiat. Eff. 14, 1-37.

Novakowski, T. J., Tripathi, J. K., Hassanein, A. (2016) Effect of high-flux, low-energy He+ ion irradiation on Ta as a plasma-facing material. Sci. Rep. 6, 39746.

Okamoto, P.R., Levine, E., Thomas, G. (1966) Kikuchi maps for hcp and bcc crystals. Appl. Phys. 38, 289-296

Odette, G.R., Alinger, M.J., Wirth B.D. (2008) Recent Developments in Irradiation-Resistant Steels. Annu. Rev. Mater. Res. 38, 471-503.

Rosenberg, J.M., Piehler, H. R. (1971) Calculation of the Taylor factor and lattice rotations for bcc metals deforming by pencil glide. Metall. Trans. 2, 257-259.

Russell, K.C., Powell R.W. (1973) Dislocation loop nucleation in irradiated metals. Acta Metall. 21,187-193.

Satta, A., Willaime, F., de Gironcoli, S. (1999) First-principles study of vacancy formation and migration energies in tantalum. Phys. Rev. B 60, 7001.

Seeger, A.K. (1958) On the theory of radiation damage and radiation hardening. Proceedings of the Second United Nations International Conference on the Peaceful Uses of Atomic Energy 6, 250-273.

Semenov, A. A., Woo, C.H. (2006) Void lattice formation as a nonequilibrium phase transition. Phys. Rev B 74, 024108.

Semenov, A.A., Woo, C.H., Frank, W. (2008) Diffusion anisotropy and void development under cascade irradiation. Appl. Phys. A 93, 365-377.

Singh, B.N., Evans, J.H. (1995) Significant differences in defect accumulation behaviour between fcc and bcc crystals under cascade damage conditions. J. Nucl. Mater. 226, 277-285.

178

Spitzig, W.A., Mitchell, T.E. (1966) Dislocation arrangements in tantalum single crystals deformed in tension at 373°K. Acta Metall. 14, 1311-1323.

Stork, D., Agostini, P., Boutard, J.L., Buckthorpe, D., Diegele, E., Dudarev, S.L., English, C., Federici, G., et al. (2014) Developing structural, high-heat flux and plasma facing materials for a near-term DEMO fusion power plant: The EU assessment. J. Nucl. Mater. 455, 277-291.

Taylor, G.I. (1934) The Mechanism of Plastic Deformation of Crystals. Part I. Theoretical. Proc. Roy. Soc. A 145, 362-387.

Trinkaus, H., Singh, B.N., Foreman, A.J.E. (1993) Impact of glissile interstitial loop production in cascades on defect accumulation in the transient. J. Nucl. Mater. 206, 200-211.

Urban, K. (1971) Growth of Defect Clusters in Thin Nickel Foils during Electron Irradiation (I). Phys. Stat. Sol. (a) 4 761-772.

Urban, K., Wilkens, M. (1971) Growth of Defect Clusters in Thin Nickel Foils during Electron Irradiation. II. Temperature Dependcnee of the Growth Rate of Interstitial Loops. Phys. Stat. Sol. (a) 6 173-185.

Wady, P.T., Draude, A., Shubeita, S.M., Smith, A.D., Mason, N., Pimblott, S.M., Jimenez- Melero, E. (2016) Accelerated radiation damage test facility using a 5MV tandem ion accelerator. Nucl. Instr. Meth. Phys. Res. A 806, 109-116.

Was, G. (2007) Fundamentals of Radiation Materials Science. Springer-Verlag, 2007.

Wiffen, F.W. (1977) The microstructure and swelling of neutron irradiated tantalum. J. Nucl. Mater. 67, 119-130.

Xu, H., Stoller R.E., Osetsky Y.N., Terentyev D. (2013) Solving the Puzzle of ⟨100⟩ Interstitial Loop Formation in bcc Iron. Phys. Rev. Lett. 110, 265503.

Yao, Z., Xu, S., Jenkins, M. L., Kirk, M. A. (2008) Preparation of TEM samples of ferritic alloys. J. Electron Microsc. 57(3), 91–94.

Yasunaga, K., Watanabe, H., Yoshida, N., Muroga, T., Noda N. (2000) Correlation between defect structures and hardness in tantalum irradiated by heavy ions. J. Nucl. Mater. 283-287, 179-182.

179

Yi, X., Jenkins, M. L., Hattar, K., Edmondson, P. D., Roberts, S. G. (2015) Characterisation of radiation damage in W and W-based alloys from 2MeV self-ion near-bulk implantations. Acta Mater. 92 163-177.

Yvon, P., Carré F. (2009) Structural materials challenges for advanced reactor systems. J. Nucl. Mater. 385, 217-222.

Ziegler, J.F., Ziegler, M.D., Biersack, J.P. (2010) SRIM – The stopping and range of ions in matter. Nucl. Instr. Meth. Phys. Res. B 268, 1818-1823.

Zinkle, S.J., Was, G.S. (2013) Materials challenges in nuclear energy. Acta Mater. 61, 735- 758.

180

7. Radiation-induced void formation and ordering in Ta-W alloys

I. Ipatovaa,b,*, P.T. Wadyb, S.M. Shubeitab, C. Barcellinia, A. Impagnatielloa, E. Jimenez-Meleroa

a School of Materials, The University of Manchester, Manchester M13 9PL, UK bDalton Cumbrian Facility, The University of Manchester, Moor Row CA24 3HA, UK

Published in Journal of Nuclear Materials

DOI 10.1016/j.jnucmat.2017.08.029

Corresponding author (*): University of Manchester School of Materials Oxford Road Manchester M13 9PL United Kingdom Tel.: +44 7849290480 Email: [email protected] 181

Abstract We have assessed the formation and evolution of void and dislocation arrrangements in Ta, Ta-5wt.%W (Ta-5W) and Ta-10wt.%W (Ta-10W) as a function of radiation level at a temperature of 345±3C, by combining proton irradiation experiments, transmission electron microscopy and nano-hardness measurments. The damaged structure of tantalum at 0.1dpa is characterized by the presence of a/2 ⟨111⟩ interstitial dislocation loops and randomly distributed voids, whereas only dislocation loops are observed in the two alloys. Void ordering occurs in tantalum at 0.25dpa, together with the apperance of dense dislocation tangles. A further increase in damage level leads to a continuous nucleation and growth of voids, and saturation is not attained at a damage level of 1.55dpa. In contrast, the average size and number density of dislocation loops increases gradually with damage level in the two alloys, and voids only form at 1.55dpa. Tungsten delays the loop evolution and therefore the formation of radiation-induced voids. Keywords: refractory metals, void lattice, dislocation analysis, proton irradiation, transmission electron microscopy

182

Introduction Tantalum and its alloys are targeted for structural applications in service environments that combine elevated temperatures and high neutron or ion fluxes. These materials offer a very appealing combination of high melting point, toughness, fabricability, radiation tolerance and compatibility with liquid alkali metals. They are, for instance, regarded as advanced high-Z plasma-facing material candidates for future tokamak and other advanced nuclear fusion designs [1]. Recent studies have demonstrated that tantalum exhibits a higher fluence threshold for He+ ion-induced surface nano-structuring, as compared to tungsten, and therefore a lower risk of compromising the material integrity and of contaminating the reactor plasma [2]. Ta-based alloys broaden the materials’ palette for fusion technology development, by potentially being used as a full-thickness structural facing material or as a coating of first wall reactor components. Furthermore, tungsten targets used at international neutron spallation sources, such as ISIS in England, LANSCE in United States or KENS in Japan, are cladded with a thin layer of tantalum to improve the resistance to water corrosion of the target under irradiation [3, 4]. Tantalum offers the advantage of retaining its ductility after proton irradiation up to a dose of 11 dpa at temperatures lower than 200C, and as a heavy atomic nucleus helps to maximise the neutron yield of the target [5]. Ta-W alloys are of particular interest due to their demonstrated high performance at elevated temperatures [6, 7] and high deformation rates [8]. Body-centered cubic (bcc) alloys, such as those belonging to the Ta-W system, are particularly attractive for structural components that will experience unprecedented levels of radiation damage in foreseen nuclear reactor environments. The attractiveness of these alloys lies to a great extent on their resistance to radiation-induced defect accumulation in the form of clusters, dislocation loops and especially voids, and therefore on their swelling resistance. Molecular dynamics simulations, supported by limited heavy ion irradiations, have revealed the production of a relatively small number of vacancy clusters and self-interstitial atom (SIA) clusters in bcc metals, as compared to face-centered cubic (fcc) metals, and also the loss of small mobile SIA clusters to sinks such as dislocations or grain boundaries [9]. Despite their relatively high resistance to void swelling of bcc materials, the formation of radiation-induced planar defect agglomerates such as voids and loops, and their evolution into three-dimensional defect agglomerates or voids and potentially void ordering, should be systematically assessed and predicted reliably in bcc metallic materials for their use in future nuclear reactor systems. Special emphasis is to be placed on the temperature and dose range

183

for those defect agglomerates to form. Swelling is traditionally interpreted in terms of the preferential absorption of SIAs to dislocation loops (dislocation bias) [10, 11] or, mainly in fcc metals, based on the dissociation of vacancies from clusters and their diffusion in the matrix (production bias) [12, 13]. The main aim of this work is to investigate the radiation-induced formation of dislocation loops and defect clusters in selected Ta-W alloys, and their evolution into void arrangements, as a function of radiation dose and tungsten content in the alloy. The existing knowledge based on experimental data about the nature and evolution of radiation-induced lattice defects in Ta-W alloys, and their relationship to hardening and swelling phenomena, is still very limited. Void-induced swelling data available from immersion density measurements and experiment-oriented modelling revealed the occurrence of void swelling in tantalum at temperatures of 200C or higher, with a peak swelling at 635C [14]. Transmission electron microscopy (TEM) data of tantalum irradiated with neutrons at a temperature of 585C and up to a neutron fluence of 2.5×1022 neutron/cm2 revealed the appearance of voids with an average diameter of 6.1nm, ordered on a bcc superlattice with a lattice parameter of 20.5nm and 80% occupancy of superlattice points. However, voids do not order at higher irradiation temperatures up to 1050C, whereas at 425C the presence of voids was not detected by TEM below the detection limit at that time of 2.5nm [15]. Furthermore, radiation-induced hardening in tantalum as well as in Ta-W alloys is reported in neutron-irradiated specimens at temperatures 350°C and radiation doses of ≤ 0.14 dpa based on mechanical testing data [16, 17]. In parallel to neutron irradiation campaigns, ion accelerators have been used successfully to simulate neutron-induced void formation in metals [18]. Tantalum specimens were irradiated with intense beams of Cu2+ or Ni+ at variable temperatures up to 1273C. Their analysis revealed the presence of radiation-induced vacancy loops that form preferentially inside large interstitial loops, and increase in size and number density with radiation dose [19]. Evidence of void formation from vacancy loops is reported at temperatures higher than 400C, with the damage level of the irradiation varied from 0.03 to 3 dpa at its peak position of damage [20]. The peak void swelling value of 8.2% is reported to occur at 1135C for a dose of 20dpa [18]. Void ordering in tantalum irradiated with Ni+ ions requires temperatures higher than 847C [21]. Neutron/ion irradiation experiments on other bcc refractory metals, namely Nb, Mo and W, revealed the existence of a damage threshold for void ordering [22-24]. In this study, we have used a proton beam as a surrogate of neutron damage, and assessed the effect of dose and tungsten content on the damage evolution at 184

345C, which corresponds to ~0.11Tm, where Tm denotes the melting point of tantalum. This temperature lies close to the lower temperature limit for void swelling in tantalum [9, 20]. The maximum W content used was 10wt.%, since Ta-W alloys with higher amounts exhibit a strong tendency to form a CsCl-type B2 superstructure [25, 26].

Experimental 1mm-thick plate material of Ta, Ta-5wt.%W (Ta-5W) and Ta-10wt.%W (Ta-10W) was annealed for recrystallization at 1400C in high vacuum (10-5Torr) for <10h. The average grain size after annealing was 50-60m. Specimens with dimensions of 15×15×1 mm3 were cut from the annealed material, mechanically ground and polished to a final surface finish using 0.3m alumina suspension. The specimens were irradiated by a 3MeV proton beam generated using a 5MV Tandem Pelletron ion accelerator and a high-current TORVIS source installed at the Dalton Cumbrian Facility of the University of Manchester [27]. Several equivalent irradiation experiments were performed, in order to generate enough irradiated material to obtain samples at selected damage levels up to 1.55 dpa for analysis. The beam was rastered over a sample area <12×12 mm2, with a frequency of 517 and 64 Hz in the x- and y-directions respectively. A summary of the irradiation conditions for the studied samples is presented in Table 7.1. Figure 7.1 shows details of the irradiation beam line, sample stage and temperature distribution during irradiation. For each experiment, samples were mounted on the sample stage equipped with heating and water cooling capabilities. The temperature uniformity was monitored during the experiment using a non-contact IR camera located at 30 with respect to the sample surface. In addition, thermocouples were spot welded on the samples outside the irradiated area. Liquid indium was used to improve the thermal contact between the samples and the NIMONIC75® alloy block of the sample stage. The accumulated charge on the samples was recorded during irradiation, in order to be able to determine the damage level achieved at the end of the experiment. Nanoindentation measurements on the cross section of the irradiation specimens were carried out using a NanoIndenter-XP (MTS Systems Corp.). A Berkovich-type diamond tip and the loading depth of 200 nm were used. The indenter tip geometry was evaluated using a silica standard. The hardness estimate for each damage level considered corresponds to the average of 25 data points around the region from where TEM discs were extracted for analysis.

185

Figure 7.1 (a) Ion accelerator beam line used for the proton irradiation experiments of Ta-W alloys; (b) main components of the sample stage; (c) top view of the stage with four samples mounted simultaneously and clamped with a tantalum shim; (d) emissivity-corrected temperature distribution over the four samples under proton irradiation, recorded on a thermal camera mounted at a 30 angle with respect to the sample surface; (e) histogram of temperature distribution corresponding to 24 hours of continuous irradiation.

Table 7.1 Summary of irradiation conditions used for the Ta-W samples in this study.

186

The damaged structures were characterised using the FEI Quanta 250 FEG Scanning Electron microscope operating at 30kV, and also the FEI Tecnai T20 Transmission Electron microscope with LaB6 crystal, equipped with a double-tilt specimen holder and operating at 200kV. 3mm-diameter discs from selected regions of the irradiated layer of the samples were prepared by mechanical pre-thinning, followed by electro-polishing using the Struers TenuPol-5 unit and an electrolyte containing 15vol.% sulphuric acid (95%) – 85vol.% methanol at a temperature of 5C. The determination of the Burgers vector b of the observed dislocations made use of the g.b=0 invisibility criterion, where g denotes the scattering vector. The foil thickness was derived from the fringes spacing of the convergent beam electron diffraction pattern, and was used to obtain the dislocation density at each selected radiation dose. Bright-field imaging of small radiation-induced voids was based on the “out- of-focus” imaging technique, where voids appear as white dots surrounded by black Fresnel fringes when recorded in an under-focused condition, and as dark dots with bright fringes in an over-focused condition [28]. During void imaging, the amount of defocus was Δf ≤ 0.6µm, which falls close to the defocus value of ~1µm reported for observing voids of 1nm in diameter [28].

Results Thickness of irradiated layer The thickness of the proton irradiated layer of each sample was determined from both the Back Scattered Electron (BSE) image and the variation in nano-hardness of the sample cross section, see Figure 7.2. The BSE image shows a relatively bright line at 32m from the irradiated surface in all samples, which coincides with the highest values in the hardness profile. The damage density caused by the 3MeV proton beam attains its highest value at the Bragg peak position. Consequently, the channeling of electrons is more impeded and the BSE signal increases. The damage profile was simulated using the SRIM software with the quick Kinchin–Pease approach [29-31], using an average displacement energy of 90 eV [32] and default values for other software settings, together with the total charge deposited on the sample during the irradiation experiment. The simulated profile confirmed experimentally for each irradiated sample was used to select regions in the irradiated layer from where TEM foils were prepared at representative damage levels.

187

Figure 7.2 (a) Backscattered electron image of the tantalum cross-section, after having been irradiated at 345±3°C using 3MeV proton beam; (b) simulated damage profile using the SRIM software with the quick Kinchin – Pease approach and the total current deposited on the sample during the irradiation experiment, together with the nano-hardness values along the cross section of the irradiated tantalum sample.

Irradiation-induced microstructural evolution Figure 7.3 shows the evolution of the microstructure of tantalum as a function of damage level. Its damaged microstructure is characterised by the presence and evolution of radiation- induced voids and dislocations with a Burgers vector of a/2 ⟨111⟩. At a damage level of 0.1 dpa, the microstructure of tantalum is characterized by the presence of interstitial dislocation loops with an average size of 7nm (length) and 5nm (width) and a density of ~5×1022m-3, together with a relative high density of approximately spherical voids that are randomly distributed in the tantalum matrix. A further increase in the damage level to 0.25dpa causes the evolution of the dislocation loops into tangles that become denser as we increase the damage level up to 1.55dpa. Simultaneously with the occurrence of dislocation tangles, the radiation-induced voids present at 0.1dpa become ordered into a bcc lattice, oriented parallel to the underlying (111) bcc lattice plane of the Ta matrix. The variation of the main characteristics of the void arrangement, namely the void number density, average diameter and void-void distance is shown in Figure 7.4. The void number density (Figure 7.4a) remains relatively constant between 0.1 and 0.25dpa, and increases continuously at higher damage levels, while the average void-void distance (Figure 7.4c) gradually decreases. 188

Moreover, the average void size (Figure 7.4b), that reflects the nucleation of voids and the growth of existing voids, also increases with damage level and seems to gradually saturate at the highest damage levels.

Figure 7.3 Formation and evolution of void (top row) and dislocation (middle and bottom row) structures in Ta irradiated at 345±3ºC using a 3 MeV proton beam. All data were obtained using the [111] zone axis.

189

Figure 7.4 Variation with damage level of (a) void size. (b) number density and (c) average void lattice distance in Ta, proton irradiated at 345±3C. The inset in (b) corresponds to the number of vacancies per void as a function of damage level (see text). The stripped region denotes the region where the voids are not ordered into a bcc lattice.

The addition of tungsten modifies the characteristics of the dislocation structure and its evolution with increasing damage levels, see Figure 7.5. The microstructure of both Ta-5W and Ta-10W alloys contains a number of interstitial dislocation loops at 0.25dpa. The change in size and number density of those loops with damage level is shown in Figure 7.6. In both alloys, the loop dimensions and number density increase with damage level. However, we have not observed the transition to dislocation tangles with increasing damage level in any of the two alloys. An increased amount of tungsten in the alloy reduces the loop size and density, but does not prevent the loop evolution with increasing damage level. In contrast with the results in tantalum, the appearance of radiation-induced voids has only been observed at the highest damage level of 1.55dpa and they are randomly distributed in the tantalum matrix (see Figure 7.7).

190

Figure 7.5 Evolution of dislocation structures in Ta-W alloys at 345±3C with damage level.

Figure 7.6 Dislocation loop size and number density as a function of damage level in Ta-W alloys irradiated at 345±3C. 191

Figure 7.7 Radiation-induced void formation in Ta-10W alloy, proton irradiated at 345±3°C to a damage level of 1.55 dpa.

Changes in hardness The hardness of the samples was measured in the non-irradiated condition, and also in the cross section of the irradiated layer of the samples, as a function of position with respect to the surface exposed to the incident proton beam. In the non-irradiated state, the addition of tungsten causes a small increase in hardness from 2.76(0.33) GPa for Ta to 3.07(0.17) GPa for Ta-10W. Figure 7.8 displays the changes in hardness for the three materials as a function of damage level. The hardness value in the three materials increases with the damage level, and does not reach saturation at the highest level of 1.55dpa. The largest change in hardness corresponds to tantalum, in good agreement with the observed radiation-induced formation of high density dislocation tangles and void arrangements. The Ta-10W alloy shows the lowest hardness increase during proton irradiation of the three materials.

192

Figure 7.8 Radiation-induced change in hardness (ΔHirr.) observed in Ta-W alloys at 345±3°C with increasing damage level.

Discussion The damaged structure in bcc metals, after the spike phase of the displacement cascades, is characterized by a relatively large fraction of single (unclustered) lattice defects, both vacancies and SIAs, together with a number of small mobile SIAs clusters [9]. The evolution of those lattice defects proceeds by the appearance and growth of interstitial dislocation loops, as a consequence of the fast diffusion and clustering of SIAs. The resultant vacancy excess in the lattice can lead to the nucleation and growth of voids during irradiation. However, an incubation period is expected to be required before any damage in the form of interstitial dislocation loops or voids is observed experimentally [33]. Our results reveal the presence of radiation-induced interstitial dislocation loops in the three materials at a lower damage level of 0.1dpa, but the presence of voids is only observed at that low level in the case of tantalum (Figure 7.3). In Ta-5W and Ta-10W, the average loop dimensions and number density increases gradually in the absence of voids, see Figure 7.5 and 7.6. This increase in both loop parameters is more pronounced in the Ta-5W alloy. This fact points to the role of solute W atoms in delaying SIA diffusion and/or loop mobility via a solute drag effect [34, 35]. The formation and growth of voids is expected to be favored when there are

193

dislocation loops in their neighborhood to act as effective sinks for SIAs. A critical loop density of 6 × 1022 m-3 seems to be required for voids to nucleate and grow in this case. The occurrence of voids in W-doped materials is observed at higher damage levels (1.55dpa), as compared to tantalum (0.1dpa), and those voids are randomly distributed in the matrix (Figure 7.7). It is worth mentioning that the use of a focused proton beam, instead of the approach adopted in this study of rastering the beam over the sample, is in general expected to accelerate the formation of voids [36, 37]. In the three materials, the voids do not present any faceted shape. Void embryos are expected to be spherical, and evolve into a faceted morphology during void growth with increasing damage level or irradiation temperature [38]. This morphological transition is predominantly governed by the anisotropy of the surface energy, based on a recent phase-field void growth model [39]. A transition from spherical to cube-octahedral shape has been observed in voids induced by Cu2+ irradiation in tantalum at damage levels 3dpa and temperatures 1135C [19]. Therefore our study probes the early stages of void formation in these materials. The further evolution of the dislocation and void structures in Ta-5W and Ta-10W would occur at damage levels higher than 1.55dpa at a temperature of 345±3C, since W seems to delay to formation and evolution of damaged structures in tantalum under irradiation. However, we can monitor the evolution of dislocation loops and random voids in unalloyed tantalum at lower damage levels. An increase in the damage level from 0.1 to 0.25dpa in tantalum induces the ordering of voids into a bcc lattice isomorphous with the crystal structure of the surrounding tantalum matrix, together with the collapse of the dislocation loops to form a relatively dense dislocation tangle. The void ordering along the (111) plane is shown in the top row of Fig 3, whereas the evolution of the main void lattice parameters with damage level is displayed in Figure 7.4. In the damage level range of 0.1-0.25dpa, the void number density remains constant whereas the average size increases. Existing models of void lattice formation are based on the anisotropic diffusion of SIAs or small interstitial dislocation loops along closed packed directions or planes [40-43], or on the anisotropic energy transfer provided by long propagating discrete breeders [44-46]. In both cases, the result is the shrinkage and dissolution of nonaligned voids, and consequently the appearance of void ordering. Our data indicates that nucleation of voids does not occur during the disorderorder transition, but the ordered voids that survive the process do increase in size. These TEM data are unique in revealing the early stages of formation and ordering of voids with a diameter <3nm, since the previously reported void lattice in tantalum irradiated with

194

neutrons at 585°C already corresponded to an average void diameter of 6.1nm [15]. Once the void lattice is established, nucleation of additional voids occurs progressively at the relatively low temperature of 345±3C, as revealed by the increase in the void number density with the damage level. As a consequence, the average void distance along the (111) lattice plane decreases with damage level (Figure 7.4c). The existing voids simultaneously grow, and this leads to a progressive increase in the average void diameter that seems to slowly saturate at the higher damage levels (Figure 7.4c). We have estimated the average number of vacancies per void (n), assuming a spherical shape of the voids, according to [47]:

(1)

where denotes the average void radius and the atomic volume. The values obtained for n at variable damage levels are collected in the inset of Figure 7.4b. The ratio between the void lattice parameter and the average void radius varies from 13.2 (0.25dpa) to 8.4 (0.85dpa) and 4.5 (1.55dpa). The number density of voids in tantalum continues to increase with the damage level after the formation of the bcc void lattice, and does not reach saturation even at the highest damage level of 1.55dpa. In fact, it turns out that once the void lattice forms, the nucleation rate increases, as revealed by the slope in Figure 7.4a. Void nucleation and growth occur simultaneously in tantalum at the temperature of 345±3C once the void lattice forms. Voids act as effective sinks for vacancies, and each void is expected to have a sphere of influence around it that contains a vacancy gradient, since inside that sphere of influence the vacancies tend to become trapped by the void. Once the soft impingement of the sphere of influence of neighboring voids occurs, only void growth would be expected to occur [33]. The formation of void lattice seems therefore to retard that impingement, so that new voids can still nucleate. This evolution of the damaged structure of tantalum also manifests itself in a lack of saturation in the change in hardness with the damage level (Figure 7.8). The change in yield stress ( due to the presence of new dislocations can be estimated using the dispersed barrier model according to [48, 49]:

√ (2) where  is the barrier strength coefficient, M the Taylor factor (2.7),  the shear modulus of tantalum (69GPa), b the Burgers vector of the dislocations (2.76Å), N the density and d the average size of the obstacles [49, 50]. Eq. (2) can be used for relatively strong obstacles, i.e.  0.25 [51, 52]. The value of  depends on the defect type and size for a given temperature

195

[52]. A value of  = 0.2 has been reported for dislocations in tantalum [20], whereas it takes a value of  = 0.25 in the case of voids with a diameter of 1-2nm [52]. The change in hardness corresponds to , where the correlation factor K  3 and both and are given in Pa [53, 54]. Using the values for N and d for the a/2 ⟨111⟩ interstitial dislocation loops and for the voids from this work for tantalum at a damage level of 0.1dpa, we obtain a value of (loops) and (voids) respectively. The experimentally determined hardening value for tantalum at 0.1 dpa amounts to . Furthermore, the change in hardness due to irradiation is lower in the two W-doped materials, as compared to tantalum, and decreases with increasing W content. These observations can be attributed to the formation and evolution of voids and dislocation structures in tantalum, and to the lower dislocation loop density in the alloys with increasing W content.

Conclusions The structural characterization of Ta, Ta-5W and Ta-10W materials, after having been proton irradiated at 345±3C, have revealed the presence of a/2 ⟨111⟩ interstitial dislocation loops at a damage level of 0.1dpa in the three materials. Additionally, randomly distributed voids are observed in the microstructure of tantalum. An increase in damage level to 0.25dpa leads to the ordering of those voids into a bcc lattice isomorphous with the crystal structure of the surrounding Ta lattice, together with the appearance of a relatively dense dislocation tangle at the expense of the radition-induced dislocation loops. Void nucleation and growth takes place continuously in tantalum with increasing damage level from 0.25 to 1.55dpa, whereas the average void-void distance gradually decreases. The occurrence of void ordering seems to facilitate the formation of new voids, and saturation is not observed up to 1.55dpa. In contrast, voids are observed at 1.55dpa in W-containing samples, and they appear to be randomly distributed in the matrix. At lower damage levels, the average size and number density of dislocation loops increase progressively with damage level. A critical loop density of 6 × 1022 m-3 is required for voids to form in these materials. The presence of W seems to delay the evolution of the dislocation loop structure, by delaying self-interstitial atom diffusion and/or loop mobility via a solute drag effect, and therefore the overall lattice damage induced by radiation.

196

Acknowledgments The work described was supported by the Dalton Cumbrian Facility Project, a joint facility of The University of Manchester and the Nuclear Decommissioning Authority. We thank A.D. Smith and N. Mason for their assistance during the proton irradiation experiment, and also A. Forrest for his help to perform the nano-hardness measurements of the irradiated samples.

References 1. J.N. Brooks, L. El-Guebaly, A. Hassanein, T. Sizyuk, Nucl. Fusion 55 (2015) 043002. 2. T. J. Novakowski, J. K. Tripathi, A. Hassanein, Sci. Rep. 6 (2016) 39746. 3. P. Hosemann, S.A. Maloy, R.R. Greco, J.G. Swadener, T. Romero, J. Nucl. Mater. 384 (2009) 25. 4. A.T. Nelson , J.A. O’Toole, R.A. Vacilenti, S.A. Maloy, J. Nucl. Mater. 431 (2012) 172. 5. J. Chen, G.S. Bauer, T. Broome, F. Carsughi, Y. Dai, S.A. Maloy, et al., J. Nucl. Mater. 318 (2003) 56. 6. Z. Lin, E.J. Lavernia, F.A. Mohamed, Acta Mater. 47 (1999) 1181. 7. S. Nemat-Nasser, R. Kapoor, Int. J. Plasticity 17 (2001) 1351. 8. M.T Pérez-Prado, J.A. Hines, K.S. Vecchio, Acta Mater. 49 (2001) 2905. 9. B.N. Singh, J.H. Evans, J. Nucl. Mater. 226 (1995) 277. 10. J.H. Evans, A.J.E. Foreman, J. Nucl. Mater. 137 (1985) 1. 11. H. Trinkaus, B.N. Singh, A.J.E. Foreman, J. Nucl. Mater. 206 (1993) 200. 12. C.H. Woo, B.N. Sing, F.A. Garner, J. Nucl. Mater. 191-194 (1992) 1224. 13. B.N. Singh, A.J.E. Foreman, H. Trinkaus, Plasma Devices Oper. 3 (1994) 115. 14. J.F. Bates, A.L. Pitner, Nucl. Technol. 16 (1972) 406. 15. F.W. Wiffen, J. Nucl. Mater. 67 (1977) 119. 16. C.D. Barklay, D.P. Kramer, J. Talnagi, Investigation of Effects of Neutron Irradiation on Tantalum Alloys for Radioisotope Power System Applications, AIP Conference Proceedings. 880 (2007) 224. 17. T.S. Byun, S.A. Maloy, J. Nucl. Mater. 377 (2008) 72. 18. R.S. Nelson, D.J. Mazey, J.A. Hudson, J. Nucl. Mater. 37 (1970) 1. 19. K. Yasunaga, H. Watanabe, N. Yoshida, T. Muroga, N. Noda, J. Nucl. Mater. 258- 263 (1998) 879.

197

20. K. Yasunaga, H. Watanabe, N. Yoshida, T. Muroga, N. Noda, J. Nucl. Mater. 283- 287 (2000) 179. 21. B.A. Loomis, S.B. Gerber, J. Nucl. Mater. 71 (1978) 377. 22. J.H. Evans, Nature 229 (1971) 403. 23. V.K. Sikka, J. Moteff, J. Appl. Phys. 43 (1972) 4942. 24. B.A. Loomis, S.B. Gerber, A. Taylor, J. Nucl. Mater. 68 (1977) 19. 25. P.E.A. Turchi, A. Gonis, V. Drchal, J. Kudrnovský, Phys. Rev. B 64 (2001) 085112. 26. K. Masuda-Jindo, Vu Van Hung, N.T. Hoa, P.E.A. Turchi, J. Alloys Comp. 452 (2008) 127. 27. P.T. Wady, A. Draude, S.M. Shubeita, A.D. Smith, N. Mason, S.M.Pimblott, E. Jimenez-Melero, Nucl. Instr. Meth. Phys. Res. A 806 (2016) 109. 28. M.L. Jenkins, M.A. Kirk, Characterization of Radiation Damage by Transmission Electron Microscopy, IOP Publishing Ltd, 2001. 29. J.F. Ziegler, J. Appl. Phys. 85 (1999) 1249. 30. J.F. Ziegler, M.D. Ziegler, J.P. Biersack, Nucl. Instr. Meth. Phys. Res. B 268 (2010) 1818. 31. ASTM E521-96, Standard Practice for Neutron Radiation Damage Simulation by Charged-particle Irradiation, 2009. 32. M. Nastasi, J.W. Mayer, J.K. Hirvonen. Ion-solid Interactions: Fundamentals and applications. Cambridge University Press, 1996. 33. D.I.R. Norris, Radiat. Eff. 14 (1972) 1. 34. Y. Wang, D.J. Srolovitz, J.M. Rickman, R. LeSar, Acta Mater. 48 (2000) 2163. 35. M. Ovaska, T. Paananen, L. Laurson, M.J. Alava, J. Stat. Mech.-Theory Exp. 4 (2016) 043204. 36. E. Getto, Z. Jiao, A.M. Monterrosa, K. Sun, G.S. Was, J. Nucl. Mater. 465 (2015) 116. 37. K.E. Sickafus, E.A. Kotomin, B.P. Uberuaga, Radiation effects in solids, Springer, 2004. 38. C.W. Chen, Phys. Stat. Sol.(a) 16 (1973) 197. 39. W.B. Liu, N. Wang, Y.Z. Ji, P.C. Song, C. Zhang, Z.G. Yang, L.Q. Chen, J. Nucl. Mater. 479 (2016) 316. 40. W. Jäger, H. Trinkaus, J. Nucl. Mater. 205 (1993) 394. 41. A. A. Semenov, C.H. Woo, Phys. Rev B 74 (2006) 024108.

198

42. A.A. Semenov, C.H.Woo, W. Frank, Appl. Phys. A 93 (2008) 365. 43. A.V. Barashev, S.I. Golubov, Phil. Mag. 90 (2010) 1787. 44. V.I. Dubinko, A.G. Guglya, E. Melnichenko, R. Vasilenko, J. Nucl. Mater. 385 (2009) 228. 45. V.I.Dubinko, A.G. Guglya, S.E. Donnelly, Nucl. Instrum. Methods Phys. Res. B 269 (2011) 1634. 46. R.T. Murzaev, A.A. Kistanov, V.I. Dubinko, D.A. Terentyev, S.V. Dmitriev, Comput. Mater. Sci. 98 (2015) 88. 47. G. Was. Fundamentals of Radiation Materials Science. Springer-Verlag, 2007. 48. G.I. Taylor, Proc. Roy. Soc. A 145 (1934) 362. 49. A.K. Seeger, On the theory of radiation damage and radiation hardening, in: Proceedings of the Second United Nations International Conference on the Peaceful Uses of Atomic Energy, Geneva, September 1958, vol. 6, pp. 250– 273. United Nations, New York. 50. J.M. Rosenberg, H.R. Piehler, Metall. Trans. 2 (1971) 257. 51. S.J. Zinkle, Y. Matsukawa, J. Nucl. Mater. 329–333 (2004) 88. 52. X. Hu, T. Koyanagi, M. Fukuda, N.A.P. Kiran Kumar, L.L. Snead, B.D. Wirth, Y. Katoh, J. Nucl. Mater. 480 (2016) 235. 53. J.R. Cahoon, W.H. Broughton, A.R. Kutzak, Metall. Trans. 2 (1971) 1979. 54. J.T. Busby, M.C. Hash, G.S. Was, J. Nucl. Mater. 336 (2005) 267.

199

8. Overall discussion Selective refractory metals, being considered as promising structural materials, will undergo extreme heat loads and radiation fields when used as plasma facing material in nuclear fusion reactors. The defects occurred in structural parts under heat-radiation loadings expect to change mechanical properties of the chosen materials and by this, may lead to early failure of construction parts. The current frontrunner for the production of plasma-facing components for fusion reactors is tungsten. Unfortunately, tungsten is intrinsically brittle, and its ductile- to-brittle transition temperature shifts to higher values with increasing radiation damage levels. This brittle nature of tungsten can lead to premature catastrophic failure of reactor components. Therefore, it is of paramount importance to design such alloyed composition to enhance its ductility, and, hence, to extend the temperature window of W-based alloys for safe reactor operation. This dissertation focused on finding the ways to improve the structural performance of tungsten exploring two potential routes, such as: 1. Alloying tungsten with tantalum; 2. Considering alternative compositions to substitute tungsten which can be adopted as a full-thickness structural material or can be used to coat reactor components. The chosen method of mimicking fusion relevant damage with proton beams in order to study radiation defects contributes to the understanding of potential material degradation due to the fast neutrons, the production of helium by injection from the plasma and in (n, α) reactions, and the formation of transmutation products which are all expected to happen in ITER. In this research, W, Ta and their mutual alloys were analyzed in a transmission electron microscope (TEM) after proton irradiation (ex-situ) and directly during irradiation exposure (in-situ). Materials under investigations were exposed to proton irradiation either of 3 MeV (ex-situ) or 40 KeV (in-situ) energies using Pelletron ion accelerator placed at the Dalton Cumbrian Facility and MIAMI 1 system in the University of Huddersfield respectively. Afterward, the samples were compared in the form of bulk material or as foils transparent enough for the incident electron beam. Irradiation-caused nano-scale crystal defects, such as dislocation loops and voids were studied in terms of (i) damage level, (ii) temperature, (iii) composition, and (iv) free surface effect by comparing results received on bulk ex-situ irradiated samples with the thin foil irradiated in-situ. Advanced electron microscopy techniques were applied for understanding the primary mechanisms of radiation damage at its initial nucleation stage in these materials in order to provide more experimental data for existing mechanistic models behind the defect formation and their further evolution in the body-centered cubic metals.

200

Main results of this thesis are split into four chapters. Chapter 4 and 5 in the thesis are dedicated to the investigation of the irradiation-induced microstructure of tungsten which is alloyed with a controlled amount of tantalum and to the study of the post-irradiation thermal development of its microstructure. Influence of incremental damage level on the microstructure of tungsten and tungsten-5wt.%tantalum alloy, the role of tantalum in the radiation damage accumulation in W-5wt.%Ta alloy and an effect of the free surface on dislocation loops generated by the proton beam can be found in chapter 4. As discussed above, tungsten is a very brittle material with known values of the ductile-to- brittle transition temperature (DBTT) set in the range of 250-350C [1, 2] with its potential increase up to 800–1000°C under conditions of neutron irradiation [3, 4]. In this research, we considered to alloy tungsten with tantalum which has similar neutron performance and, can potentially improve the ductility of tungsten [5]. Recently reported self-ion irradiations data on W-Ta alloys at temperatures 500C and up to a damage level of 33dpa confirmed the absence of radiation-induced Ta clustering. The presence of Ta impeded the formation of Re clusters under irradiation [6]. Tungsten transmutation into Re and subsequently Os under neutron bombardment is known for its consequent formation of brittle  (ReW) and 

(WRe3) precipitates which may cause premature failure during low-temperature reactor work. Data on irradiated W-5wt.%Ta found in the literature indicated an increase in hardness caused by irradiation already at relatively low damage levels, e.g. from 7.3±0.2 GPa in the unimplanted condition to 8.8±0.2 GPa at 0.07 dpa, which saturates at 13 dpa with a value of 10.9±0.3 GPa [7]. Limited microstructural analysis of tungsten alloyed with tantalum revealed that the number density of forming dislocation loops in W-Ta alloys is greater than in non-doped tungsten at the same damage level, but the average loop size is reduced [8]. Tantalum may potentially hinder the mobility of self-interstitial atoms and interstitial loops in tungsten [8] which in its turn prevent rapid loop coalescence and therefore loop growth in non-alloyed tungsten. Despite these available limited results, the mechanistic understanding of the structural radiation-caused defects nucleation and evolution in W-Ta alloys leading to the observed hardness increase required extra experimental evidence in order to predict reliably the lifetime performance of the W-Ta material and its potential failure in real reactor conditions. Indeed performed proton irradiation at 350oC induced the appearance of a/2 <111> dislocation loops at 0.1 dpa in both W and W-5wt.%Ta materials and the dislocation loop

201 length increased progressively in tungsten up to 0.7 dpa, whereas the loop width saturates at 0.3 dpa, giving a rise to loop strings. Following irradiation of both W and W-5Ta, post-radiation annealing study is presented in chapter 5. In this study, we have assessed the thermal behavior of the radiation-induced defect structure in W-5wt.%Ta alloy applying in-situ transmission electron microscopy. Reduction of dislocation loop density and size, change of the dislocation loops from interstitial to vacancy following with generation of voids, during post-irradiation annealing process were assessed to observe stages of the material recovery and constitute the main body of chapter 5. The main results can be identified as follows: interstitial-type a/2 <111> dislocation loops are induced during proton irradiation at 350°C and 0.7 dpa and escape to the free sample surface during the early stages of post-irradiation annealing. Above 500oC the vacancy excess in the matrix triggers the occurrence of vacancy-type a/2 <111> dislocation loops. As reported at 650-1000C vacancy-type defects become mobile in this temperature regime, which relates to annealing stage IV [9]. These loop characteristics establish the growth of vacancy-type loops by the loop and vacancy absorption and, by loop coalescence [10]. Subsequently, voids form at 1000C by either absorption of free vacancies present in the matrix, or by vacancy loop shrinkage and collapse via thermal emission of vacancies. The voids have a tendency to grow progressively up to 1300C, whereas the void number density gradually decreases. Due to the observed fact that tantalum delays the void formation by vacancy-solute trapping mechanism [11], is therefore expected restriction of the void-induced swelling in tungsten under tokamak radiation conditions involving neutron irradiation damage. Chapter 6 and 7 are focused on the second potential route which is consideration of alternative tantalum-based alloys. The existing knowledge about the nature of radiation- induced lattice defects in Ta-W alloys (e.g. dislocation loop development), as well as the loop dynamic behavior in the material, and especially their relationship to potential hardening and embrittlement phenomena is still very limited. Dislocation loops and voids formation are reported to occur in pure tantalum at damage level of ≤ 0.3 dpa at a relatively high temperature of 700°C [12]. However, radiation-induced hardening in pure Ta as well as in Ta-W alloys seems to take place already at a dose of ≤ 0.3 dpa and temperatures up to 350°C, based on mechanical testing data of irradiated samples [7, 13, 14]. However, the correlation of the hardening phenomenon with the characteristics of the irradiated structures still remains unknown. We aimed to assess the nature, size and distribution of radiation-induced defects in

202 these materials for selected W contents up to 10wt.%, and to link this structural information to the hardening of the alloys measured with hardness indentations. Tungsten also has been reported to influence the dislocation density and dynamics in non-irradiated Ta-W alloys during mechanical testing [15]. In the current work, proton irradiation was applied to irradiate tantalum and tantalum- tungsten compositions in order to mimic fusion-relevant plasma and neutron irradiation damage. In the completive chapter 6, dependence of dislocation loop and void evolution on irradiation temperature in tantalum was considered. 3 MeV proton irradiation of tantalum at 345oC induces the formation of a/2<111> interstitial loops at 0.15dpa and the vacancy excess leads to the nucleation of randomly distributed voids. Increase in the radiation damage level from 0.15 to 0.55dpa triggers the ordering of these voids into a body-centered cubic lattice with an average void distance of 7.3(2)nm and the simultaneous appearance of dense dislocation tangles. Vacancy mobility at 180oC is reduced which prevents them from coalescing into visible voids. After irradiation at 590oC new, a<100> type of dislocation loops with limited mobility prevailed which also hinders void formation. The throughout electron microscopy analysis of irradiated tantalum as a function of temperature constitutes a key contribution to the mechanistic understanding of the radiation-induced dislocation loop formation and stability, together with the void nucleation and disorder-to-order void transition Chapter 7 reflects unique data of the void ordering observed in tantalum after proton beam exposure and the void lattice development in terms of damage level impact. In the previous investigations, void lattice formation in tantalum has been observed after neutron exposure at a minimum temperature of 585°C [16]. The influence of tungsten concentration in tantalum and the interplay of the void characteristics with the radiation-induced interstitial dislocation loops and tangles that act as sinks for self-interstitial atoms, and therefore create the necessary vacancy excess for voids to form, is under discussion. It was shown that a damage level of 0.1dpa is sufficient to induce a relatively high density of interstitial dislocation loops in the studied alloys up to 10wt.%W. The formation of random voids in W-containing alloys was observed to be delayed due to the impeded self-interstitial atom diffusion and/or loop mobility via a solute drag effect, and therefore in the overall lattice damage induced by radiation. A critical dislocation loop density of ~6 × 1022 m-3 was considered to be required for voids to nucleate in these alloys. The disorderorder transition of voids occurring in tantalum at 0.25dpa enables the nucleation of additional voids with increasing damage level

203 which occur simultaneously up to 1.55dpa, and therefore saturation was not attained at that damage level. All in all, as a big attempt to improve tungsten performance under irradiation condition, this thesis revealed a common pattern in both, W- and Ta-based compositions exposed to proton beams. Addition of alloying component prevents the formation and evolution of nano-scale radiation defects such as dislocation loop and voids as well as void growth and void ordering comparing with pure metals. Structure evolution in W-Ta system as a function of radiation damage level and irradiation temperature constitutes a complete body of knowledge of high interest to the nuclear fusion community and can be used to validate predictive models of the expected in-service degradation of fusion construction materials

References 1. T. Shen, Y.D., Y. Lee, Microstructure and tensile properties of tungsten at elevated

temperatures Journal of Nuclear Materials, 2016. 468: p. 348-354. 2. Philipps, V., Tungsten as material for plasma-facing components in fusion devices. Journal of Nuclear Materials, 2011. 415: p. S2-S9. 3. H. Bolt, V.B., G. Federici, J. Linke, A. Loarte, and K.S. J. Roth, Plasma facing and high heat flux materials – needs for ITER and beyond. Journal of Nuclear Materials, 2002. 307-311: p. 43-52.

4. S.J. Zinkle, N.M.G., Operating temperature windows for fusion reactor structural materials Fusion Engineering and Design, 2000. 51-52: p. 55-71. 5. M. Rieth, S.L.D., S. M., J.A. Gonzalez de Vicente, T. Ahlgren, S. Antusch, D. E. J. Armstrong, M., and N.B.e.a. Balden, A brief summary of the progress on the EFDA tungsten materials program. Journal of Nuclear Materials, 2013. 442 p. 5173-5180.

6. A. Xu, D.E.J.A., C. Beck, M.P. Moody, and P.A.J.B. G.D.W. Smith, et al., Ion- irradiation induced clustering in W-Re-Ta, W-Re and W-Ta alloys: An atom probe tomography and nanoindentation study. Acta Materialia 2017. 124 p. 71. 7. D.E.J. Armstrong, A.J.W., S.G. Roberts, Mechanical properties of ion-implanted tungsten–5 wt% tantalum. Physica Scripta, 2011. 2011(T145): p. 014076. 8. X. Yi, M.L.J., K. Hattar, P. D. Edmondson, S. G. Roberts, Characterisation of radiation damage in W and W-based alloys from 2 MeV self-ion near-bulk implantations. Acta Materialia, 2015. 92: p. 163-177.

204

9. L.K. Keys, J.M., Neutron irradiation and defect recovery of tungsten. Journal of Nuclear Materials, 1970. 34: p. 260-280. 10. X. Yi, M.L.J., M. and S.G.R. Briceno, Z. Zhou, M. Kirk, In-situ study of self-ion irradiation damage in W and W–5Re at 500°C. Philosophical Magazine 2013. 93: p. 1715-1738.

11. F. A. Smidt, J.A.S., Suppression of void nucleation by a vacancy trapping mechanism. Scripta Metallurgica 1973. 7: p. 495-502.

12. Yasunaga, K., et al., Microstructure of tantalum irradiated with heavy ions. Journal of Nuclear Materials, 1998. 258–263, Part 1: p. 879-882. 13. Byun, T.S. and S.A. Maloy, Dose dependence of mechanical properties in tantalum and tantalum alloys after low-temperature irradiation. Journal of Nuclear Materials, 2008. 377(1): p. 72-79. 14. Chadwick, D.B., P.K. Daniel, and T. Joseph, Investigation of Effects of Neutron Irradiation on Tantalum Alloys for Radioisotope Power System Applications. 2007. p. 224. 15. Briant, C.L. and D.H. Lassila, The Effect of Tungsten on the Mechanical Properties of Tantalum. Journal of Engineering Materials and Technology, 1999. 121(2): p. 172- 177. 16. Wiffen, F.W., The microstructure and swelling of neutron irradiated tantalum. Journal of Nuclear Materials, 1977. 67(1): p. 119-130.

205

9. Conclusions

In this project we tackled the challenge of improving the structural performance of tungsten, the primary candidate for use as plasma facing material in fusion reactors, by alloying with tantalum as well as exploring tantalum-based compositions as a potential alternative. The formation and evolution of nano-scale structural defects induced by proton irradiation have been studied in tungsten, tantalum and their mutual binary alloys. In-depth in-situ/ex-situ transmission electron microscopy analysis of the structural damage caused by proton exposure, revealed the similar tendency of hindrance of the lattice defect development in the binary alloys with respect to the pure metals. This effect relates to both the dislocation loop conversion into dislocation tangles, and also to the void growth and its disorder→order transition at higher damage levels.

The following is a summary of the main conclusions:  The b=1/2〈111〉 loop motion is hindered by the addition of 5wt.% of Ta to W and results in a reduction in hydrogen bubble formation and blister growth in W.  The recovery process in W-5t.%Ta, irradiated with protons at 350C, occurs in several sequential stages. First, a/2<111> interstitial loops have disappeared during heating up to 500oC, and the microstructure is characterized by the presence of vacancy-type a/2<111> dislocation loops grow in size between 600-900oC. The loops change to vacancy nature in expense of vacancy excess at higher irradiation temperature. With a further increase in temperature, the vacancy loops collapse, and/or absorb additional vacancies, and consequently voids form at 1000oC. They grow in diameter with temperature, but void density decreases, up to 1300oC.  The highest dislocation loop density in proton irradiated Ta was observed at 180oC, due to relatively high SIA mobility as compared to vacancy motion at that temperature. At 350oC we have detected the formation of a void lattice in Ta at a damage level of 0.25dpa. An increase in irradiation temperature up to 590oC leads to the predominance of radiation-induced a<100> dislocation loops in Ta structure, which prevents the appearance of voids and void lattice ordering.  A delayed dislocation loop evolution and void formation was observed in Ta with an increasing concentration of W up to 10wt.%. The change in hardness due to proton irradiation is smaller in the two Ta-based materials doped with W as compared to Ta. This has been attributed to the lower dislocation loop density found in these materials with increasing W content.

206

The study constitutes the foundation stone in the mechanistic understanding of the lattice defect formation and evolution in W-Ta system in radiation environments relevant to fusion reactors. The results obtained in this thesis can be used to validate predictive models of the expected in-service degradation of fusion construction materials, so that more advanced alloy compositions with optimised properties can be designed and developed in the future.

207

10. Future work

Based on the unique experimental results shown throughout this thesis, it is important to mention a few potential avenues for future research. As a direct extension of the present structural characterization analysis, we can propose the following specific studies:

 A quantitative chemical characterization in the vicinity of the radiation-induced dislocation loops observed in these alloys in order to build a complete understanding of the delay damage mechanism due to alloy element in each case. In order to do so, electron energy loss spectroscopy (EELS) measurements could be performed to investigate the solute atom distribution close to the loops and their local atomic environment.

 In order to correlate the microstructural modifications induced by irradiation with changes in mechanical behaviour, micro tensile testing could be performed using advanced plasma Focus Ion Beam (FIB) to machine micro-tensile specimens, together with in-situ TEM during straining. These experiments could provide valuable data on the interaction of lattice defects with gliding dislocations. Another possibility to study the interaction of dislocations with irradiation-induced defects would be to perform post-mortem TEM investigations of irradiated regions after deformation by nanoindentation. Electron transparent regions could be extracted by FIB.

 Higher temperature irradiations are required for tungsten and tungsten-based alloys, in order to observe when void formation takes place under proton beam exposure at temperatures close to the recommended upper limit for safe operation for these materials. Additionally, it would be beneficial to probe the early stages of radiation damage by collecting in-situ TEM data as a function of damage level below 0.2- 0.3dpa.

 Additional data are required in the case of tantalum at higher irradiation temperature (590oC), in order to evaluate the susceptibility to void formation in prospect of its future application in the fusion community.

 Irradiation of the binary alloys of this study with H+ + He dual beams would be of great interest, since helium atoms may stabilise irradiation-produced defects. Hence, this would help to understand if the proposed alloys would experience a reduced damage under simultaneous helium and proton fluxes.

208