Potential ultra hard phases of
 ZrO2-HfO2 solid-solutions, derived
 from high P-T conditions
 Neil Jackson
 Faculty of Science
 University of the Witwatersrand
 A dissertation submitted to the Faculty of Science, University of the Witwatersrand,
 Johannesburg, in ful lment of the requirements for the degree of
 Master of Science.
 2010

Abstract
 The Merrill-Bassett gem anvil cell with a newly developed anvil polishing tech-
 nique and the CO2 laser heated diamond anvil cell was used to search for possi-
 ble ultra hard phases of the unstabalized ZrxHf1 xO2 (x=0.5) solid-solution. The
 pressure-temperature phase diagram of this compound was explored for the  rst
 time using angle dispersive x-ray di raction. The pressure-temperature behaviour
 of this solid-solution was then compared to the di raction data of the end mem-
 bers ZrO2 and HfO2. The investigative techniques of Raman spectroscopy and x-ray
 di raction were used to clarify previous Raman spectroscopy studies. The struc-
 tural changes from the sevenfold coordinated monoclinic to orthorhombic-I to the
 ninefold orthorhombic-II (cotunnite) structure under high pressure-temperature con-
 ditions were con rmed. Thus the potential for the Zr0:5Hf0:5O2 solid-solution as a
 candidate for an ultra-hard ultra-tough \superceramic" was con rmed.
To my family for all their love and support.
Acknowledgements
 I am especially grateful to my original supervisor, Professor Giovanni Hearne for his
 guidance and encouragement during the course of this work. It has been a pleasure
 to work with such an incredible scientist who is such an expert in the  eld of high
 pressure physics. His scienti c ethic and love for his work was a constant source of
 encouragement and motivation during the numerous challenging experiments per-
 formed. I am thankful to Dr Naidoo, my supervisor since July 2009, for the support
 in administration, for his motivation in keeping my schedule on track and the  nal
 proof reading of my dissertation in the closing months of this dissertation. I ac-
 knowledge Professor Dave Billing my co-supervisor in the Department of Chemistry
 for assistance with regards to x-ray di raction and allowing great ease of access to
 the Structural Chemistry resources.
 This work could not have been completed without the invaluable advice and assis-
 tance from a number of people:
 Professor F. Schoning for expert assistance in understanding the complexities in-
 volving high pressure x-ray di raction and advice on the analysis of results. Mr R.
 Erasmus for the many hours of Raman measurement and assistance in investigating
 the many challenges involving high pressure Raman measurement. He has certainly
 played an integral role in the advancement of this dissertation. The technical sta 
from the workshop in the School of Physics for their assistance provided in machining
 and modifying high pressure apparatus. Dr A. Forbes at NLC CSIR, who provided
 much assistance in the CO2 laser heating set up and provided much of the optics
 for the laser heating system through the rental pool system. Thanks goes to Mr
 C. Kasl for his suggestions and the valuable assistance with proof reading the  nal
 document.
 The high pressure studies group and the structural chemistry students for general
 support and stimulating discussions during the experiments.
The support of the DST/NRF Centre of Excellence in Strong Materials (CoE-SM)
 towards this research is hereby acknowledged. Opinions expressed and conclusions
 arrived at, are those of the author and are not necessarily attributed to the CoE-
 SM. My gratitude goes to the wonderful people working at the CoE-SM for their
 assistance in administration and  nancial support provided to me during the course
 of this work.
 Lastly a big thank you to my wonderful family and friends for their generous encour-
 agement, understanding and patience which kept me motivated during the course of
 this degree.
Contents
 List of Figures ix
 List of Tables xii
 Symbols and Acronyms xiii
 1 Introduction 1
 1.1 Originality and novel features of the research . . . . . . . . . . . . . . . . . . . . 3
 1.2 Aims and approaches of the project . . . . . . . . . . . . . . . . . . . . . . . . . . 4
 1.3 Structure of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
 2 Literature Review and Description of Experimental Methodologies 6
 2.1 Ultra hard materials - unstablized Zr0:5Hf0:5O2 as a candidate . . . . . . . . . . . 7
 2.1.1 Sol-gel processing of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . . . . . . . 7
 2.1.2 Similarities between ZrO2 and HfO2 . . . . . . . . . . . . . . . . . . . . . 8
 2.1.3 The high pressure and temperature phases of ZrO2 and HfO2 . . . . . . . 9
 2.2 High pressure methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
 2.2.1 The gem anvil cell (GAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
 2.2.1.1 CZ hand polished anvils . . . . . . . . . . . . . . . . . . . . . . . 17
 2.2.2 The Merrill-Bassett diamond anvil cell (MB-DAC) . . . . . . . . . . . . . 18
 2.2.3 Determining pressure within the pressure cell . . . . . . . . . . . . . . . . 20
 2.2.4 Pressure transmitting medium (PTM) . . . . . . . . . . . . . . . . . . . . 22
 2.2.5 Gasket materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
 2.3 Heating the sample in a MB-DAC . . . . . . . . . . . . . . . . . . . . . . . . . . 24
 2.3.1 CO2 laser heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
 2.4 Raman spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
 vi
CONTENTS
 2.4.1 Vibrational energies and polarization e ects . . . . . . . . . . . . . . . . . 28
 2.4.2 Raman selection rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
 2.4.3 Confocal laser Raman micro-spectroscopy instrumentation . . . . . . . . . 30
 2.4.4 Using Raman spectroscopy in the pressure cell . . . . . . . . . . . . . . . 31
 2.5 X-ray di raction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
 2.5.1 X-ray di raction theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
 2.5.2 Monocapillary optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
 2.5.3 Crystallography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
 2.5.4 X-ray di raction at extreme pressures in practice . . . . . . . . . . . . . . 36
 3 Experimental 39
 3.1 Sample characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
 3.1.1 Raman spectroscopy characterization . . . . . . . . . . . . . . . . . . . . . 40
 3.1.2 XRD characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
 3.2 High pressure experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
 3.2.1 Manual fabrication of CZ anvils . . . . . . . . . . . . . . . . . . . . . . . . 40
 3.2.2 Diamond anvils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
 3.2.3 Anvil mounting and alignment . . . . . . . . . . . . . . . . . . . . . . . . 44
 3.2.4 Gasket fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
 3.2.5 Sample and ruby loading . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
 3.2.6 Pressure transmitting media (PTM) . . . . . . . . . . . . . . . . . . . . . 48
 3.2.7 Pressure measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
 3.2.8 Gem anvil cell (GAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
 3.2.9 Diamond anvil cell (DAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
 3.3 High temperature experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
 3.3.1 CO2 laser heating the MB-DAC . . . . . . . . . . . . . . . . . . . . . . . 51
 3.3.2 High pressure Raman spectroscopy . . . . . . . . . . . . . . . . . . . . . . 53
 3.3.3 High pressure XRD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
 4 Experimental Results and Analysis 57
 4.1 Ambient Raman and XRD measurements on Zr0:5Hf0:5O2 . . . . . . . . . . . . . 57
 4.2 Hand polished CZ anvils and initial Raman TiO2 results . . . . . . . . . . . . . . 58
 4.3 High pressure and high temperature Raman and XRD experiments . . . . . . . . 63
 4.3.1 MB-GAC strong scatterer TiO2 and weak scatterer Zr0:5Hf0:5O2 experiments 63
 vii
CONTENTS
 4.3.2 MB-DAC high pressure with Fluorinert PTM experiments . . . . . . . . . 68
 4.3.3 MB-DAC high pressure-temperature with Ar PTM experiments . . . . . . 75
 4.3.4 Comparison of the 20 GPa high pressure experiments as a function of
 temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
 5 Discussion and Conclusions 86
 Bibliography 89
 viii
List of Figures
 1.1 Hardness of materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
 1.2 Phase diagram for ZrO2 and HfO2 . . . . . . . . . . . . . . . . . . . . . . . . . . 3
 2.1 Bulk modulus of some hard materials . . . . . . . . . . . . . . . . . . . . . . . . 8
 2.2 Phase diagram of ZrO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
 2.3 Phase diagram of HfO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
 2.4 Monoclinic ZrO2 or HfO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
 2.5 High pressure orthoI phase of ZrO2 or HfO2 . . . . . . . . . . . . . . . . . . . . . 12
 2.6 High pressure orthoII phase of ZrO2 or HfO2 . . . . . . . . . . . . . . . . . . . . 12
 2.7 Shock wave vs. DAC experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
 2.8 DAC pressure generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
 2.9 FT-IR absorbance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
 2.10 Raman spectrum for sapphire anvils . . . . . . . . . . . . . . . . . . . . . . . . . 15
 2.11 CZ anvil Raman bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
 2.12 Properties of diamond, sapphire and cubic zirconia . . . . . . . . . . . . . . . . . 17
 2.13 Interface between the CZ anvil and glass showing Newtons interference fringes . 18
 2.14 Exploded view of the MB cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
 2.15 Ruby  uorescence R1 and R2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
 2.16 Electron-energy level term scheme . . . . . . . . . . . . . . . . . . . . . . . . . . 21
 2.17 Cryogenic loading apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
 2.18 Laser heating setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
 2.19 Principle of Stokes and anti-Stokes . . . . . . . . . . . . . . . . . . . . . . . . . . 27
 2.20 Molecular vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
 2.21 Molecular vibrational modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
 2.22 Principle of confocal Raman spectroscopy . . . . . . . . . . . . . . . . . . . . . . 31
 ix
LIST OF FIGURES
 2.23 Electromagnetic spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
 2.24 XRD spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
 2.25 X-ray di raction from a powder sample . . . . . . . . . . . . . . . . . . . . . . . 35
 2.26 Monocapillary glass optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
 2.27 Boehler-Almax anvils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
 2.28 Siemens SMART di ractometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
 3.1 Merrill-Bassett cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
 3.2 CZ anvil polishing holder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
 3.3 CZ culet  nal soft cloth polish . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
 3.4 CZ anvil balancing test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
 3.5 Gem alignment holder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
 3.6 Newtons interference fringes for the diamond-culet/glass interface . . . . . . . . . 45
 3.7 Spark erosion machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
 3.8 Gasket indentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
 3.9 Ruby  uorescence microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
 3.10 Two samples TiO2 and Zr0:5Hf0:5O2 in the MB-GAC . . . . . . . . . . . . . . . . 50
 3.11 22 mm diameter MB-DAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
 3.12 CO2 laser heating facility details . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
 3.13 Siemens SMART MB-DAC stage . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
 3.14 XRD peak identi cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
 4.1 Raman Zr0:5Hf0:5O2 powder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
 4.2 Ambient Raman spectra of Zr0:5Hf0:5O2 mixed with PTM . . . . . . . . . . . . . 59
 4.3 Raman spectrum of the CZ anvil . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
 4.4 Intensity comparison of sample Raman bands . . . . . . . . . . . . . . . . . . . . 61
 4.5 Raman MB-GAC high pressure measurement of TiO2 . . . . . . . . . . . . . . . 61
 4.6 Sharp and rounded CZ anvil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
 4.7 Pressure measurement to 13.75 GPa . . . . . . . . . . . . . . . . . . . . . . . . . 62
 4.8 MB-GAC loaded with TiO2 and Zr0:5Hf0:5O2 sample . . . . . . . . . . . . . . . . 63
 4.9 Raman MB-GAC TiO2 high pressure measurements . . . . . . . . . . . . . . . . 64
 4.10 Raman MB-GAC Zr0:5Hf0:5O2 high pressure measurements . . . . . . . . . . . . 66
 4.11 MB-DAC cavity loaded with Zr0:5Hf0:5O2 and Fluorinert PTM. . . . . . . . . . . 68
 4.12 XRD of raw Zr0:5Hf0:5O2 powder . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
 x
LIST OF FIGURES
 4.13 XRD of Zr0:5Hf0:5O2 in the MB-DAC . . . . . . . . . . . . . . . . . . . . . . . . 69
 4.14 11 GPa XRD data of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . . . . . . . . . . 71
 4.15 XRD high pressure orthoI phase of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . . . 71
 4.16 Raman MB-DAC high pressure of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . . . 72
 4.17 Raman MB-DAC decompression from 20 GPa to 2 GPa of Zr0:5Hf0:5O2 . . . . . 73
 4.18 XRD MB-DAC decompression from 20 GPa to 2 GPa of Zr0:5Hf0:5O2 . . . . . . 74
 4.19 MB-DAC cavity loaded with Zr0:5Hf0:5O2 and Ar PTM . . . . . . . . . . . . . . 76
 4.20 Raman Zr0:5Hf0:5O2 and Ar PTM . . . . . . . . . . . . . . . . . . . . . . . . . . 76
 4.21 Raman spectra at 15 GPa of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . . . . . . 77
 4.22 XRD data of Zr0:5Hf0:5O2 with Ar PTM . . . . . . . . . . . . . . . . . . . . . . . 79
 4.23 d-spacing obtained from XRD data of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . 80
 4.24 OrthoII Raman comparison at 20 GPa of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . 83
 4.25 XRD comparison at 20 GPa of Zr0:5Hf0:5O2 . . . . . . . . . . . . . . . . . . . . . 84
 xi
List of Tables
 4.1 d-spacing of Ar PTM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
 4.2 20 GPa d-spacing of fully converted high P orthoI phase . . . . . . . . . . . . . . 81
 4.3 20 GPa d-spacing of high P orthoI and orthoII mixture after laser heating . . . . 81
 xii
Symbols and Acronyms
  HV - vickers hardness
  x - mols of Zr in ZrxHf1 xO2
  P - pressure
  T - temperature
  CZ - cubic zirconia
  MB - Merrill-Bassett
  GAC - gem anvil cell
  DAC - diamond anvil cell
  (hkl) - Miller indicees
  OrthoI - orthorhombic-I
  OrthoII - orthorhombic-II (cotunnite)
  SG(#) - space group number
  PTM - pressure transmitting medium
  IR - infra-red
   - wavelength
  BA - Boelher-Almax
  TP - thrust plate
  BP - backing plate
 xiii
1
 Introduction
 Zirconium dioxide (zirconia, ZrO2) and hafnium dioxide (hafnia, HfO2) are well known binary
 oxide materials that are used in applications such as structural ceramics, high temperature solid
 electrodes and also substrates for growth of other materials [1]. Materials of high hardness and
 toughness, with erosion and corrosion resistance, are increasingly important for modern industry.
 High hardness corresponds to high bulk modulus or low compressibility and some non-metallic
 materials with a high bulk modulus are shown in  gure 1.1 [2]. Ultra hard materials are especially
 important in South Africa?s billion dollar mining industry, where they can be used in grinding
 and cutting tools such as drill bits for oil exploration.
 Zirconia (ZrO2) based ceramics exhibit high temperature strength and high toughness. Pos-
 sible applications of these ceramics include high-K (high permittivity) materials for use in
 Ge channel materials in metal-oxide-semiconductor  eld-e ect transistor (MOSFETS) and to
 replace SiO2 as the gate dielectric in complementary metal-oxide-semiconductor (CMOS) tech-
 nology [3]. These ceramics are used extensively as implants in the dental industry in the form
 of dental caps and in optical  lms since zirconia based ceramics are transparent and have a high
 refractive index.
 At ambient pressure and temperature both ZrO2 and HfO2 are stable in the so-called bad-
 deleyite or monoclinic structure. At temperatures above  1000 C their structures change to
 tetragonal and then to the cubic  uorite structure as shown in  gure 1.2 [1, 4]. At temperatures
 below 500 C the phase changes from monoclinic to orthorhombic-I (orthoI) at around 5 GPa,
 and to orthorhombic-II (orthoII or the so-called cotunnite structure) between 10 and 15 GPa
 in these materials. The high-pressure stabilized cotunnite phase can be recovered to ambient
 conditions (displays irreversibility) whereas orthoI cannot. The transformation to tetragonal
 1
Figure 1.1: Hardness of materials - Hardness of some non-metallic materials as a function of
 bulk modulus [2].
 from the orthoI phase is independent of pressure and occurs above 600 C for ZrO2 [4] and above
 1200 C in the case of HfO2 [1]. Hafnia therefore has higher application temperatures than zir-
 conia and is a better refractory material. The bulk modulus of cotunnite zirconia has been
 determined experimentally to be 278 GPa [5], compared with the calculated value of 305 GPa
 by Dewhurst [6]. The calculated and experimental bulk modulus of HfO2 has been determined
 to be 306 GPa and 320 GPa respectively [7]. These high bulk moduli values demonstrate that
 both ZrO2 and HfO2 have properties of hard materials.
 To achieve these ultra hard phases it is necessary to generate high pressures and high tem-
 peratures. To generate these extreme conditions anvil cells, multi anvil presses and shock com-
 paction pressure techniques are used in conjunction with laser, resistive or furnace heating
 techniques. This project will investigate the ZrO2-HfO2 solid solution using the Merrill-Basset
 (MB) type diamond anvil cell (DAC) and gem anvil cell (GAC) in conjunction with CO2 laser
 heating. Two diamond anvils with truncated tips are mounted in opposed orientation with a
 sample con ned to a cavity drilled in a metal foil gasket sandwiched between the diamond tips.
 Pressurized, this assembly can be mounted in the CO2 laser heating system. Thereafter the
 CO2 laser can be focused directly on the sample and the entire sample raster scanned so that a
 2
1.1 Originality and novel features of the research
 Figure 1.2: Phase diagram for ZrO2 and HfO2 - (a) ZrO2 phase diagram, dotted and solid
 lines represent previously reported phase boundaries. Closed triangles, open circles, and closed circles
 represent the orthoI, tetragonal and orthoII phases respectively [4]. (b) HfO2 phase diagram, dotted
 and solid lines represent previously reported phase boundaries. Open triangles, closed squares, open
 squares and open circles represent the monoclinic, orthoI, tetragonal and orthoII phases, respectively
 [1].
 comparison between the XRD and Raman spectrum can be made.
 Structure changes were monitored via Raman spectroscopy and then combined with ex-
 perimental x-ray di raction (XRD) patterns and then compared to results obtained from the
 literature of the end members.
 1.1 Originality and novel features of the research
 To date, there have been no published investigations of the pressure-temperature (P-T) phase
 diagram for the Zr0:5Hf0:5O2 solid-solution system. A recent in situ high pressure Raman
 spectroscopy study on Zr0:5Hf0:5O2 at room temperature to pressures of  30 GPa suggests
 that it adopts a phase transition sequence similar to the end-members ZrO2 and HfO2 (mono-
 clinic/baddeleyite to orthoI to orthoII/cotunnite) [8]. Raman spectroscopy does not readily give
 structural details and there has been no prior (XRD) structural characterization of the high P-T
 (orthoI and orthoII) phases of Zr0:5Hf0:5O2. The better characterization of high P-T phases of
 the solid solution Zr0:5Hf0:5O2 using XRD and Raman spectroscopy will be the main goal of this
 study. New hand polishing cubic zirconia (CZ) anvil methods were developed for use in the gem
 anvil cell and tested in the region of  15 GPa. These newly developed CZ anvils were used to
 3
1.2 Aims and approaches of the project
 study the phase change from monoclinic to orthoI and the metastability of the orthoI phase of
 the Zr0:5Hf0:5O2 solid-solution at low pressures.
 The facilities for processing to extreme conditions of high pressure/temperature (P-T) in the
 laser heated diamond anvil cell in the School of Physics at the University of the Witwatersrand
 (WITS) are perhaps unique in the Southern Hemisphere.
 1.2 Aims and approaches of the project
 The aim of this project was to investigate the potential ultra hard phases of the sol-gel processed
 ceramic Zr0:5Hf0:5O2 solid-solution. High importance was placed on gaining extensive experience
 in high pressure and high temperature experimental techniques. The gem and diamond anvil
 cells were prepared for experiments using Raman spectroscopy and the limitations of the sample
 thickness, pressure transmitting medium and anvils investigated. Familiarity of laboratory based
 analysis techniques was a priority, therefore training on the modi ed Siemens SMART XRD
 system was extensive. Emphasis was placed on the development of the hand polishing of CZ
 anvils for use in the gem anvil cell to use at low pressures. Zr0:5Hf0:5O2 (weak scatterer) and
 TiO2 (strong scatterer) were investigated in the gem anvil cell using the triple confocal Raman
 conditions.
 The project was divided up into four stages. The  rst stage was to become familiar with high
 pressure methodology. As the diamond anvils are expensive and easily breakable, a method was
 devised to hand polish cubic zirconia anvils. Moderate pressures ( 10 GPa) can be achieved
 using these gem anvils and valuable experience gained with the Zr0:5Hf0:5O2 solid-solution system
 and the high pressure apparatus. With experience gained in gem anvil cell methodology the
 diamond anvil cell was prepared and loaded. Learning and using Raman spectroscopy and
 XRD analysis techniques followed including training on a myriad of software packages. The
 last stage was to perform CO2 laser heating on the sample. Liquid Ar was chosen to be the
 pressure transmitting medium to insulate the sample and was cryogenically loaded. The CO2
 laser heating system involves setting up and performing preliminary alignment and modi cations
 speci c to the experimental conditions.
 4
1.3 Structure of the dissertation
 1.3 Structure of the dissertation
 The background to the research topic, the aims and the novel aspects of this project are dis-
 cussed in this introduction. A literature review containing details surrounding the similarities of
 the zirconia and hafnia systems including the structural similarities of the system under extreme
 conditions of pressure and temperature is presented in chapter 2. The literature review will in-
 clude detailed discussions on the Merrill-Bassett gem anvil cell (MB-GAC), the Merrill-Bassett
 diamond anvil cell (MB-DAC), Raman spectroscopy, XRD and CO2 laser heating. Experimen-
 tal details are described in chapter 3. In chapter 4, the results obtained from high pressure
 experiments and laser heating experiments are presented in detail. The general discussion and
 conclusions of the results are presented in chapter 5.
 5
2
 Literature Review and Description
 of Experimental Methodologies
 This chapter presents background information regarding the techniques, methods and theory
 covered in this dissertation. It will cover information regarding the sample chosen. The exper-
 imental apparatus, including the high pressure cells and the CO2 laser heating system, Raman
 spectroscopy and XRD will be discussed in detail.
 The literature review is divided into  ve main sections. Section 2.1 will discuss the impor-
 tance of ultra hard materials in the Southern African context. Comparisons on the unstabalized
 Zr0:5Hf0:5O2 solid-solution will be limited to information available on the end members ZrO2
 and HfO2, and Raman spectroscopy work completed by Mojaki [8] in her PhD thesis.
 Section 2.2 describes high pressure gem and diamond anvil cell methodology. The merits
 of static pressure generation will be compared to shock compression pressure generation. This
 section will introduce the various Merrill Basset (MB) high pressure cells and the MB pressure
 cell used in the current investigation will be described in detail.
 This section will also include details on the new hand polishing technique used to manufacture
 CZ anvils as developed in our high pressure laboratory at WITS. The details of the metal foil
 used to make gaskets will be reviewed along with details concerning the pressure transmitting
 media and pressure measurement.
 Section 2.3 describes the heating techniques using a 30 W CO2 laser and its implementation
 in attaining high temperatures in the MB-DAC. The components of the laser heating system
 will be discussed along with the improvements made. Focusing the laser into a  200  m di-
 ameter sample cavity is challenging and the theory of this setup is discussed. The temperature
 6
2.1 Ultra hard materials - unstablized Zr0:5Hf0:5O2 as a candidate
 measurement is non-trivial and some basic theory will be covered.
 Section 2.4 describes the theory of the Raman spectroscopy technique and the implementa-
 tion of Raman spectroscopy with the gem and diamond anvil cells is discussed.
 Section 2.5 will focus on XRD applications in high pressure research. High pressure and
 high temperature are key techniques to fully explore the P-T phase diagram. Using XRD, phase
 transformations can be quali ed and with su ciently high quality data, quanti ed.
 Raman inelastic light scattering by long wavelength optical phonons is a convenient and
 sensitive probe used to characterize local structural changes and able to monitor structural
 transitions in compounds. Combining this technique with XRD data will help qualify the phase
 information contained in the Raman spectra for the  rst time.
 2.1 Ultra hard materials - unstablized Zr0:5Hf0:5O2 as a candidate
 The hardness of materials is related to the plastic deformation after indentation [2]. Ultra hard
 materials are very important in many sectors in South Africa and the world. The search for
 compounds as hard as diamond is very competitive and a lucrative industry. Materials with
 high hardness have many applications in high performance engineering applications like wear
 resistive coatings and machining [2] (a selection of these materials is shown in  gure 2.1).
 To achieve these ultra hard materials, high pressure and temperature is used as a means of
 stabilizing the material into more closely packed structures and also increases the de-localization
 of the valence electrons [9]. The e ect of temperature results in less ordered or higher en-
 tropy structures, ionization and the disruption of chemical bonds [9]. By investigating the
 phase changes of a material we can potentially form interesting metastable phases which can
 be quenched to ambient conditions. The following section will describe the current informa-
 tion regarding Zr0:5Hf0:5O2 as a possible candidate for a \superceramic". To my knowledge
 Zr0:5Hf0:5O2 has not been studied by XRD before and information in the literature pertains to
 the end members ZrO2 and HfO2 and initial Raman spectroscopy performed by Mojaki [8] who
 synthesized the samples via the sol-gel method.
 2.1.1 Sol-gel processing of Zr0:5Hf0:5O2
 The sol-gel method, a low temperature method that produces better purity than conventional
 high temperature processes, was used to synthesize the Zr0:5Hf0:5O2 ceramic [8]. The sol-gel
 method involves the precipitation of the metal oxide particles in a solution to form a colloidal
 7
2.1 Ultra hard materials - unstablized Zr0:5Hf0:5O2 as a candidate
 Figure 2.1: Bulk modulus of some hard materials - Modi ed  gure of bulk modulus as a
 function of the mean volume per atom (log-log scale), with the solid diamond symbol representing
 high pressure phases of dioxides [1].
 suspension called a sol [8]. Sol-gel processing can produce a variety of forms of a material, for
 example ultra- ne or spherical powders [8]. Zr0:5Hf0:5O2 was chosen as the project candidate
 because of promising preliminary high P-T work performed by Mojaki [8].
 2.1.2 Similarities between ZrO2 and HfO2
 Zirconium and hafnium have similar atomic radii of 0.1442 nm and 0.1454 nm, respectively [8].
 The compounds are so similar that HfO2 was  rst discovered by x-ray spectroscopy in a ZrO2
 sample [10]. Both ZrO2 and HfO2 have high melting temperatures and form glassy materials
 when heated [10]. The ionic radii are 0.79  A and 0.78  A for Zr4+ and Hf4+ with electronegativity
 values 1.22 and 1.23, respectively. These  gures demonstrate the similarities between Zr, Hf and
 their oxides. Due to the similarities of the crystal structure and phase transformation between
 ZrO2 and HfO2, continuous solid-solutions can be formed over the entire range of composition
 [8].
 8
2.1 Ultra hard materials - unstablized Zr0:5Hf0:5O2 as a candidate
 2.1.3 The high pressure and temperature phases of ZrO2 and HfO2
 At ambient pressure and temperature, zirconia is stable in the so-called baddeleyite structure
 (monoclinic P21=c). It changes to the tetragonal structure (P42=nmc) at 1100 C, and then
 transforms to the cubic  uorite structure (Fm3m) at  2300 C [4]. Five structures of ZrO2
 have been identi ed at di erent pressures and temperatures. At ambient pressure zirconia is
 stable at room temperature in the monoclinic phase. It then transforms to tetragonal-I structure
 (P42=nmc) at  950 C and to the cubic  uorite phase at  2300 C with space group (Fm3m) [8].
 Under high pressure of about 5 GPa at room temperature, ZrO2 transforms to the tetragonal-I
 structure and then to an orthorhombic phase at about 15 GPa [8]. Ohtaka et al: [4] however,
 reported that the pressure induced transformation at 5 GPa goes to the orthoI phase and not the
 tetragonal-I phase and then to orthoII at  15 GPa. These transformations are reversible when
 the compound cools back to ambient conditions. When the tetragonal phase reverts back to the
 monoclinic phase, there is a volume increase of about 5% which causes cracking and structural
 failure, thus making it unsuitable to use pure zirconia as a starting material for synthesizing a
 new ultra hard material [8].
 Figure 2.2: Phase diagram of ZrO2 - The pressure-temperature phase diagram of ZrO2. Re-
 ported phase boundaries are given by dashed lines. Phase boundaries by Ohtaka et al: [4] given
 by solid lines. Closed triangles, open circles, and closed circles represent the orthoI, tetragonal and
 orthoII phases, respectively [4].
 9
2.1 Ultra hard materials - unstablized Zr0:5Hf0:5O2 as a candidate
 The P-T phase diagram of ZrO2 is depicted in  gure 2.2. Zirconia changes from monoclinic
 to orthoI (Pbca) with a distorted  uorite structure, to orthoII (Pnam) structure with increasing
 pressure [4]. Zirconia changes to the tetragonal phase (above 600 C), after starting from the
 orthoI phase, as this transformation is independent of pressure, it then transforms to orthoII
 above 12.5 GPa accompanied by a 9% volume reduction [4].
 Ohtaka et al: [4], found that the phase transformation of ZrO2 from orthoI to tetragonal
 to be independent of pressure. For single crystals, the transformation of orthoI occurs between
 4 GPa and 12 GPa [11] and transformation pressures are dependent on grain size. The high-
 pressure stabilized cotunnite phase can be recovered to ambient conditions whereas the orthoI
 phase cannot. The cotunnite phase is stable up to 1800 C and pressures well beyond 24 GPa
 [4]. Ohtaka et al: [5] determined the bulk modulus of cotunnite zirconia experimentally to
 be 278 GPa compared with the calculated value of 305 GPa [6]. Based on the bulk modulus
 hardness correlation plot of Yang et al: [12], cotunnite zirconia can be considered to be a hard
 material.
 Figure 2.3: Phase diagram of HfO2 - Pressure-temperature phase diagram of HfO2 determined
 from in situ XRD measurements. Reported phase boundaries are given by dashed lines and the
 triangle, solid square, square and circle symbols represent the monoclinic, orthoI, tetragonal and
 orthoII phases, respectively [1].
 10
2.2 High pressure methodology
 Hafnia (HfO2) changes to the same structures as zirconia under the increase of pressure
 at similar values (ZrO2 in  gure 2.2 and HfO2 in  gure 2.3). As the temperature increases,
 the phase changes from orthoI to tetragonal but at a higher temperature than in zirconia. It
 is isostructural to zirconia and more stable. Hafnia has applications at higher temperatures
 than zirconia and therefore is a better refractory material. Similarly to zirconia, as shown in
  gure 2.3, hafnia transforms from monoclinic to orthoI to orthoII (cotunnite), structures with
 increasing pressure. Liu [11] determined the transition pressures to be 4 GPa and 14.5 GPa and
 like zirconia the transformation pressures are also dependent on grain size. The cotunnite phase
 in hafnia can be retained to room temperature and atmospheric pressure as in zirconia, while
 the orthoI phase cannot. The calculated and experimental bulk modulus has been determined
 to be 306 GPa and 320 GPa, respectively [7], therefore giving properties of a hard material (high
 bulk modulus).
 Figure 2.4: Monoclinic ZrO2 or HfO2 - Monoclinic ambient phase of ZrO2 or HfO2, space group
 P121/c1 SG(14) ICSD-89426 [13]. Zr or Hf represented by small white spheres and O are the red
 spheres. The monoclinic phase of ZrO2 or HfO2 are sevenfold coordinated dioxides.
 Figures 2.4, 2.5 and 2.6 show the monoclinic, orthoI and the orthoII structures of the end
 members ZrO2 and HfO2, respectively.
 2.2 High pressure methodology
 High pressure can be generated using two techniques, via shock wave and static compression.
 Shock wave experiments in increasing pulse duration consists of laser irradiation, electric dis-
 charge and high explosive shock experiments shown in  gure 2.7 [14]. These techniques are
 11
2.2 High pressure methodology
 Figure 2.5: High pressure orthoI phase of ZrO2 or HfO2 - High pressure orthoI phase of
 ZrO2 and HfO2, space group Pbcm SG(57) ICSD-77716 [13]. Zr or Hf represented by small white
 spheres and O are the red spheres.
 Figure 2.6: High pressure orthoII phase of ZrO2 or HfO2 - High pressure orthoII (cotunnite)
 phase of ZrO2 or HfO2, space group Pnma SG(62) ICSD-83862 [13]. Zr or Hf represented by small
 white spheres and O are the red spheres. The orthoII phase of ZrO2 or HfO2 are ninefold coordinated
 dioxides.
 12
2.2 High pressure methodology
 mainly used to simulate natural pressure and temperature occurrences such as meteorite im-
 pacts [14].
 Figure 2.7: Shock wave vs. DAC experiments - Comparison of shock wave pressure generation
 vs. DAC pressure generation [14].
 Shock wave experiments consist of a high velocity projectile that hits a target thus inducing a
 shock wave within the target. The Huguenot relations are used to analyse the target by treating
 the shock wave as an ideal discontinuity. The conservation of mass, momentum and energy must
 be applied across the shock wave giving the following relations [8]:
 P  P0 =  0 s P ; (2.1)
  =
  0 s
 ( s   P )
 ; (2.2)
 E  E0 =
 1
 2
 (V0  V )(P + P0); (2.3)
 where  s is the velocity of the shock wave and  P the velocity of the projectile. E0, P0, V0
 and  0 are the initial energy, pressure, volume and density, respectively.
 The DAC is used to generate static pressures as shown in  gure 2.8, with
 P =
 F
 A
 ; (2.4)
 where P is the pressure and F the force applied to area A, which is the truncated polished tip
 of the anvil called the culet. This relation shows a large force applied to a small area generates
 13
2.2 High pressure methodology
 large pressures. The DAC takes advantage of both of these requirements by having small anvil
 faces and the thrust screws generating large forces. The gasket acts as a container ring and the
 pressure transmitting medium (PTM) ensures that the pressure is transmitted hydrostatically
 (from all sides) to the sample.
 Figure 2.8: DAC pressure generation - Schematic of the principle of pressure generation in a
 DAC with F representing the applied force to the area A [15].
 The details involving the anvil cells will be further described in the sections 2.2.1 and 2.2.2.
 2.2.1 The gem anvil cell (GAC)
 Various gems have been utilized to generate pressure in anvil cells including CZ, sapphire and
 diamond [16]. Diamond anvils have been used to generate the highest static pressures due
 to their high hardness (442 GPa shown in  gure 2.12), de ned as having high bulk modulus
 (measured in GPa) and high shear modulus (values over 100 GPa) [2]. Diamonds also exhibit
 the best thermal conduction at room temperature and excellent electrical insulation [2].
 Diamond anvils are however expensive, di cult to manufacture and the DAC methodology
 is considered to be somewhat specialized, thus limiting widespread application in many labo-
 ratories. In addition, diamonds are not optically transparent over the entire electromagnetic
 spectrum, having experimental limitations such as an intense broad infra-red (IR) absorption
 from 2600 cm 1 to 1800 cm 1 (for type II diamonds) and are also unstable when heated in
 air at around 700 C, as they graphitize [17]. Diamond anvils have a high intensity background
  uorescence which can mask or interfere with the scattering from the sample [17]. In type I
 diamonds there are absorption bands at 2000 cm 1 and 1250 cm 1 and for type II diamonds
 the absorption bands are at 2500 cm 1 and 1850 cm 1 which can mask the area of scattering
 interest (refer to  gure 2.9) [18].
 14
2.2 High pressure methodology
 Figure 2.9: FT-IR absorbance - FT-IR (fourier transform - infra red) absorbance of (A) type I
 diamond, (B) type II diamond, (C) sapphire and (D) cubic zirconia [17].
 Figure 2.10: Raman spectrum for sapphire anvils - Raman intensities for sapphire anvils in
 comparison with the R1 and R2 ruby  uorescence peaks, showing that the sapphire Raman bands,
 excited by 514.5 nm Ar ion laser beam, interfere with the ruby lines at around 5000 cm 1 [19].
 15
2.2 High pressure methodology
 Gem substitutes for diamond anvils have been investigated, each having associated advan-
 tages and disadvantages over diamond. The advantages of using gems is that they are cheaper
 and are electromagnetically transparent in di erent regions (shown in  gures 2.9 to 2.11) to
 that of diamond anvils. Sapphire and quartz would be good choices due to comparable hardness
 with diamond, cost e ectiveness and machining ease. These substitutes are more stable at high
 temperature but have the same disadvantage as diamond in that they absorb light in the mid-IR
 wavelengths as shown in  gure 2.10 [17]. Sapphire cannot be used below 1850 cm 1 ( gures 2.9
 (C) and 2.10) and quartz cannot be used above 2900 cm 1 [17]. Sapphire is inferior to CZ due
 to the IR absorption of two unknown peaks at 692.8 nm and 694.2 nm [19]. Metastable brilliant
 cut single crystal cubic zirconia (ZrO2 stabilized with approximately 10-mol% Y2O3) has found
 widespread acceptance since 1976 as a diamond substitute in the jewellery industry [17]. CZ
 is extremely inert and has a high hardness (Knoop hardness = 1370). CZ can be used up to
 2400 C in air and is transparent in a large part of the electromagnetic spectrum as shown in
  gures 2.9 D and 2.11 [17].
 Figure 2.11: CZ anvil Raman bands - Raman bands for CZ anvils showing (a) excitation by a
 432 nm He-Ne laser beam; (b) by a 514.5 nm Ar ion laser beam [19].
 The GAC has been used to obtain moderate pressures of  15 GPa [20] depending on the culet
 size, pressure transmitting medium and the gasket material used, which will be expanded upon
 in sections 2.2.4 and 2.2.5. Russell and Piermarini [18] highlighted that generating pressures
 16
2.2 High pressure methodology
 above 10 GPa using gem anvils is non-trivial in e orts spanning the last 20 years or so using
 various experimental set-ups. Experimentation of various culet sizes by pairing a 254  m culet
 with a 762  m culet to replace the larger failed anvil thus minimising down time has been
 explored [17]. Russell and Piermarini [18] have used 500  m to 1 mm culets with thin gaskets of
 125  m T301 stainless steel. Patterson and Margrave [17] achieved pressures up to 2 GPa using
 culet sizes of 500  m and 1 mm. Xu and Huang [21] achieved pressures of 8.9 GPa [18]. Russell
 and Piermarini have achieved pressures of 13.2 GPa with a noteworthy signi cant broadening
 of the ruby  uorescence lines above 10 GPa [18]. They highlighted the absence of the CZ anvil
  uorescence or CZ Raman scattering when between 800 cm 1 and 1600 cm 1 wave numbers, for
 a 432 nm He-Ne laser. In  gure 2.12 a comparison of diamond, sapphire and CZ anvils is made
 and presents favourable data supporting CZ anvils as good substitutes for diamond anvils.
 Figure 2.12: Properties of diamond, sapphire and cubic zirconia - The properties of some
 anvils that have been used to generate pressure in the miniature anvil cells, reproduced from reference
 [21].
 2.2.1.1 CZ hand polished anvils
 Simpli ed manual polishing procedures have been developed for fabricating CZ anvils for GAC
 high-pressure experiments. To the best of my knowledge there have been no reports indicating
 the polishing of CZ anvils manually.
 17
2.2 High pressure methodology
 The manual polishing method was developed using guidelines similar to machine polishing.
 The hand held gem holder is a modi cation of the holder used in an industrial polishing machine
 as shown in the next chapter, section 3.2.1. The hand polishing technique produces very rounded
 anvil faces (and a rounded bevelled edge) seen by the circular Newtons rings on the CZ culet
 as opposed to the straight Newtons fringes seen on diamond anvils. This can be clearly seen in
  gure 2.13 and  gure 3.6 in section 3.2. The degree of rounding is given by the radius of the
 Nth Newton ring:
 rN = [(N  
1
 2
 ) R)]
 1
 2 (2.5)
 R is the radius of the lens through which the light passes, N is the bright ring number, and  
the wavelength of light. In this case R would give an indication of distortion of the culet surface
 that acts as the lens, i:e: if N is small and rN large it will imply that the lens (in this case the
 culet) is relatively  at.
 Figure 2.13: Interface between the CZ anvil and glass showing Newtons interference
 fringes - Circular Newtons fringes showing the degree of rounding of the CZ anvils viewed through
 one of the two opposing CZ anvils. The rings originate from the centre of the culet and are sym-
 metrical indicating that the anvils are parallel. The shadow or outline of both anvils overlap and
 indicate that the anvils are concentric to one another.
 2.2.2 The Merrill-Bassett diamond anvil cell (MB-DAC)
 Diamond anvil cells can be used for a variety of experimental measurements such as Raman
 spectrometry, IR spectrometry, magnetism and XRD measurements [17, 22]. The range of
 compounds that have been studied in the last 30 years is extensive in  elds such as chemistry,
 biology and physics and have included experiments on condensed gases, amino acids, proteins,
 18
2.2 High pressure methodology
 pharmaceutical compounds and various materials [22]. Jayaraman classi ed the design of DACs
 into  ve categories. Four are piston cylinder type cells, namely, the National Bureau of Standards
 cell, the Bassett cell, the Mao-Bell cell and the Syassen-Holzapfel cell [23]. The  fth cell, using
 a simple concept of thrust plates and driving screws, is called the Merrill-Bassett cell [23].
 Originally these 45 mm diameter cells would be used in single crystal experiments and would
 only reach pressures of about 5 GPa [23]. The MB type diamond anvil cell has been through
 various iterations and geometry changes but the basic mode of operation remains the same [24].
 Sterer et al: [23] noted that in conventional MB type anvil cells, the three screw design is
 only stable up to about 25 GPa and that the stability of the cell is highly dependant on the
 elastic limit of the screws. The modi ed design used in this study incorporates 6 allen thrust
 screws as shown in  gure 2.14 [23]. The MB-DAC is of the transmission geometry type and is
 one of the most widely used types of cells [22]. Originally the design of the MB cell was based
 on the constraints that it should  t on a standard goniometer head and be small enough to
  t the space requirements of a standard precession camera and a single-crystal orienter [22].
 This small and simple design consists of two platens that are pulled together by allen screws.
 The platens are located on posts for stability and are kept parallel by careful measurement.
 Vertical posts are included for two reasons, namely, to position the metal gasket for precise and
 consistent alignment when indenting, and secondly to position the top thrust plate and insure
 precise vertical movement. Three additional holes are located in both thrust plates so that the
 cell can be positioned on a specially designed torque lever for the pressure allen screws to be
 torqued. The MB-DACs are manufactured in the School of Physics (WITS) using INC718 (steel
 alloy) and have a Vickers hardness of 440 HV after annealing in a furnace at 718 C for 8 hours
 under hydrogen or argon.
 Anvils are mounted on the polished backing plates, which are inserted into the thrust plates.
 The anvils are aligned with their culets parallel and concentric. The cell is closed after a 100  m
 thick glass slide is inserted between the anvils. This is done to protect the anvil faces from
 coming into contact with each other which would cause chipping or scratching thus lowering
 the integrity of the culets. The interference fringes between the diamond anvils disappear when
 aligned, whereas the hand polished anvils give the circular Newtons rings originating in the
 centre of the anvil face. A metal gasket of SS301 (diamond anvils) and Ta90W10 (CZ anvils)
 is pre-indented up to a thickness of about 30  m and is sandwiched between the anvils [18].
 The gasket is drilled and the cavity is then loaded with a pressure transmitting medium along
 with the sample and a pressure sensitive marker [18]. Measurements are then taken through the
 19
2.2 High pressure methodology
 diamonds as the platens and backing plates have exit cones of between 80 and 30 depending
 on the experimental requirements. This is a versatile design that allows single crystal di raction
 experiments incorporating various materials and substances using laboratory based equipment
 and synchrotron sources [22].
 Figure 2.14: Exploded view of the MB cell - Cut-away exploded view of a modi ed MB cell.
 (1) Thrust screw, (2) thrust plate, (3) backing plate, (4) gasket (5), diamonds and epoxy, (6) bottom
 backing plate, (7) supporting posts and (8) lateral adjusting screw.
 2.2.3 Determining pressure within the pressure cell
 There are several methods of measuring the pressure in a DAC. One method is to include a
 material with a known pressure-volume equation of state (EoS) and the other is to load a pressure
 calibrant such as ruby chips/ balls (Cr3+ doped  -Al2O3) or MgO:V2+ into the gasket cavity
 along with the sample into the high pressure cell [17]. The lattice parameter of the marker can
 then be determined by XRD by tracking signi cant peaks and calculating the lattice parameter
 from the EoS. The ruby  uorescence R1 and R2 peaks can be measured and the pressure can
 be determined from the excitation wavelength using equations 2.6 and 2.7.Originally Russel [18]
 applied this idea of including a marker and by using the Decker EoS of NaCl up to 19.5 GPa
 to calibrate the ruby scale. The assumption is that both the marker and ruby are at the same
 pressure and then the line shift of the ruby  uorescence R1 taken along with di raction data
 can be used to calibrate the ruby scale [25].
 The ruby pressure scale is an easy way to determine pressure within the DAC and has been
 informally adopted within the high pressure community as a standard. Ruby has spectral lines
 20
2.2 High pressure methodology
 Figure 2.15: Ruby  uorescence R1 and R2 - R1 and R2 lines excited with an Ar ion laser
 (514.4 nm) showing pressures of 0 GPa (R1 at 694.31 nm) and 5.5 GPa (R1 at 696.37 nm).
 R, U , B and Y in the visible spectrum with the higher U , B and Y being able to be pumped with
 a laser which relaxes to the R lines R1 and R2 as shown in  gure 2.16 [25]. The R1 and R2 lines
 are metastable with long lifetimes that  uoresce down to the ground state [25]. Zero pressure
 represents  0 = 694:24 nm and  0 = 692:81 nm for R1 and R2 at 298 K, respectively [25]. The
 peak of the usually stronger R1 line is used for the calibration and pressure measurement. A
 zero pressure calibration is always performed for each experiment as the  0 value does not always
 match the quoted values as shown in  gure 2.15. The luminescence and absorption spectrum
 is shown in  gure 2.16. Piermarini et al: calibrated the R1 line shift against the NaCl EoS in
 1975 [25] and since then a few more calibrations have been necessary to extend the calibration
 to higher pressures.
 Figure 2.16: Electron-energy level term scheme - Luminescence and absorption spectra for
 ruby [24].
 21
2.2 High pressure methodology
 Piermarini et al: calibrated the wavelength shift against the Decker EoS of NaCl and found
 a linear dependency of
   
 P
 = 0:365 (2.6)
 Mao et al: [25] and Bell [25], using quasi-hydrostatic and non-hydrostatic pressure to megabar
 pressures to re-calibrate the ruby scale, found the empirical quasi-linear relationship
 P = A[(
  
 0
 )B  1]=B; (2.7)
 using A = 1904 and B = 7.665 (the slope and curvature parameters, respectively) for quasi-
 hydrostatic and B = 5 for non-hydrostatic conditions. Silvera et al: [25] have since calibrated
 the R1 line to 150 GPa using equation 2.7 with A = 1876 and B = 10.71.
 Even though the  uorescence method is ideal to do fast and easy pressure calibration be-
 tween 0.01-0.1 GPa, the precision is limited to about 0.02-0.05 GPa due to di erential thermal
 expansion of the components of the DAC and to time dependent mechanical relaxation within
 the gasket [24]. Pressure can be determined by means of an internal pressure marker such as
 NaCl or Pt Black and calculating the pressure from the unit-cell parameters using the materials?
 EoS. The Birch-Murnaghan equation of state is:
 P (V ) = 1:5B0[(
 V
 V0
 ) 7=3  (
 V
 V0
 ) 5=3] f1 + 0:75(B00  4)[(
 V
 V0
 ) 2=3  1]g (2.8)
 where B0, V0 and B00 are the bulk modulus, the atomic volume and the pressure derivative
 of the bulk modulus, respectively.
 Having a hydrostatic environment is ideal for the cavity contents and reliable pressure mea-
 surements need to be taken. The choice of pressure transmitting medium is determined via
 experimental parameters.
 2.2.4 Pressure transmitting medium (PTM)
 The identi cation of a suitable PTM is experiment speci c. This is considerably more di cult
 when using CZ anvils. There is a parabolic stress distribution across the anvil face under
 pressure, with the maximum stress at the centre of the anvil [18]. When a liquid hydrostatic PTM
 like ethanol:methanol (4:1) or glycerine is employed the pressure gradient is undetectable so the
 pressure distribution is a truncated shape across the gasket hole diameter [18]. When a PTM like
 NaCl and AgCl solids are employed, there is additional support across the sample and thereby
 22
2.2 High pressure methodology
 decreasing the truncation of the parabolic stress across the gasket hole diameter [18]. The
 decision on what PTM to use was based on my own experimental data using Raman spectroscopy.
 This is due to the anvils, PTM and the sample giving o Raman bands necessitating the PTM
 chosen to optimise the Raman signal and to minimise the interference from any Raman bands
 originating from the PTM. For the initial high pressure experiments, the PTM considered was
 Si oil (vacuum pump oil), Fluorinert FC70/FC77 (1:1) or NaCl  akes. The NaCl  akes look
 similar to the sample  akes and therefore makes discerning the sample  akes di cult. For the
 high pressure laser heating experiments, liquid Ar was used because it insulates the diamond
 anvils from the sample [26, 27].
 Figure 2.17: Cryogenic loading apparatus - Apparatus for making liquid Ar PTM using a bath
 of liquid N. The Ar gas  ows from the gas bottle through the coil submerged in liquid N. The Ar
 gas is lique ed and collected in a vessel for the MB-DAC to be submerged in.
 As seen in  gure 2.17, the high pressure cell can be loaded with liquid Ar (boiling point of
 87 K). Commercial welding argon (purity better than 99.996%) is lique ed by  owing the Ar
 gas through a heat exchanger coil submerged in liquid nitrogen (boiling point of 77 K). The
 cell is then lowered into the liquid Ar and the screws cracked open to reference positions and
 the PTM allowed to  ow in. The screws are then tightened. This low temperature process is
 tricky as it a ects the movement of the screws and the platens and it is not possible to view the
 process easily due to the slight convection of the liquid [26].
 If the cell is pressurized above 1.2 GPa, the Ar has already crystallized in the hexagonal
 close packed (h.c.p.) structure before being heated up [26]. If the cell is below 1.2 GPa further
 pressurization is required to crystallize the argon and it goes to the face centred cubic (f.c.c.)
 structure once the cell is at room temperature.
 23
2.3 Heating the sample in a MB-DAC
 2.2.5 Gasket materials
 The choice of gasket material as in the case of PTM is also highly dependent on the experimental
 setup. It depends on the type of anvils being used and also the analysis technique employed.
 Typically a metal gasket of 200  m starting thickness is used. Various gasketing materials that
 are suitable such as tungsten, rhenium, hardened stainless steel, martensitic tool-steel or alloys
 for example inconell or Cu-Be may be used [24]. CZ anvils require a softer gasketing material
 due to the possibility of breaking these anvils when pre-indenting. CuBe and Ta90W10 were
 explored because they are not considered very hard but are able to fully support the sample
 cavity. Xu and Huang [21] used T301 stainless steel gaskets of 0.125 mm thickness in CZ anvil
 cells successfully to only 7.8 GPa. Pb was also considered a possible gasketing material but
 reports show that it will only support the cavity to pressures of around 2 GPa [17]. Tungsten
 gasket has been tested to 50 GPa with the cavity diameter shrinking from 130  m to 75  m
 [28]. In extremely high pressure experiments rhenium gaskets are used in high pressure cells of
 up to 222 GPa [29].
 2.3 Heating the sample in a MB-DAC
 Three methods of heating the DAC are possible. A purpose built mini furnace for heating of
 the entire DAC under H2 or Ar gas and a CO2 laser heating facility were considered. The
 third method of heating a sample is to place resistive wires in the cavity but this method was
 not explored. Furnace heating results in the entire cell being heated and the behaviour of the
 cell due to the expansion of the metal body adds uncertainty to the pressure measurement,
 as the pressure measurement is not done in situ. The cell body is made from INC718 alloy
 (heat treated at 718 C) therefore furnace heating is limited to below this value. The anvils
 are typically glued to the backing plates with epoxy. Furnace heating is advantageous due to
 ease of measuring and controlling the temperature via the furnace thermocouple and the trivial
 placement of the cell within the furnace as compared with the challenging setup of the CO2 laser
 heating facility and DAC alignment.
 2.3.1 CO2 laser heating
 The wavelength of the CO2 laser is convenient to use due to the fact that we are able to focus the
 beam down to about 30-50  m on the sample and reach temperatures of up to about 3300 C [27].
 This is advantageous when using a cavity size of 100  m in diameter which is an average cavity
 24
2.3 Heating the sample in a MB-DAC
 size used in DAC experiments. This allows for heating of very speci c areas of sample while
 leaving other areas unheated that can be used as a reference in the cell. Most importantly, ZrO2
 and HfO2 absorb 10.6  m wavelength from the CO2 laser as metal oxides are strongly absorbing
 at the 10.6  m wavelength [27]. The CO2 laser is therefore suitable for heating the Zr0:5Hf0:5O2
 solid-solution. The attenuation is believed to be about 70% using the type Ia anvils (used in
 this study) compared with the highly transparent type IIa diamond anvil windows which are
 rare and expensive [27].
 Figure 2.18: Laser heating setup - The modi ed CO2 laser heating setup in the School of
 Physics.
 The CO2 laser heating setup is shown in  gure 2.18. The setup was modi ed from the setup
 mentioned in the paper by Hearne et al: [27]. The system consists of a co-linear laser of the
 30 W continuous wave CO2 laser and HeNe laser (yellow and red paths combined) due to the
 CO2 being invisible. The lasers are aligned by doing a near  eld and far  eld test by using heat
 sensitive paper. Laser light passes through the attenuator (calibrated between 0 and 100%).
 The convex lens expands the CO2 laser on to the CO2 focusing lens (ZnSe) mounted on a three
 axis stage and the light is then focused onto the sample through the copper mirror. The copper
 mirror has a hole drilled through it to allow the laser light through onto the sample and was
 newly installed. The light is then collected on the mirror (blue and green paths) and re ected
 through the camera zoom lens on to the bullet camera (blue and green paths) to be viewed on
 the monitor. A 50:50 beam splitter can be inserted before the camera to re ect light onto an
 25
2.4 Raman spectroscopy
 optical  bre for temperature measurements (not shown). The re ected light from the sample is
 not parallel to the incoming CO2 laser light and as the stage is moved horizontally the re ected
 light takes a di erent path. This system therefore displays the image moving horizontally across
 the monitor and can only be used for vertical alignment. This alignment problem was not
 discovered in the paper by Hearne et al: [27]. A second colour CCD camera (pink) was installed
 on the sample stage for easier alignment of the CO2 laser.
 From the phase diagrams of the end members in  gures 2.2 and 2.3, we can see the tempera-
 tures used to synthesize orthoI and orthoII were between 500 C and 1000 C. Using these phase
 diagrams as guidelines we hoped to achieve similar processing temperatures. On the current
 laser heating system two methods of temperature measurement are possible. The  rst involves
 a system of changing the angle of the polariser and comparing the resulting signal to a reference
 value [27]. The second, uses a spectro-radiometric method that measures grey body emission
 and the temperature extracted from the Planck black-body radiation formula. The Wien ap-
 proximation to the Planck formula is often used as the temperature is more easily extracted
 from the linear  t:
 J( ) = ln( ( )) !( )T 1; (2.9)
 where J( ) = ln(I  5=c1), I is the spectral intensity, !( ) = c2= ,  the emissivity and
 T the temperature. The constants are c1 = 2 hc2 = 3:7418  10 16 Wm2 and c2 = hc=kB,
 hc=kB = 1:4388 10 2 mK. Focus from the various optical elements can be determined easily
 using a substitute sample in place of the DAC by emission. However, the diamonds in the DAC
 act as lenses themselves and have di erent refractive indices requiring the focus to be calculated
 using the following formula:
 AD
 RD
 =
 n1
 n2
 (2.10)
 where RD is real depth, AD the apparent depth, n1 and n2 the indices of refraction of
 diamond and air, respectively. This formula was used to roughly calculate how far into the
 second diamond the laser focuses.
 2.4 Raman spectroscopy
 Raman scattering in the DAC provides a convenient way of investigating the e ect of pressure
 on optical phonons in crystalline solids experimentally to several hundred kilobars [15]. In this
 26
2.4 Raman spectroscopy
 section Raman Spectroscopy will be brie y introduced and information is summarized from texts
 by Dr Soss e Ndiaye [30], Dhamelincourt et al: [31], Jayaraman [15] and Paula [32]. Textbooks
 by Long [33] and Gregora [34] were also consulted and a more comprehensive review can be
 found therein.
 Raman spectroscopy is a type of vibrational spectroscopy. It is based on the inelastic scat-
 tering of light. Incident radiation is created with the use of a suitable laser source. Laser light
 produces a monochromatic, polarised narrow beam which has a well de ned wave vector kl.
 Light scattered in the direction ks is collected over a certain solid angle   with wave num-
 ber v. The bulk of the intensity would be due to Rayleigh scattering, where light of the same
 wavelength is elastically scattered by the sample. Weak lines to higher and lower energies would
 correspond to Raman scattering.
 Figure 2.19: Principle of Stokes and anti-Stokes - Visualization of the Stokes and anti-Stokes
 scattering processes [32].
 When a monochromatic light (eg: a laser), is shone on a material most of the light is trans-
 mitted or re ected elastically, however some inelastic scattering of the radiation is observed.
 Stokes Raman lines (~v0  ~vi) and the corresponding anti-Stokes Raman lines (shown in  gure
 2.19 (~v0 + ~vi) appear symmetrically on both sides of the Rayleigh line (~v0) with (~v0) and (~vi)
 the absolute wave numbers (~v = v=c). Raman bands can be regarded as unique  ngerprints
 that also give information about the local molecular environment. Suppose that a photon of
 a given energy (and wavelength) collides with a sample. When energy is transferred from a
 molecule to the photon, light of higher energy (shorter wavelength) will be scattered by the
 sample (anti-Stokes scattering). When energy is transferred from the photon to the molecule,
 light of lower energy (longer wavelength) will be scattered by the sample (Stokes scattering).
 Both the Stokes and anti-Stokes scattering are associated with transitions between rotational,
 vibrational and electronic levels [33]. The new wave numbers are called Raman lines or bands
 and constitute the Raman spectrum [33].
 For Stokes scattering,
 27
2.4 Raman spectroscopy
 ~vout  ~vin =   ~v = G(v + 1) G(v) +
 1
 2 c
 s
 kQ
  Q
 (2.11)
 and anti-Stokes scattering,
 ~vout + ~vin =  ~v (2.12)
 where kQ and  Q are the force constant and reduced mass of the normal mode, respectively;
 G(v + 1) and G(v) are vibrational terms for the v + 1 and v states; and  ~v is the so-called
 Raman shift, expressed in units of cm 1 (shown in 2.19). Molecules are in motion inside the
 lattice and consist of rotational, translational and vibrational modes.
 2.4.1 Vibrational energies and polarization e ects
 From quantum mechanics, only well de ned modes are allowed. These are the normal modes or
 phonons which are responsible for physical qualities in materials including thermal and electrical
 conduction (for example, the heat conduction in insulators or the propagation of sound in solids).
 Raman spectroscopy has consistently been shown to be a useful tool in the characterisation of the
 structure of metal oxides in both the bulk and the surface supported phases [8]. Visible Raman
 spectra are sensitive to terminal oxygen vibrations (M = O) [8]. These atomic displacements
 and frequencies are known as normal modes of vibration of the molecule. Figures 2.20 and 2.21
 show the various types of motion that contribute to the creation of a normal mode. Each normal
 mode, denoted by Q, can be described as an independent, multi-atom harmonic oscillator with
 allowed energies given by:
 GQ(n) = (n+ 1=2)~nQ (2.13)
 where
 ~nQ = 1=2 c
 q
 kQ= Q (2.14)
 and where GQ(n) is the energy of the oscillator in a state labelled by the quantum number
 n (n = 0, 1, 2, 3, ....), and ~nQ is the characteristic frequency (in cm 1) of the oscillator.
 In vibrational spectroscopy, electromagnetic radiation is used to excite the normal modes.
 At room temperature, most normal modes of large molecules may be aptly described with n = 0.
 28
2.4 Raman spectroscopy
 This means that, even in its lowest energy state, each of the normal modes of a molecule are
 moving with energy
 GQ(0) = GQ(1) GQ(0) = ~nQ (2.15)
 The ability of a molecule to be distorted by an electric  eld is referred to as its polarizability
  . The induced dipole moment is given by:
 ~uind(t) =   ~E(t) (2.16)
 If there is a change in the polarizability of a molecule it means there will be an active
 vibrational mode in the Raman spectrum. Bonds possessing covalent character tend to give
 strong Raman signals because of the change of polarizability associated with the vibration [35].
 Figure 2.20: Molecular vibrations - (a) 3N -6 vibrations for a non-linear entity of N atoms, (b)
 3N -5 for a linear entity of N atoms (for N=3 atoms).
 Figure 2.21: Molecular vibrational modes - Since 3N = 9, this leaves 3N -6 = 3 vibrational
 modes: (a) Vibration (q1) is called the symmetric O-H stretching mode. It can be formed by
 combining the stretching of the two O-H bonds so that expansions and contractions of the two bonds
 occur in phase. (b) Vibration (q2) is the bending mode. (c) Vibration (q3) is called the antisymmetric
 O-H stretching mode. It is formed by combining the expansion of one O-H bond with the contraction
 of the other O-H bond, i:e: out of phase.
 29
2.4 Raman spectroscopy
 2.4.2 Raman selection rules
 For a molecule with a centre of symmetry, vibrations with symmetrical vibrations are inactive in
 the IR and vibrations with anti-symmetrical vibrations inactive in Raman scattering [36]. The
 distortion of a molecule in an electric  eld, and therefore the vibrational Raman cross section,
 is determined by its polarizability.
 A Raman transition from one state to another, and therefore a Raman shift, can be activated
 optically only in the presence of non-zero polarizability and dependent on the symmetry of the
 vibration or rotation [36].
 Raman-active vibrations/rotations can be identi ed by using almost any textbook that treats
 quantum mechanics or group theory for chemistry. Raman-active modes can then be found for
 molecules or crystals that show symmetry by using the appropriate character table for that
 symmetry group.
 2.4.3 Confocal laser Raman micro-spectroscopy instrumentation
 The intensity of the Raman signal is given by
 S  I0  N T s (2.17)
 where I0 is the laser irradiance at the sample,   is the di erential cross section for the
 Raman line, N is the number of molecules in the probed volume V ,  is the solid angle of
 collection of the Raman light, T is the throughput of the instrument, and s the sensitivity of
 the detector at the wavelength  . Using microscope objectives for both the illumination and
 collection of light at the sample was found to be the best way to decrease V and increase  
and I0 [31]. By using high numerical aperture (NA) optics the laser beam can be focused to
 a small volume collecting under a wide angle. The micro-Raman spectrometer can focus the
 laser down to 5  m and thus small sample areas in the DAC can easily be probed. A vastly
 greater amount of local irradiance I0 can be collected, and due to the wide angle of collection,
  helps to compensate for the small number of molecules N [31]. Originally the micro-Raman
 spectrometer was conceived as a confocal microscope.
 The pinhole acts as a  lter to remove rings and noise and produces a clean point source
 which is then imaged in the plane of the microscope objective. The pinhole arrangement,
 as shown in  gure 2.22, ensures that only the light coming from the focal plane reaches the
 detector and that light from the upper and lower planes is partially attenuated. By using a
 30
2.4 Raman spectroscopy
 micro-Raman spectrometer it is possible to discriminate between di erent depths of the sample
 allowing \optical sectioning". This arrangement is useful in illuminating stray light from either
 the background or  uorescence which is outside the focal volume [31].
 Figure 2.22: Principle of confocal Raman spectroscopy - Comparison of conventional micro-
 scope with the confocal microscope setup [31].
 2.4.4 Using Raman spectroscopy in the pressure cell
 Optical spectroscopic techniques are well suited to probing materials under extreme conditions.
 Information from pressure induced shifts on inter-atomic binding forces in crystals, charge trans-
 fer e ects and structural instabilities is obtained from Raman spectra [15]. In the case of the
 GAC and DAC, measurements can be made quickly on samples with micrometer dimensions [9].
 Thermal gradients created by laser heating through the sample is less of a factor when using
 Raman spectroscopy because the bulk of a material is not probed by spectra taken from localised
 areas. Care must be taken to minimize the laser-induced heating of the sample as it may alter
 the surface structure of the bulk metal oxide [8]. However, from the phase diagrams by Ohtaka
 et al: (in  gure 2.2), the orthoI phase only starts at approximately 5 GPa for both ZrO2 and
 HfO2 and at about 500 C. Consequently this means that any laser heating from the Raman
 laser is negligible for the Zr0:5Hf0:5O2 sample. The advantage of using Raman spectroscopy over
 31
2.4 Raman spectroscopy
 IR spectroscopies is that even for non-crystalline compounds, the Raman bands are narrower
 than IR bands [37]. Other advantages include:
  Characteristic spectra for qualitative identi cation of compounds/phases.
  Relative lack of interference of adjacent peaks; aqueous solutions not a problem.
  Spatial resolution of  1  m (micro-Raman).
  No sophisticated sample preparation necessary.
  It is a non-destructive, in situ technique.
  It is relatively fast (spectrum acquisition time 1 sec to 2 min).
 Some disadvantages of Raman spectroscopy are:
  Inherent weakness of Raman e ect - can be overcome by intense sources (lasers) and
 e cient detectors (PMT/CCD).
  Fluorescence - can be suppressed by using longer wavelength excitation.
  Not all compounds are Raman active - need complementary technique such as IR spec-
 troscopy to access all vibrational modes.
  Can only probe bulk properties of materials which are transparent to incident laser wave-
 length (otherwise used as a surface or near-surface technique).
 Certain corrections must be taken into account because the observer is outside the crystal
 in real experiments. These corrections are due to refraction, re ection and transmission of the
 incident and scattered light at the interfaces and through the anvil windows [34]. The absorption
 of light by the sample must also be taken into account [34]. It is di cult to take all of these
 aspects into account experimentally and therefore the measurement of the absolute Raman
 intensities are rare [34].
 32
2.5 X-ray di raction
 2.5 X-ray di raction
 X-ray di raction is a technique that probes the structure of materials. It is a powerful technique
 to determine the phase changes of a material under extreme conditions. These changes can be
 closely monitored to get a good understanding of the nature of the material being probed. XRD
 is a well known technique and covered extensively in numerous sources. Information contained
 here is taken from a set of lecture notes by S. Verryn and the texts by B. E. Warren [38] and
 M. M. Woolfson [39].
 Figure 2.23: Electromagnetic spectrum - The electromagnetic spectrum (modi ed from Aus-
 tralian synchrotron).
 2.5.1 X-ray di raction theory
 X-rays are electromagnetic radiation with wavelengths of the order 0.01-10 nm. This radiation is
 produced from radioactive materials, electron beams, synchrotron sources, proton bombardment
 and photons as secondary x-rays. The laboratory based XRD machines use a tube with a hot
 tungsten  lament and electrons are accelerated by a voltage in a vacuum and then strike an
 anode made of molybdenum in the case of the Siemens SMART di ractometer. Large amounts
 of heat are produced as a by-product of the process (2% converted into x-rays). Various designs
 have overcome the challenges of producing x-rays in a tube by incorporating a rotating anode
 and constant cooling of this anode. Classical scattering theory considers the primary beam an
 electromagnetic wave with a perpendicular electric  eld exerting force on the electrons creating
 radiation in all directions with the same frequency as the primary beam and is called scattered
 33
2.5 X-ray di raction
 radiation. Experimentally this unmodi ed scattering and Compton modi ed (longer wavelength)
 scattering is observed. This scattering is incoherent because of the change in wavelength and
 produces a di use background. The unmodi ed scattering gives rise to Bragg re ections. The
 wavelength of x-rays are of the order of bond lengths between atoms and is therefore suitable
 for di raction measurements as illustrated in  gure 2.23, by looking at the region of ?soft? and
 ?hard? x-rays. X-rays are di racted o planes of atoms within the cell as if they were re ected.
 These re ecting planes have an inter planar distance d with an associated scattering angle (2 )
 that leads us to Bragg?s law (equation 2.18) below. The typical emission spectrum from an x-ray
 tube is shown in  gure 2.24 (a) with a typical Ni-K  lter. Figure 2.24 (b) shows the typical
 synchrotron radiation which highlights the high brilliance and the broad range of wavelengths
 available.
 Braggs law:
 n = 2dhklsin hkl (2.18)
 The d-spacing for an orthorhombic crystal with lattice parameters a, b and c, for example,
 can be determined by:
 dhkl =
 1
 p
 (h=a)2 + (k=b)2 + (l=c)2
 (2.19)
 where h, k and l are the Miller indices which specify the re ection plane (a peak in a
 di raction pattern).
 A powder di raction pattern consists of a number of rings (as collected via a CCD detector
 or position sensitive detector). There are numerous randomly orientated crystals (in nite orien-
 tations assumed, giving Debye rings on the Edwald sphere in  gure 2.25) for any set of planes
 with spacing dhkl that satisfy the Bragg equation 2.18. This is shown in  gure 2.25.
 On a laboratory based system the collimator is placed close to the sample which results in
 a more divergent beam than if placed further away but resulting in the intensity of the beam
 decreasing. The solution on our system is the inclusion of monocapillary optics.
 2.5.2 Monocapillary optics
 X-rays are very e ciently re ected o glass surfaces when the angle of incidence is at or below
 grazing incidence [41]. Monocapillaries function on the principle of total internal re ection of
 34
2.5 X-ray di raction
 Figure 2.24: XRD spectrum - XRD spectrum from laboratory based di ractometer showing the
 e ect of a Ni-K  lter as the dotted line, (b) typical synchrotron emission spectrum [40].
 Figure 2.25: X-ray di raction from a powder sample - X-rays di ract o a powder sample
 as cones on to the CCD detector or  lm (image modi ed from lecture notes).
 35
2.5 X-ray di raction
 the x-rays o the glass surface. Monocapillaries are carefully shaped ellipsoidal glass and for
 single-bounce optics are nearly 100% e cient [41].
 Figure 2.26: Monocapillary glass optics - Principle of single-bounce ellipsoidal glass monocap-
 illary [41].
 The solid angle from the x-ray source to the capillary is relatively large therefore coupling
 between the two is e cient. Very high intensities of x-rays can be channelled into the capillary
 and can be used to generate high intensity micro-beams with diameters as small as 5  m, with
 the disadvantage being line widths are increased by  10% [41]. Monocapillary glass optics
 display near perfect collimation [42].
 2.5.3 Crystallography
 Materials are commonly found in the crystal state. Crystals are regular shaped with a high degree
 of symmetry that possess considerable variation in size and shape. A crystal is a 3-dimensional
 periodic repetition of a structure. This can be represented as a lattice which is a framework on
 which the crystal is built. There are 14 Bravais lattices within 7 crystal systems namely, cubic,
 triclinic, monoclinic, orthorhombic, tetragonal, trigonal and hexagonal. The bravais lattices are
 made up of unique unit cells which are described as the smallest building blocks to complete
 the entire lattice and have lattice types: primitive side-centred, face-centred, body-centred, and
 rhombohedral lattice types. There are 32 crystal systems. There are 230 space groups which
 are a description of the total symmetry of a crystal.
 2.5.4 X-ray di raction at extreme pressures in practice
 The sample volume size for single crystal di raction studies is not signi cantly larger than the
 sample size loaded into the gasket cavity in a DAC and therefore the DAC is ideally suited
 for XRD studies [24]. The Merrill-Bassett cell made routine XRD studies possible due to its
 compact and simplistic design [24].
 X-rays di ract o lattice planes in solid materials so a clever design is needed for the DAC
 so that the x-ray beam is free of other solid materials besides the sample during a measurement.
 36
2.5 X-ray di raction
 Diamond is a single crystal so when a measurement on diamond using XRD is performed the
 resulting pattern returns intense spots which are easy to mask out of the powder ring pattern
 obtained from the sample. The second consideration is the design of the diamond supporting
 plates, or backing plates. One method is to use solid beryllium backing plates; these plates
 however di ract along with the sample and can prove tricky in the case where the beryllium
 lines can?t be collimated out on a CCD [22]. The beryllium lines also become more intense
 and ?spotty? in synchrotron patterns [22]. Using steel backing plates also has challenges because
 steel is opaque in nature to x-rays. Wide opening angles must be made in the backing plates,
 requiring larger anvils which are more expensive and the anvil support is also weakened thus
 lowering the maximum pressure obtainable [22]. Another beryllium-free design is to make the
 backing plates out of a type Ib synthetic diamond [22]. Brister [43] notes  ve limitations af-
 fecting XRD experiments at extreme pressures. Firstly, the x-ray beam must travel through
 approximately 5 mm of diamond that absorbs low energy x-rays of about 15 keV and below.
 Secondly, the e ect of Compton scattering from the diamonds as the incoherent scattering from
 carbon dominates the total cross section over most of the useful x-ray energy range. Thirdly,
 the exit aperture of the diamond backing plates limits the available access to reciprocal space.
 Previously this has been overcome by choosing a slit for the exit aperture but only a satisfac-
 tory solution when doing energy dispersive di raction and is problematic for angle dispersive
 di raction. Non-hydrostatic pressure environments are the fourth limitation of the MB-DAC
 but hydrostatic pressure transmission mediums help to reduce or eliminate the pressure gradient
 over the diamond anvil face. Lastly, the small sample sizes and the corresponding x-ray beam
 must be matched, requiring the precise positioning of the monocapillary  tted to the SMART
 di ractometer. Type IIa diamonds would be preferable due to problems with optical viewing
 with type Ib, but are far more costly [22]. Particular to this study is the use of the novel
 Boehler-Almax (BA) anvil and special backing plate design shown in  gure 2.27.
 The BA anvils have the advantage of a larger girdle of 4 mm compared with typical brilliant
 cut anvils of 2.5 mm. This is to take advantage of the full 80 opening angle of the BA anvil [22].
 The brilliant cut and BA anvils have a similar heights of 1.70 mm and 1.95 mm, respectively
 and path lengths are therefore comparable to standard brilliant cut diamond anvils [22]. The
 design of the BA anvil backing plate design incorporates a backing plate that is 3.7 mm thick to
 compensate for the BA anvil being partly embedded in the backing plate shown in  gure 2.27,
 also the exit cone is ground to between 40 and 90 . The anvil seat is ground to 60 to match
 the 60 conical shape of the BA anvil.
 37
2.5 X-ray di raction
 Figure 2.27: Boehler-Almax anvils - BA anvils and backing plate used for XRD wide exit [44]
 These wide exit angles are extremely useful on a laboratory based machine as the wavelength
 typically used is that of Mo-K (0.707109  A). This wavelength is not tunable as is the case with
 a synchrotron sourced wavelength allowing  30 (2 ) maximum scattering angle to be measured
 on the laboratory based SMART x-ray di ractometer (determined experimentally).
 Figure 2.28: Siemens SMART di ractometer - Modi ed Siemens SMART x-ray di raction
 system  tted with a 0.1 mm monocapillary.
 Part of the study was to perform preliminary laboratory based experiments on the Siemens
 SMART di ractometer ( gure 2.28) before envisaged synchrotron-based experiments. This
 included loading the cell, aligning the cell with respect to the x-ray beam and analysis of
 powder patterns in situ using the European Synchrotron Radiation Facility (ESRF) software
 FIT2D [45].
 38
3
 Experimental
 In this chapter, the experimental techniques employed during this study will be described. The
 chapter has been divided into the following sections: sample preparation in section 3.1, high
 pressure experiments in section 3.2 and high temperature experiments in section 3.3. These
 sections include the techniques employed in developing the manual fabrication of cubic zirconia
 anvils, setting up of the Merrill-Bassett miniature pressure cells, anvil mounting, gasket inden-
 tation and sample loading. Experimental procedures of confocal Raman spectroscopy through
 the CZ anvils using two sample types in conjunction with preliminary MB-DAC measurements
 are discussed. Finally MB-DAC high pressure/temperature experiments using the CO2 laser
 heating system and set up is discussed. The phase diagram of the Zr0:5Hf0:5O2 system was
 explored with XRD in conjunction with Raman spectroscopy.
 3.1 Sample characterization
 The sample Zr0:5Hf0:5O2 was prepared using the sol-gel process and is well documented in
 Mojaki?s thesis [8]. It was received in powder form and no further preparation was required.
 The characterization of samples is extremely important in the investigation of materials and
 their properties. There are various techniques available for sample characterization, with the
 technique selected dependent on the information required. In this study, Raman spectroscopy
 and XRD was used.
 39
3.2 High pressure experiments
 3.1.1 Raman spectroscopy characterization
 Raman spectroscopy was used to establish whether the solid solution of zirconia and hafnia were
 indeed the same powder used in the previous study ( gure 4.1). The 514.5 nm line of a coherent
 argon ion laser was used, with an interference  lter imposed to remove the plasma lines. A power
 source of 600 mW, reduced to approximately 10 mW on the sample, and a focused spot size
 of about 5  m was used. The confocal pin hole of the apparatus used for powder samples was
 0.2 mm. A Jobin-Yvon T64000 Micro-Raman spectrometer with BX40 Olympus microscope
 attachment,  tted with X20 and X50 ultra long working distance (ULWD) objectives was used
 and spots of about 5  m could be measured.
 3.1.2 XRD characterization
 The Brukker D8 di ractometer (Cu-K ) was used to determine the ambient XRD di raction
 pattern ( gure 4.12). Some preliminary internal pressure calibration studies were also performed
 with this instrument but the idea was ultimately discarded due to the broad nature of the peaks
 from the SMART di ractometer. The scanning range was 25-55 (2 ) using the step size of
 0.014 and step time of 7327.2 s at room temperature (25 C). This ambient pattern was used
 to compare against the end members ZrO2 and HfO2. The low pressure XRD data measured in
 the MB-DAC was also compared with the ambient pattern.
 3.2 High pressure experiments
 Static high pressure experiments were carried out using the MB type anvil cell. The exploded
 view of the cell is shown in  gure 2.14 in section 2.2.2 and the cross section is shown in  gure 3.1.
 Initial experiments were performed using CZ anvils polished using a manual fabrication method
 developed in our high pressure laboratory at the School of Physics at WITS. The sample was
 then subjected to even higher pressures using the MB-DAC loaded with one Boehler-Almax
 anvil and one brilliant cut diamond anvil. High temperatures were achieved by using the in-
 house CO2 laser heating system. The samples were then analysed with Raman spectroscopy
 and XRD.
 3.2.1 Manual fabrication of CZ anvils
 For the gem anvil experiments brilliant cut CZ single crystal gems of 170 mg ( 0.85 carat) with
 a 5 mm girdle obtained from commercial suppliers were used. The composition of the anvils was
 40
3.2 High pressure experiments
 Figure 3.1: Merrill-Bassett cell - Cross section view of the miniature MB-DAC with a total
 diameter of 22 mm [23].
 CZ stabilized with 10-mol% Y2O3. The program of hand polishing was continued after initial
 experiments conducted several years ago by Professor G. R. Hearne. CZ anvils had previously
 been machine polished in the School of Physics at WITS.
 The hand polishing technique required three steps. The same polishing mats and lubricants
 are used in the hand polishing method as in the machine polishing method. These comprise of
 SiC 4000 grit polishing mat, UltraPrep self adhesive Diamond Lapping Film of 1  m, 0.5  m
 and 0.01  m, MetaDi Diamond Suspension monocrystalline spray and MicroCloth self adhesive
 obtained from Apollo Scienti c cc. [46]. The polishing mats are stuck to a glass sheet to create
 a su ciently  at polishing surface.
 Referring to  gure 3.2, the CZ (iv), is mounted in the holder as shown. The gem is mounted
 with the table of the CZ on the stainless steel adjusting screw (i), (polished parallel to the
 stainless steel polishing ring (ii)). The CZ can be adjusted up and down by turning this screw.
 The CZ is clamped on the top by a locating plate (iii) that is tightened by the three screws.
 The CZ needs to be constantly adjusted up or down depending on the degree of rounding and
 the required  nal anvil size. With the tip of the gem protruding it is blunted o using a SiC
 4000 grit polishing mat following a  gure of eight or forward-back pattern. Constant downward
 pressure on the anvil and holder is required for a high quality, symmetrical culet. The tip
 must be adjusted so that it does not cause the holder to wobble or spin on the mat. This  ne
 adjustment achieves very precise parallel polishing of gem tip with respect to the table. The
 culet can be viewed in the holder through the hole (v) in the adjusting screw (i).
 41
3.2 High pressure experiments
 Figure 3.2: CZ anvil polishing holder - CZ anvil polishing holder showing the top, bottom and
 cross-section (labels de ned in the text).
 Figure 3.3: CZ culet  nal soft cloth polish - Edges polished round by the 1  m soft cloth.
 42
3.2 High pressure experiments
 The culet is polished with the SiC 4000 grit polishing mat until it is 20-40  m larger than
 the desired culet size. The soft cloth is used to round the edges of the culet octagon giving
 a rounded bevel shape to the anvil face similar to the bevel on diamond anvils as shown in
  gure 3.3. The next step is to polish using a 1  m diamond impregnated polymer polishing mat
 keeping a constant downward pressure. It is necessary to alternate between the 1  m polishing
 mat and the soft mat with diamond lapping spray to round o the corners and edges of the
 anvil. This technique is used to relieve the stresses at the edges as they tend to crack and chip
 and thus contaminate the polishing mats with hard CZ chips which scratch the anvil face during
 subsequent polishing. If the  nished CZ anvil has any defects such as chips and scratches there
 is a likelihood of the anvil failing at lower pressures.
 The 0.5  m and 0.1  m polymer polishing mats are used to polish out any micro-scratches
 and to give a higher quality  nish to the culet. When the culet is viewed under a microscope
 at 20X magni cation, there should not be any visible scratches. The culet face should be fully
 illuminated without optical shadow e ects when adjusting the light level. If asymmetric shadow
 e ects are observed, it could indicate non-uniformity on the culet face. A  nal test of parallelism
 and circular symmetry of the culet is to remove the  nished anvil from the polishing holder;
 clean it thoroughly and try balance the anvil on the culet face on a glass slide and view the
 Newton interference fringes through a microscope as illustrated in  gure 3.4.
 Figure 3.4: CZ anvil balancing test - CZ anvil balancing on the culet. The circular Newtons
 rings are very nearly centred and this anvil displays the characteristics of a well polished anvil.
 If the anvil is of satisfactory quality and  atness it will balance on the culet and secondly the
 interference fringes will be circular and originating from the centre of the culet as in  gure 3.4.
 If the rings are broad, the anvil has a relatively  at surface and if there are many  ne rings, the
 anvil face is relatively rounded (see equation 2.5). Indications of a poorly polished anvil is that
 43
3.2 High pressure experiments
 the anvil does not balance as a  rst test and second if it lists and balances o centre creating
 more oval shaped interference fringes.
 3.2.2 Diamond anvils
 Diamond anvils are typically manufactured by Almax industries [44]. The top anvil used was a
 Boehler-Almax anvil of 0.55 carats and a 5 mm girdle of type Ia polished with an exit cone of
 83 , and a culet size of 550  m with an 8 bevel. The bottom anvil was a brilliant cut diamond
 of 0.55 carats and 5 mm girdle of type Ia, with a culet size of 650  m with an 8 bevel.
 3.2.3 Anvil mounting and alignment
 The backing plates (BP) were polished by hand to a mirror  nish and have 1 mm exit holes
 opening up in a cone of 60 for the top anvil and 80 for the bottom BA anvil. A holder ( gure
 3.5) clamps the anvil on the top with a screw (i) tight onto the backing plate (iii), and was then
 centred with three lateral adjustment screws (iv). The anvils, whether gem or diamonds, were
 mounted on the backing plates with a high temperature epoxy; Stycast (mixed with catalyst
 9) and is stable to  200 C. The epoxy was de-gassed after mixing with the catalyst using a
 vacuum pump. By viewing from the bottom, the backing plate was moved such that the hole in
 the BP aligns with the culet by changing the focus on the microscope. The culet and the 1 mm
 hole must be concentric. The use of an inverted microscope such as the Olympus BX70 helps
 to achieve this, or a vertical telescopic system can be constructed using cross hairs or a similar
 arrangement can be used to get the culet and BP hole concentric.
 The thrust plates (TP) with the BP secured on them were brought together with a glass
 slide between the anvils. Figure 3.6 shows the Newtons interference fringes (broad and  ne
 fringes between the two separate glass/diamond interfaces). The fringes must disappear for
 both interfaces with the culets concentric by adjusting the lateral screws of the moveable BP
 and the three pressure generating thrust screws (that tilt the TP).
 With the parallelism and concentricity of the culets completed, the screws were marked with
 a reference position. The height of the TP was measured at each thrust screw and noted as
 the reference of parallelism and concentricity (which was used when indenting the gasket and
 generating pressure). A  nal check that the anvils were polished parallel to their respective
 tables, was to recheck that these thrust screw height measurements were all within  150  m of
 each other and were equally tight.
 44
3.2 High pressure experiments
 Figure 3.5: Gem alignment holder - Gem alignment holder for centring anvils on the backing
 plate. (i) Clamping screw, (ii) clamping screw lock nut, (iii) backing plate seat, (iv) lateral adjusting
 screws to centre backing plate on culet.
 Figure 3.6: Newtons interference fringes for the diamond-culet/glass interface - Straight
  ne and broad fringes of the ( at) diamond/glass interface of each anvil indicating that the culets
 are not parallel. The anvil culet diameters are 550  m for the brilliant cut diamond and 650  m for
 the BA culet.
 45
3.2 High pressure experiments
 3.2.4 Gasket fabrication
 The glass slide was removed and a metal gasket of approximately 200  m sandwiched between
 the anvils and the screws closed to the reference positions mentioned above. Initially stainless
 steel 301 (SS301) was used with the CZ anvils but the anvils cracked in the indentation phase.
 CuBe was then used but determined to be too soft, as the cavity deformed at pressures below
 10 GPa. Experimentally Ta90W10 was found to be suitable. The gasket material was cut to
 about 0.25 cm2 and  xed parallel to the culet using non-permanent sticky clay or positioned
 on the vertical posts. The gasket was marked so that it can be returned to the same position
 after the sample cavity is drilled. During the pre-indentation period, the anvils are most likely
 to break due to the radial extrusion of the gasket material putting maximum tensile stress on
 the anvils? tip [24].
 The recorded relative heights are kept constant during the indentation of the metal gasket.
 It was indented at 5 to 15  m intervals and pressures of up to 2 GPa are possible [23]. It was
 also required that the gasket be left for tens of minutes to let the stresses in the metal relax,
 or the cell opened periodically during this indentation process. A hole was then spark eroded
 in the centre of the indent. The diameter was determined by the maximum pressures needed.
 The bigger the ratio between the cavity diameter and the culet (indent) diameter, the higher
 the pressures the gasket can support [24].
 Figure 3.7: Spark erosion machine - Spark erosion machine used to start the hole in the gasket
 material after which a pin vice and micro drill are used to clean up the hole.
 The gasket was inserted centrally in the clamp of the spark erosion machine ( gure 3.7) so
 that it was visible on both cameras. The needle was centred on the indent. Dielectric liquid of
 46
3.2 High pressure experiments
 para n or alcohol was dropped in the indent. The coarse screw was used to lower the needle to
 within a few millimetres of the indent and then the  ne adjustment screw used for sparking. The
 gasket can be removed and inspected at various stages and corrections made. The cavity was
 then cleaned out and  ne tuned with the use of a micro-drill bit of between 100-200  m ( gure
 3.8). Once the gasket was drilled it was inspected to ensure the cavity was drilled centrally
 on both sides of the gasket. The hole can be spark eroded before the  nal indent thickness is
 achieved as with further indenting the hole will close and relieve some of the stresses on the
 anvils. In our experience, a cavity with a  nal diameter of about a 1/3 of the diameter of the
 gasket indent, is optimal.
 Figure 3.8: Gasket indentation - SS301 stainless steel foil of starting thickness 200  m. Indent
 with 140  m diameter cavity sparked and drilled.
 At this stage, the parallelism and concentricity of the anvils was rechecked and any new
 references noted. The cell has undergone pressure during the indenting stage and the thrust and
 backing plates have had a chance to seat themselves in their  nal positions. It was expected
 that there would be a small change in reference heights.
 3.2.5 Sample and ruby loading
 Sample powder was collected on a surgical needle and placed in the indent of the gasket (with
 the top TP removed) while monitoring through a microscope. The top TP assembly was then
 slid onto the posts and clamped down between  ngers thus pressing the powder into  akes in
 the indent. The TP was then removed and pieces of sample  akes manoeuvred into the cavity,
 after the excess powder was cleaned away. Usually 2-3 ruby balls are loaded onto the centre of
 the top anvil face. It is di cult to control where the ruby balls and sample  ake end up after the
 47
3.2 High pressure experiments
 PTM is loaded into the cavity and sometimes several loading attempts are necessary to ensure
 a satisfactory distribution of sample  akes and ruby balls.
 3.2.6 Pressure transmitting media (PTM)
 The  nal experiments relied on three types of PTM. These were Si-oil for the TiO2 experiment,
 Fluorinert (FC70/FC77, 1:1) for the TiO2 and Zr0:5Hf0:5O2 sample loading experiment in the
 MB-GAC. Fluorinert was used in the 20 GPa experiment and liquid Ar for the MB-DAC CO2
 laser heating experiment in the MB-DAC. The Si-oil and Fluorinert are liquids that are loaded
 with a needle and dropped into the cavity. Argon is a gas at room temperature but must be
 loaded in its liquid form. To achieve this, the Ar gas was lique ed using liquid nitrogen (N).
 The boiling temperature of Ar is 87 K and that of N is 77 K [26]. The method is described in
 section 2.2.4. The DAC was pre-cooled in liquid nitrogen and then the liquid Ar poured into
 the vessel. The Ar droped into the vessel and gently creeped into the sample cavity. Wittlinger
 [26] states that only weak convection arises in the liquid Ar which is necessary if the ruby and
 sample are to remain undisturbed and not washed away from the diamond surfaces. After the
 PTM was loaded, the cell was closed to the reference marks to increase the pressure to between
 0.5-1 GPa to ensure the PTM was sealed in the cavity. This was checked via ruby  uorescence.
 3.2.7 Pressure measurements
 Using the standard ruby  uorescence technique, the pressure was evaluated by monitoring the
 R1  uorescence line shown in  gure 3.9. Initially three thrust screws are employed to create
 pressure in the MB cell. After the cell was closed to seal the PTM in the cell, the pressure was
 between 0.5-1 GPa. The screws were tightened slowly to protect the anvils from failing. The
 pressure response to tightening the screws and screw reference positions were closely monitored.
 By experimentation it was determined desirable to increase pressure in steps of about 1 GPa.
 Constant checking was important at higher pressures, especially when using CZ anvils. A
 20 mW, 532 nm green laser was used as an excitation source. This was coupled to an Ocean
 Optics, USB-2000 or HGR-4000, spectrometer and optical  bre to the trinocular turret of the
 Olympus BX60 compound microscope. At ambient conditions (in our laboratory) theR1 line was
 located at 695 nm. The  uorescence was collected via a 200  m optical  bre and displayed using
 the software package Ocean Optics Spectra Suite. Two additional internal pressure calibrants
 were tried, NaCl and Pt Black, with the Zr0:5Hf0:5O2 sample using XRD. It was found that it
 48
3.2 High pressure experiments
 Figure 3.9: Ruby  uorescence microscope - A 20 mW, 532 nm green laser excitation source,
 Ocean Optics USB-2000 or HGR-4000 spectrometer, coupled to a 200  m optical  bre to the Olympus
 BX60 compound microscope.
 49
3.2 High pressure experiments
 was di cult to get quantitative data from the di raction patterns using the current set up as
 the peaks were too broad. The idea was consequently discarded.
 3.2.8 Gem anvil cell (GAC)
 Initial high pressure work was carried out by static compression in the MB-GAC. The orthoI
 phase of Zr0:5Hf0:5O2 starts evolving at 5 GPa . Using static compression in the MB-GAC in
 conjunction with the hand polished CZ anvils it makes for a good initial experiment on the
 Zr0:5Hf0:5O2 sample. Pressures of up to 10-15 GPa were anticipated with this set up considering
 the preliminary experiments to  13 GPa. The preparation of the CZ anvils in the MB cell is
 similar to the method described in sections 3.2.3 to 3.2.7 above for the MB-DAC. Initial Raman
 measurements were performed on the strong scatterer TiO2 using the confocal pinhole in triple
 subtractive mode to determine if the CZ anvils created a signi cant detectable background. The
 laser was focused inside the CZ anvil and on the TiO2 sample and compared. Si-oil was used
 as the PTM. Pressure generated on imperfectly polished CZ anvils reached 5 GPa. Further
 experiments were done at ambient conditions on a glass slide. Measurements on Si-oil and
 Fluorinert were compared. TiO2 and Zr0:5Hf0:5O2 Raman intensities were compared as well as
 the Raman spectrum of the CZ anvils themselves. These experiments helped show the possible
 Raman bands that may contaminate the spectrum of our sample at pressure. The experimental
 parameters of the Raman spectrometer were also explored and gave valuable information to use
 in the pressure experiments. In the  nal MB-GAC experiment both TiO2 and Zr0:5Hf0:5O2 were
 loaded into the same cavity as shown in  gure 3.10.
 Figure 3.10: Two samples TiO2 and Zr0:5Hf0:5O2 in the MB-GAC - TiO2 and Zr0:5Hf0:5O2
 loaded into the same cavity.
 50
3.3 High temperature experiments
 3.2.9 Diamond anvil cell (DAC)
 The MB-DAC (W302 #1) was loaded with the BA anvil with culet size of 550  m with 8 bevel.
 The brilliant cut diamond was mounted on the bottom BP and had a culet size of 650  m with
 8 bevel. For the second run using liquid Ar as the PTM and laser heating experiments, the
 INC718 #1 MB-DAC was loaded with a BA anvil on the top backing plate (BA anvil still
 mounted on W302 #1 backing plate used) and the brilliant cut diamond on the bottom BP. A
 stainless steel SS301 gasket of  200  m starting thickness was used in both cases for the XRD
 pressure measurements and a  200  m cavity was drilled. The side view of the cell is shown
 in  gure 3.1 and the cooling of the MB-DAC shown in 3.11 after it had been loaded with the
 liquid Ar PTM.
 Figure 3.11: 22 mm diameter MB-DAC - MB-DAC cooling after liquid Ar loading.
 3.3 High temperature experiments
 The two methods available for heating the MB-DAC were furnace heating and laser heating.
 It was decided to laser heat the cavity to avoid damage to the diamond anvils and the glue.
 The INC718 cell is designed to handle temperatures of  700 C. Liquid Ar was loaded into the
 160  m cavity is shown in  gure 4.11 to insulate the sample from the diamonds.
 3.3.1 CO2 laser heating the MB-DAC
 The system was aligned to try to focus the beam on speci c parts of the cavity. The heated
 regions were measured with Raman spectroscopy. The reference positions (unheated by the
 laser) displayed no clear evidence of heating between the heated and unheated regions. The
 entire cavity was then heated by raster scanning because speci c areas could not be reheated
 51
3.3 High temperature experiments
 once the cell was moved. The advantage of heating the entire cavity was that XRD measurements
 could be performed and qualitative data obtained for direct comparison to the  rst high pressure
 run.
 Figure 3.12: CO2 laser heating facility details - (i) 30 W continuous wave CO2 laser, (ii)
 attenuator and HeNe laser behind it, (iii) ZnSe CO2 focusing lens on three-axis stage, (iv) copper
 re ecting mirror with hole cut in for the laser to pass through and zoom lens, (v) bullet camera with
 polariser and (vi) MB-DAC mounted in water cooling jacket.
 Figure 3.12 shows the laser heating system. It consists of a co-linear 30 W continuous wave
 CO2 laser and HeNe laser that are combined by the beam combiner. The CO2 laser was focussed
 onto the sample by a ZnSe lens mounted on a three-axis stage. The alignment of the sample
 was achieved by using the test sample (Al2O3). The test sample was heated until emission
 was reached so that this emission spot could be monitored with both the cameras. The copper
 mirror is used to re ect the laser light through to the \re ection camera" as marked in  gure
 3.12. A 50:50 beam splitter can be  tted that allows half of the emitted light to the camera
 and half to the optical  bre for temperature measurements. Temperature measurements were
 unsuccessful due to hot spot alignment problems. The temperature was estimated from sample
 melting experiments done outside the DAC cavity. The zoom lens achieves a magni cation
 on the monitor of  220X. The second colour CCD camera was installed on the sample stage
 utilising the built-in lens giving a modest magni cation of about 10X. This was further magni ed
 52
3.3 High temperature experiments
 by zooming the picture via the software. The position of the emission spot was marked on both
 of the screens as a reference. The DAC was then placed on the stage and the sample cavity
 illuminated with the cold light (light transmitted through bundled  bre optic glass cables [47])
 and the laser reference spot and cavity moved to be on top of each other by means of the sample
 stage. This ensured that the laser beam was positioned on the cavity. Con rmation that the
 system was correctly aligned was that the camera seemed to pick up radiation from the CO2
 laser through the cavity and it was then a matter of observing this spot on the screen and raster
 scanning over the cavity. The sample stage is water cooled and the MB-DAC is constantly
 cooled to avoid damaging the type Ia diamonds that have high thermal absorption at 10.6  m
 wavelength. The absorption by the diamond anvil window (type-Ia, 2 mm thick) could be as
 high as 70% [48]. There was no emission during the laser heating experiment and it is believed
 that temperatures lower than 1000 C were achieved. The 10.6  m wavelength of the CO2 laser
 beam was focused to  50  m into the second diamond but attenuated by the sample and cavity
 contents. This set up results in the laser not being focussed into the front diamond, avoiding
 damage.
 3.3.2 High pressure Raman spectroscopy
 The Raman e ect is described as frequency shifts that are either positive (resulting in anti-Stokes
 Raman lines) or negative (resulting in Stokes Raman lines) when light is scattered by molecules.
 The Raman shifts correspond to vibrational energy changes in the scattering species and there-
 fore display characteristic  ngerprints. A Jobin-Yvon T64000 Micro-Raman spectrometer with
 BX40 Olympus microscope attachment, holographic gratings (1800 grooves mm 1) and a liquid
 nitrogen cooled CCD detector with a 514.5 nm laser line of Ar+ with output of 55 mW was used.
 The darker regions of the sample in  gure 4.11 are labelled (A), (B), (C), (D) and (E) and gave
 the best signal to noise ratio when measured by the Raman laser. The sample was speci cally
 pressed thin (10-20  m thick, and near transparent), for e cient heat dissipation when laser
 heating. There is a  ne balance between heat dissipation and being able to measure the thin
 sample with Raman and XRD. In Raman spectroscopy, most of the probing radiation was in
 fact transmitted rather than back-scattered. The diameter of the laser spot at the sample was
 5-10  m. The entrance slit width of the pre-monochromater was set at 150  m. The confocal
 pinhole was set to between 100-200  m to reduce the depth of the focal plane region as much
 as possible and thus to minimise sampling background signal from the CZ anvil. Under these
 conditions data acquisition times for each spectrum was between 5-8 minutes for the CZ anvils
 53
3.3 High temperature experiments
 and 8 minutes in the case of the laser heated MB-DAC. The maximum Raman laser power was
 set between 600 and 700 mW.
 3.3.3 High pressure XRD
 XRD data has been obtained using the modi ed Siemens SMART equipped with a 1K CCD
 detector based in the structural chemistry group in the School of Chemistry at WITS. The
 system consists of a conventional x-ray generator with Mo-K radiation ( = 0.71073  A). The
 beam was directed normal to the diamond table and culet through a 0.1 mm monocapillary.
 Angle-dispersive di raction patterns were in the form of Debye-Scherrer powder rings recorded
 on the 512 x 512 pixels of the area detector. A special stage was built for the SMART x-ray
 di ractometer. It has three-axis manipulation as well as tilting and rotation of the MB-DAC as
 shown in  gure 3.13. MB-DAC is positioned in a \V" such that the cavity/sample of the MB-
 DAC is in the same place as if a crystal sample is mounted on the standard goniometer. This
 allows for correct sample to detector distance determination by means of the built-in viewing
 telescope. The cavity was then further manipulated by optimizing the count rate through to
 the CCD by adjusting the MB-DAC horizontally and vertically.
 A new 0.1 mm monocapillary was attached to the x-ray tube and the beam is totally in-
 ternally re ected and focused into a narrow beam. The ?OPTICAL? setting on the SMART
 [49] software rotates the goniometer and stage 30 2 so that it was aligned with the optical
 telescope. The MB-DAC was then put back into the zero position. The beam stop was then
  tted taking care not to disturb the DAC.
 Two dark  eld measurements were taken and averaged and automatically subtracted out of
 the powder ring pattern [49]. The length of acquisition (30 minutes) was determined experimen-
 tally to give the best sample signal to noise ratio. Few adjustments had to be made to get an
 uncontaminated powder pattern as the gasket would sometimes give strong powder rings. After
 each pressure increase, the set up process would have to be completed. As the backing plates
 provide limited 2 re ections, the stage can be skewed to get the outer rings. Once the powder
 pattern has been collected the SMART software [49] unwarps the rings and the powder ring
 pattern imported into the FIT2D program [45]. This program converts 2-dimensional powder
 ring patterns into a powder peaks pattern and requires manual input of instrument parameters,
 like detector to sample distance and pixel size must be calculated.
 To determine the pixel size (calculated to be 122.3  m), the dimensions of the detector are
 divided by the number of pixels. The sample to detector distance was calculated by taking two
 54
3.3 High temperature experiments
 Figure 3.13: Siemens SMART MB-DAC stage - MB-DAC loaded onto the stage. The beam
 stop is not  tted in this  gure.
 55
3.3 High temperature experiments
 di erent sample to detector measurements and calculating where the face of the 2-dimensional
 detector was exactly positioned with respect to the sample. This was calculated to be 7.687 mm
 added to the detector distance used in the FIT2D program.
 The DAC was removed from the purpose built stage ( gure 3.13) to increase or decrease
 pressure and heat the sample. Initially some experimentation was performed to determine the
 optimal time and detector distance settings. 60 mm, 45 mm and 35 mm were chosen and 10 min,
 15 min and 30 min for each detector distance and time respectively. 45 mm was calculated to
 be the optimum distance from the detector to sample, however it was found experimentally
 that 35 mm (30 minutes) was a better compromise due to the dark  eld subtraction (dark  eld
 impinges on the pattern as shown in  gure 3.14).
 In a similar way as shown in  gure 3.14 the XRD lines were identi ed. We compare the Ar
 lines to the end members ZrO2 and HfO2 to both the orthoI and orthoII XRD patterns. We
 look at the gasket interference lines and describe any of the other prominent features in the
 di raction patterns. As the pressure increases the gasket hole closes and further intrudes into
 the di raction pattern. Figure 3.14 is an example of the raw powder rings and the resulting
 integration of these rings using the FIT2D program.
 Figure 3.14: XRD peak identi cation - Figure showing identi cation after integration using
 FIT2D, of raw data from the SMART XRD di ractometer. Dark  eld subtraction has not been done
 as can be seen by the white border in the ring pattern. The MB-DAC was prepared with a SS301
 gasket, loaded with Fluorinert PTM, Zr0:5Hf0:5O2 sample and ruby balls. (i) Unknown peaks, (ii)
 diamond spots, (iii) gasket and unknown peaks and (iv) monoclinic sample peaks.
 56
4
 Experimental Results and Analysis
 This chapter will be a logical  ow of results obtained using the various experimental techniques
 learned with discussions included. High pressure techniques were learned by using gem anvils
 in the Merrill-Basset cell. A new cubic zirconia manual anvil fabrication technique was devel-
 oped and the CZ gems? usefulness demonstrated for two extreme sample scattering cases, TiO2
 and Zr0:5Hf0:5O2. Following these preliminary high pressure studies and Raman spectroscopy
 measurements, diamond anvil cell techniques were used. Pressures of up to 20 GPa were ex-
 plored. The  nal part of the study used a combination of high pressure and high temperature
 by means of CO2 laser heating. These techniques were used to investigate the response of the
 Zr0:5Hf0:5O2 sample in conjunction with cryogenic PTM loading. Raman spectroscopy analysis
 techniques were used on both the MB-GAC and MB-DAC, while XRD was used exclusively on
 the MB-DAC. The results will be compared to the preliminary results obtained by Mojaki [8]
 and additional data will be presented on the P-T response of this compound as deduced from
 XRD and Raman data.
 4.1 Ambient Raman and XRD measurements on Zr0:5Hf0:5O2
 Ambient measurements were performed on the Zr0:5Hf0:5O2 powder sample. These were to
 obtain a reference Raman spectrum and reference XRD di raction pattern for comparison with
 the high pressure-temperature modi ed sample. Figure 4.1 shows the Raman spectroscopy
 results, while the XRD results are shown in  gure 4.12 of section 4.3.2. Additionally, the
 Raman spectrum was used to compare with the original Raman spectrum taken by Mojaki [8]
 and thus a direct comparison could be made to their results.
 57
4.2 Hand polished CZ anvils and initial Raman TiO2 results
 Figure 4.1: Raman Zr0:5Hf0:5O2 powder - (i) Raman spectrum of Zr0:5Hf0:5O2 powder at
 ambient conditions outside of the high pressure cell. (ii) Ambient ZrO2 Raman spectrum measured
 by Carlone [50]. (iii) Ambient Raman spectrum of Zr0:5Hf0:5O2 as measured by Mojaki [8]. The
 stars indicate the signature monoclinic Raman bands.
 In the spectrum of Mojaki ( gure 4.1 (ii)) there is an extra low intensity peak between
 250 cm 1 and 300 cm 1 and a low intensity doublet between 500 cm 1 and 550 cm 1 wave
 numbers. These additional peaks could be attributed to laser plasma lines. The Zr0:5Hf0:5O2
 sample displays the prominent monoclinic features of Mojaki?s [8] sample and those of Carlone
 [50].
 4.2 Hand polished CZ anvils and initial Raman TiO2 results
 The information regarding manual fabrication of CZ anvils can be found in an as yet unpub-
 lished paper by our high pressure research group [20]. The manual fabrication technique was
 important for a number of reasons, namely, learning high pressure methodology and concurrently
 developing the new manual CZ fabrication technique for use in the MB-GAC. Additionally by
 conducting Raman measurements while re ning the polishing methods, experimental parame-
 ters were uncovered regarding the use of PTM, gasketing materials and the optimum  ake size
 and thickness that needs to be loaded (critical in laser heating experiments).
 An ambient measurement of Zr0:5Hf0:5O2 when mixed with Si-oil and Fluorinert is shown
 58
4.2 Hand polished CZ anvils and initial Raman TiO2 results
 in  gure 4.2 and reveals a high attenuation of Zr0:5Hf0:5O2 and additional peaks compared to
 the ambient sample. The inset shows Si-oil and Fluorinert PTM which were compared to the
 Zr0:5Hf0:5O2 ambient sample. Figure 4.2 shows that there is an intense signal from the Si-oil
 below 100 cm 1 wave numbers and clearly swamps out the lower Zr0:5Hf0:5O2 sample peaks.
 These peaks are of interest at higher pressure as they are monitored when the orthoII phase
 develops. By comparing (b), (c) and (d) it is clear the extent to which the Si-oil obscures the
 Zr0:5Hf0:5O2 Raman signal. By comparing the (a) and (c) it is clear that the Flourinert, (e),
 intrudes minimally in the Zr0:5Hf0:5O2 spectrum. Even though the Fluorinert displays its own
 Raman signal, the signal is much less intense than Si-oil, but becomes more intense at higher
 pressures.
 Figure 4.2: Ambient Raman spectra of Zr0:5Hf0:5O2 mixed with PTM - (a) Zr0:5Hf0:5O2
 powder (b) Zr0:5Hf0:5O2 mixed with Si-oil on slide, (c) Zr0:5Hf0:5O2 mixed with Fluorinert. The
 inset shows the Raman bands of the (d) Si-oil and (e) Fluorinert PTM.
 Figure 4.3 shows the Raman spectrum of the CZ anvils for the region of interest 0-750 cm 1
 wave numbers. The Raman spectrum from 0-4000 cm 1 was measured by Xu [19] using a
 514.5 nm Ar ion laser beam. From the inset in  gure 4.3, it can be seen that the CZ anvils
 display intense Raman bands below 2000 cm 1 and more importantly below 1000 cm 1, the
 region of interest.
 In a subsequent experiment where the two samples of TiO2 and Zr0:5Hf0:5O2 were loaded
 (with results to follow), evidence is revealed that Zr0:5Hf0:5O2 is a weak scatterer and TiO2 a
 59
4.2 Hand polished CZ anvils and initial Raman TiO2 results
 Figure 4.3: Raman spectrum of the CZ anvil - Raman spectrum of CZ anvil over the region
 of interest from 0-700 cm 1 wave numbers. The inset shows the Raman spectrum as measured by
 Xu [19].
 strong scatterer ( gure 4.4). The intensities and data acquisition times of the Raman measure-
 ments at 5 GPa show the weak scatterer Zr0:5Hf0:5O2, display 2200 counts over 2400 seconds. By
 contrast the strong scatterer TiO2 displays 8000 counts over 5 seconds with the signal completely
 dominating any background signal from the CZ anvil or the PTM.
 Some valuable information was learned from these initial experiments. The sample of
 Zr0:5Hf0:5O2 is a weak scatterer and the TiO2 sample a strong scatterer. The CZ anvils display
 Raman bands and attenuate the Raman signal. Experimentally it was found that the PTM
 should be matched to the type of sample and anvils used. From preliminary analysis of these
 experiments, TiO2 and Si oil PTM were loaded into the 200  m Ta90W10 gasket cavity using the
 MB-DAC (labelled INC718 #1) with 380  m imperfectly manually polished CZ culets. Raman
 spectra were taken of the cavity contents and are shown in  gure 4.5.
 The mode hardening (shift to higher wave number) is clearly distinguished by way of the Eg
 mode [51] at 140 cm 1 of the TiO2 anatase phase. The other weaker intensity bands are also
 readily observable at 397 cm 1, 518 cm 1 and 641.3 cm 1. This  rst pressure run resulted in a
 failure of one of the CZ anvils at 5.56 GPa.
 The reason for this failure was believed to be caused by stresses at the edges of the culet face
 as the anvils were polished with very sharp edges as can be seen in  gure 4.6 (a). The polishing
 technique was then re ned to include a rounding on the edges of the culet, to distribute the
 60
4.2 Hand polished CZ anvils and initial Raman TiO2 results
 Figure 4.4: Intensity comparison of sample Raman bands - The intensity of (a) Weak scat-
 terer Zr0:5Hf0:5O2 showing 2200 counts over 2400 seconds with contaminating background contribu-
 tions from the CZ anvils and PTM that obscure some of the signatures of the sample. (b) Strong
 scatterer TiO2 reaching 8000 counts over 5 seconds which completely dominates any background
 signal from the CZ anvil or PTM, at 5 GPa.
 Figure 4.5: Raman MB-GAC high pressure measurement of TiO2 - Raman spectra of the
  rst TiO2 pressure experiment to 5.56 GPa using imperfectly hand polished anvils.
 61
4.2 Hand polished CZ anvils and initial Raman TiO2 results
 stresses at the octagon corners more evenly, as described in the previous chapter. This was
 noted in  gure 4.6 (b).
 Figure 4.6: Sharp and rounded CZ anvil - (a) Early type sharply polished CZ culet of poor
 quality. (b) Rounded CZ anvil resulting in high quality culet due to the soft cloth polishing step.
 Two anvils of 380  m were polished with the sole purpose of determining the quality of the
 anvils by measuring the maximum pressure that could be reached ( gure 4.7). The maximum
 pressure attained in this successful pressure experiment was 13.75 GPa and recorded with the
 ruby  uorescence method, without taking a Raman spectrum.
 Figure 4.7: Pressure measurement to 13.75 GPa - Rounded CZ anvil test showing ruby
  uorescence to 13.75 GPa.
 As proof of concept for manual hand polished anvils, a follow up experiment to take data
 at these pressures was designed. New 380  m CZ anvils were polished (for MB-GAC, INC718
 62
4.3 High pressure and high temperature Raman and XRD experiments
 #3) and TiO2 and Zr0:5Hf0:5O2 were both loaded into a sample cavity shown in  gure 4.8. The
 initial  ndings suggested that the Raman signal quality of the Zr0:5Hf0:5O2 sample may be weak
 at high pressure using CZ anvils. High pressure phase transformations of the TiO2 could be
 monitored should the Zr0:5Hf0:5O2 bands be too weak.
 4.3 High pressure and high temperature Raman and XRD ex-
 periments
 This section is divided into four subsections reporting results on MB-GAC experiments using
 the manually fabricated CZ anvils, preliminary XRD experiments in the MB-DAC and thirdly
 laser heating experiments with the cryogenically loaded Ar PTM in the MB-DAC. The  nal
 subsection compares the laser heated MB-DAC XRD data and Raman spectra to those of the
 unheated MB-DAC at 20 GPa.
 4.3.1 MB-GAC strong scatterer TiO2 and weak scatterer Zr0:5Hf0:5O2 exper-
 iments
 In the case exempli ed here, thin samples of TiO2 and the Zr0:5Hf0:5O2 sample were compacted
  at to form thin  akes in the indentation and then loaded into the sample cavity. Figure 4.8
 shows the loading of the samples, Fluorinert PTM and two ruby balls into the Ta90W10 metal
 gasket cavity which had been spark drilled to 150  m and closed to 0.7 GPa to create a seal.
 Figure 4.8: MB-GAC loaded with TiO2 and Zr0:5Hf0:5O2 sample - Non-trivial loading of
 both samples TiO2 and Zr0:5Hf0:5O2 into 150  m diameter sample cavity,  uorinert as the PTM
 along with ruby balls. The MB-GAC was closed and pressurized to 0.7 GPa to create a seal.
 63
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.9: Raman MB-GAC TiO2 high pressure measurements - Raman spectra taken
 in the MB-GAC from strong scatterer TiO2. The phase transition to more complex spectra of the
  -PbO2-type phase (blue dashed lines indicate signature peaks) and the baddeleyite phase (green
 dashed lines indicate signature peaks) are readily distinguished (see text for details). The  -PbO2-
 type phase develops between 8 and 10 GPa and the baddeleyite phase develops above 12 GPa. The
 TiO2 spectra completely dominates any background signal from the CZ anvil and PTM. The inset
 shows an example of the ruby  uorescence spectra taken at some of the highest pressures (14.8 GPa)
 attained in the MB-GAC. Note that the ruby  uorescence peaks are still well resolved and sharp at
 this pressure. This indicates that the cavity is relatively hydrostatic. Notice that the signal to noise
 ratio of the ruby  uorescence decreases at higher pressures.
 64
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.9 shows the TiO2 high pressure Raman spectra taken in the MB-GAC. As in section
 4.2, the mode hardening (shift to higher wave number) is clearly discerned by way of the Eg mode
 at 140 cm 1 of the TiO2 anatase phase [51]. The other weaker intensity bands are also readily
 observable at 397 cm 1 (B1g), 518 cm 1 (A1g / B1g unresolved doublet) and 641.3 cm 1 (Eg)
 [51]. Again, these are part of six assigned Raman active modes (A1g + 2B1g + 3Eg) derived
 from a factor group analysis of tetragonal anatase [51].
 As the pressure was increased to  8 GPa, several additional weak peaks appear in the range
 330-380 cm 1, characteristic of the  -PbO2-type phase. As pressure was further increased,
 the intensities of these new peaks in the range 330-370 cm 1 increase and they were clearly
 distinguished at pressures of  10 GPa and beyond, there is a broadening of the intense band
 at low wave numbers of 180 cm 1. The structural transition of bulk polycrystalline anatase
 to the  -PbO2-type intermediate is slow, as found previously [52], coexisting with the original
 anatase structure up to a pressure of  12 GPa. Beyond  12 GPa, the intense mode at 180 cm 1
 broadens appreciably and develops shoulder features. Furthermore as its pressure dependence
 decreases (signalling hardening), new Raman bands become apparent in the range 225-280 cm 1
 in addition to a prominent band at 500 cm 1. These are characteristic of the high-pressure (and
 more incompressible) baddeleyite phase and is fully stabilized at  15 GPa, consistent with the
 literature [48]. It may be noted how all aspects of this phase transition sequence are readily
 discerned, comparable to the case studies in the MB-DAC. This serves to demonstrate how the
 pressure signatures and phase transformations of strong to moderate scatterers may be readily
 monitored with this methodology, even down to low wave number lattice modes in regions
 where the contaminating background from the anvils is at its most intense. Two subsequent
 pressures were attained while trying to pressurize to 16 GPa, and the pressures were 15.5 GPa
 and 15.8 GPa, at which point the top anvil of 380  m failed.
 The spectra of the weak scatter Zr0:5Hf0:5O2 ceramic are depicted in  gure 4.10. The Raman
 spectrum of the monoclinic phase of this compound has been previously studied. The low
 symmetry of the crystal structure results at the  point in 18 Raman-active modes of symmetry
 (Ag) or (Bg). There are known modes in the wave number range up to  250 cm 1 [20]. Two
 main modes are discerned at 100 cm 1 and 150 cm 1. These modes have merged by 3 GPa (see
 later section 4.3.3 for veri cation in the MB-DAC).
 Modes in the range 300-350 cm 1 are partially obscured by the CZ background. A medium
 strength intensity doublet is discerned at 400 cm 1. The known intense mode at 490 cm 1
 is also discerned above the background. The rest of the modes in the range 500-700 cm 1
 65
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.10: Raman MB-GAC Zr0:5Hf0:5O2 high pressure measurements - Raman spec-
 tra taken in situ in the MB-GAC from the weak scatterer, Zr0:5Hf0:5O2 ceramic. Contaminating
 background contributions from the CZ anvils (orange dotted line) and PTM obscure some of the
 signatures of the sample. Raman bands from the Zr0:5Hf0:5O2 sample (blue dotted lines) are also
 readily discerned. The peaks at 100 cm 1 and 150 cm 1 merge by 5 GPa as shown in the Raman
 spectra. Inset shows the Raman spectrum of the CZ anvil.
 66
4.3 High pressure and high temperature Raman and XRD experiments
 are obscured by the strong contribution from the CZ. Part of the limitation of this low-scatter
 sample is perhaps the very thin (10-20  m thick and near transparent) Zr0:5Hf0:5O2 sample  ake
 used, so that most of the probing radiation is in fact transmitted rather than back-scattered.
 It is anticipated that the quality of the spectra should improve if there is an increase in the
 relative sample-to-cavity volumes. The purpose of introducing this data is to con rm, together
 with the preceding mention of anatase TiO2 behaviour, that low wave number lattice Raman
 modes at pressure may be tracked in spite of this being a weak scatterer in the region where the
 background from the CZ anvil tip is at its most intense. This is the  rst time that Raman bands
 at pressure in the lattice mode region have been presented in MB-GAC experiments involving
 CZ.
 These results demonstrate that the CZ-GAC may now be used quite routinely for Raman
 pressure measurements. It will su ce entirely for the case of strong to moderate scatterers or
 with Raman spectra very di erent to that of the anvils (background). It may be possible to give
 valuable preliminary indicators of the pressure response of weak scatterers (even though there
 may be background contamination from the CZ anvils) ahead of more suitable DAC experiments.
 The CZ-GAC is already known to be very suitable for IR spectroscopy work because of the low
 background in the high wave number regime, perhaps even superior to that of DAC work [53].
 This work demonstrates that complementary Raman data in the low wave number regime
 (lattice mode information) can also be derived from the same pressure cell, to appreciable
 pressures of 10-15 GPa. This easily covers and even exceeds the pressure range in which the
 pressure response of bio- and macro- molecular systems are investigated [20]. Smaller CZ culet
 sizes can be polished to say 300  m that would allow for an increase in pressure.
 These ideas may  nd similar applicability in moissanite [20] which has higher strength and
 stability compared to CZ, so that pressures up to 20 GPa for relatively convenient culet sizes
 (of 380  m) would be feasible. Although there is need to be aware of where there may be a
 strong background contribution from Raman signals of the anvil. Emphasis must be placed on
 the modest cost-e ective tooling and polishing techniques that may be used for the appropriate
 gem-anvil preparation. This methodology could in the least case, be used as an educational
 tool to school novices in analogous DAC preparation and pressurization techniques, without
 having to resort to expensive acquisition of diamond-anvils at the outset. Furthermore, materials
 characterization by optical spectroscopy at extreme high P-T conditions in excess of 10 GPa
 (previously the exclusive domain of DACs) is now much more accessible with modest e ort,
 expertise and cost.
 67
4.3 High pressure and high temperature Raman and XRD experiments
 4.3.2 MB-DAC high pressure with Fluorinert PTM experiments
 The experiments on the Zr0:5Hf0:5O2 ceramic in the MB-DAC is presented in this section. XRD
 will be used to clarify the phase transition at room temperature to 20 GPa complemented by
 Raman spectroscopy measurements. The results are compared to Raman spectra and XRD
 data of the end members from literature. In this experiment Fluorinert was used as the PTM,
 thus no laser heating or furnace heating was done due to probable unpredictable behaviour of
  uorinert (and anvil epoxy) at high temperatures. This initial investigation was performed to
 determine what further experiments would be necessary to further explore the phase diagram
 of Zr0:5Hf0:5O2.
 Figure 4.11: MB-DAC cavity loaded with Zr0:5Hf0:5O2 and Fluorinert PTM. - (i) Ruby
 balls  uorescing using the 20 mW, 532 nm green laser excitation source, (ii) SS301 gasket, (iii)
 Flourinert PTM, (iv) 200  m diameter sample cavity, (v) Zr0:5Hf0:5O2 sample  akes.
 The MB-DAC of tungsten 302, (HV = 560) was used and a gasket of SS301 (HV = 350)
 preindented to 40  m, was spark drilled to obtain a sample cavity of 200  m diameter. Samples
 of Zr0:5Hf0:5O2 were compacted  at to form thin  akes in the indentation and then pushed into
 the sample cavity with Fluorinert PTM, with  lling factor of  60% ( gure 4.11). Figure 4.12
 shows the ambient XRD measurement mentioned in section 4.1 above using the Bruker D8 x-ray
 di ractometer with Cu-K radiation.
 Figure 4.13 shows XRD data of Zr0:5Hf0:5O2 sample loaded in the MB-DAC. The  rst XRD
 measurement is at 0.6 GPa (sealing pressure) and clearly shows the monoclinic phase (two
 signature peaks at 12.5 and 14 , measured in 2 ) by comparing it to the di raction pattern
 of the ambient powder sample in  gure 4.12 (Mo-K radiation). These signature monoclinic
 68
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.12: XRD of raw Zr0:5Hf0:5O2 powder - Ambient Zr0:5Hf0:5O2 measurement on powder
 sample with 3 wt% Pt Black mixed in for internal pressure marker ( 40 ) experiment using copper
 K (Cu-K ) whereas the MB-DAC XRD measurements used molybdenum-K (Mo-K ). The inset
 shows the ambient XRD measurement performed by Mojaki [8].
 Figure 4.13: XRD of Zr0:5Hf0:5O2 in the MB-DAC - (i) XRD MB-DAC at high pressure
 displaying two peaks at 12.5 (11-1) and 14 (111) indicating monoclinic Zr0:5Hf0:5O2. At 8.3 GPa
 the orthoI peak 13.5 (211) starts emerging. Insets (ii) show end member ZrO2 monoclinic and (iii)
 orthoI high pressure phase (98426-ICSD and 41011-ICSD respectively). The pink stars correspond
 to monoclinic sample peaks and the blue stars correspond to orthoI sample peaks.
 69
4.3 High pressure and high temperature Raman and XRD experiments
 sample peaks at 12.5 and 14 are labelled the (11-1) and (111) re ections, respectively taken
 from the corresponding labelled peaks of the end members [1, 4]. The XRD pattern in  gure
 4.13 shows that the monoclinic phase dominates to 8.2 GPa although the monoclinic to orthoI
 phase change is initiated at  5 GPa [4]. The peaks at 12.5 and 14 decrease in intensity as the
 phase change proceeds to orthoI with a single peak at 13 developing. This is seen by tracking
 it from 0 GPa through to 20 GPa and con rmed by the literature [1, 4]. There are additional
 signature peaks at 16 and 22 indicated by pink and blue stars for the monoclinic and the
 orthoI phase, respectively.
 At 4.7 GPa the signature monoclinic peaks are still present from the 0.6 GPa XRD data.
 There is no indication of the orthoI phase. This con rms that the monoclinic to orthoI phase
 boundary is above 4.7 GPa and therefore certainly in the  5 GPa region as stated for the end
 member phase boundaries. The monoclinic peaks are slightly shifted right under the application
 of pressure. The other sample peaks are of low intensity and su er from being broad and thus
 some of the doublets are obscured. The MB-DAC was slightly misaligned in the sample holder
 resulting in a strong gasket peak at 20 .
 From 8.2 to 11.2 GPa, there is a phase mixture as the two monoclinic peaks decrease in inten-
 sity. The peaks shift slightly to the right under pressure as expected. There is the appearance of
 a third peak between the two monoclinic peaks and this indicates the orthoI phase developing at
 13.5 . The highest intensity orthoI peaks are at 13.5 (labelled the (211) re ection) and 16 (the
 combination of the (020), (002) and (400) re ections, respectively). The appearance of these
 peaks con rm the phase diagram boundary for ZrO2 and HfO2 end members in the literature
 (shown in  gure 2.2 and 2.3) [1, 4]. Ohtaka et al: [4] observed that low pressure measurements
 of ZrO2 had slow phase transformations and that there was a mixture of the phases over a wide
 range of pressures. The monoclinic phase coexists with the orthoI phase up to a pressure of
  15 GPa. This sluggishness may be overcome by heating at temperatures above 600 C [1].
 Figure 4.14 compares the two 11 GPa measurements taken a few weeks apart and show that
 the orthoI phase had further developed over time (the kinetics is on a time scale of days).
 The orthoI phase is dominant in the 18.2 GPa and 20 GPa XRD patterns by comparing
  gure 4.15 to  gure 4.13 (iii). There is no indication of orthoII from the XRD because of the
 absence of any signature orthoII peaks at 15 or 15.5 .
 The sluggishness of the transition from monoclinic to orthoI was con rmed in the Raman
 spectra. As pressure increases further to 15 GPa the intensities of these new peaks (150 cm 1,
 200 cm 1 and 400 cm 1) increase beyond 15 GPa there is an intense mode at 400 cm 1. These
 70
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.14: 11 GPa XRD data of Zr0:5Hf0:5O2 - XRD high pressure data indicating the
 sluggish monoclinic to orthoI conversion of Zr0:5Hf0:5O2. The second 11.01 GPa XRD pattern was
 taken a few weeks later. The cell had relaxed to a lower pressure of 11.01 GPa form 11.2 GPa. The
 conversion to the orthoI phase from monoclinic has advanced.
 Figure 4.15: XRD high pressure orthoI phase of Zr0:5Hf0:5O2 - XRD MB-DAC high pressure
 di raction pattern displaying the two monoclinic peaks at 13 and 14 weakening in intensity and
 the orthoI peak dominating the XRD pattern. XRD taken from a very well aligned MB-DAC as
 there is hardly a trace of the gasket line at 20 .
 71
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.16: Raman MB-DAC high pressure of Zr0:5Hf0:5O2 - Raman spectrum of
 Zr0:5Hf0:5O2 powder at ambient P-T, MB-GAC at 5 GPa and the MB-DAC at 11 GPa, 15 GPa
 and 20 GPa. The orthoI signature peak is at 380 cm 1 and is discerned in the 11 GPa spectra
 (green dotted line). Zr0:5Hf0:5O2 in MB-DAC loaded with Fluorinert PTM with no laser heating
 performed.
 72
4.3 High pressure and high temperature Raman and XRD experiments
 are supposed to be characteristic of the high-pressure (and more incompressible) orthoI phase.
 This orthoI phase is not fully stabilized at  15 GPa consistent with the literature [1, 4]. The
 pressure evolution of the Raman spectra of Zr0:5Hf0:5O2 is shown in  gure 4.16. Under ambient
 conditions the prominent Raman bands are from 100 cm 1 through to 700 cm 1. The ambient
 powder spectrum and CZ anvil cell data (5 GPa) is used as a reference, and it is slightly
 attenuated compared to the MB-DAC spectra. In this experiment, the  rst MB-DAC Raman
 measurement was taken at 11 GPa. It displays a broad peak at 150 cm 1, peaks between
 300 cm 1 and 450 cm 1and between 500 cm 1 and 600 cm 1. When the pressure is increased
 to 15 GPa a number of new Raman bands of low intensity appear at wave numbers 150 cm 1
 and 200 cm 1. These appear to merge to a single peak at 200 cm 1 at 20 GPa and are speci c
 to the high pressure orthoI phase. The bands at high pressure are broader than the low pressure
 monoclinic phase. The Raman spectra at 20 GPa can thus be used as a reference for the Raman
 signatures of orthoI. Using XRD it may be noted how all aspects of this phase transition sequence
 are readily discerned through the qualitative tracking of peaks. This is the  rst time that the
 Raman spectrum of the Zr0:5Hf0:5O2 solid solution has been unambiguously identi ed as orthoI.
 Figure 4.17: Raman MB-DAC decompression from 20 GPa to 2 GPa of Zr0:5Hf0:5O2 -
 Figure showing spectra at 20 GPa and 2 GPa. Blue dotted lines indicate monoclinic signature peaks
 and the green dotted line the orthoI signature peak.
 73
4.3 High pressure and high temperature Raman and XRD experiments
 In ZrO2 both orthoI and orthoII high pressure phases are quenchable to ambient conditions
 [4]. HfO2 orthoI is not quenchable to near ambient conditions but the orthoII phase is [1].
 Figure 4.17 shows the Raman spectra of the sample after it has been decompressed to 2 GPa.
 The pressure was decreased slowly from 20 GPa to 2 GPa and then immersed in a bath of liquid
 nitrogen to attempt to quench in the orthoI phase. The comparison of the Raman spectra in
  gure 4.17 to the XRD data in  gure 4.18 shows that there is a mixture of phases. Monoclinic
 modes as well as orthoI modes are discernible and the 2 GPa spectrum looks similar to the
 Raman spectra taken at 11 GPa. The XRD patterns con rm the monoclinic and orthoI mixture.
 The XRD data shows the two monoclinic peaks at 13 and 14 reappear with the orthoI peak
 observed between the two peaks at 12.5 . This shows that there is some reversion to the original
 monoclinic phase. It is highly probable that the recovered mixture of the phases monoclinic and
 orthoI has enhanced hardness and toughness properties compared to the original monoclinic
 phase. This result is inferred as Block et al: discovered enhanced hardness and toughness for
 the end member ZrO2 for pressures and temperatures as low as 8.6 GPa and 500 C [54].
 Figure 4.18: XRD MB-DAC decompression from 20 GPa to 2 GPa of Zr0:5Hf0:5O2 - XRD
 data at (i) 0.6 GPa (close to ambient) displaying monoclinic structure, (ii) 2 GPa decompression data
 which is a mixture of monoclinic (higher percentage) and the orthoI phases. (iii) 20 GPa showing
 the full conversion to orthoI.
 This experiment con rms that the Zr0:5Hf0:5O2 solid solution cannot be quenched to the or-
 thoI phase exclusively from 20 GPa whereas ZrO2 can be. This is the  rst time that Zr0:5Hf0:5O2
 74
4.3 High pressure and high temperature Raman and XRD experiments
 has been quenched to ambient from the orthoI phase. By using the combination of Raman and
 XRD we are able to track phase changes of Zr0:5Hf0:5O2. The emphasis of this  rst experiment
 was to learn high pressure methodology in the MB-DAC. Secondly the set up of the XRD was
 completed and di raction measurements taken. A reference di raction pattern involving no
 heating was required for comparison with the high pressure laser heated sample. The XRD data
 shows conclusively that the orthoI phase is dominant and there is compelling evidence that the
 orthoII phase cannot be synthesized by pressure (up to 20 GPa) at room temperature alone.
 This is the  rst time that XRD at pressure on Zr0:5Hf0:5O2 have been presented in MB-DAC
 experiments and compared with Raman spectra. The following experiment explores the conver-
 sion of Zr0:5Hf0:5O2 to the orthoII high pressure phase (cotunnite) when CO2 laser heating is
 used to accelerate phase transition kinetics.
 4.3.3 MB-DAC high pressure-temperature with Ar PTM experiments
 In this section the pressure-temperature phase diagram of Zr0:5Hf0:5O2 is further explored to
 20 GPa and laser heating to  1000 C. The Raman spectra results are compared to the previous
 pressure experiment and Mojaki?s [8] data along with the phase diagrams of the end member
 components (ZrO2 and HfO2).
 The MB-DAC (labelled INC718 cell # 1) was prepared with a gasket of SS301 preindented to
 40  m and spark drilled to obtain a cavity of 200  m diameter. Thin sample  akes of Zr0:5Hf0:5O2
 were pushed into the sample cavity ( gure 4.19). Liquid Ar was cryogenically loaded into the
 cavity to insulate the sample from the diamonds during the CO2 laser heating experiment.
 Raman spectra ( gure 4.20) were taken on the darker regions due to a better signal to
 noise ratio. These are labelled (A) through (E) in  gure 4.19. The pressure evolution of
 the Zr0:5Hf0:5O2 sample is comparable to previous high pressure experiments from 2 GPa to
 15.4 GPa (the MB-GAC experiment shown in  gure 4.10 and the MB-DAC experiment shown
 in  gure 4.16).
 At 15.4 GPa, a laser heating experiment was attempted. The cavity was heated for ap-
 proximately 15 minutes at about 75% full power with the laser beam focused into the second
 diamond as described in subsection 3.3.1. Figure 4.21 shows the pre- and post- laser heating
 Raman measurements at 15.4 GPa. It is clear that there is no change in the spectra. This
 implies that the laser was not intense enough, not focused properly on the sample or the heating
 was for too short a time. The only change that was noticed was the increase in pressure of
  0.4 GPa to 15.8 GPa after the laser heating.
 75
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.19: MB-DAC cavity loaded with Zr0:5Hf0:5O2 and Ar PTM - 200  m diameter
 stainless steel gasket cavity. (A), (B), (C), (D) and (E) are the regions that were measured by Raman
 spectroscopy. (i) Ruby balls and (ii) liquid Ar PTM.
 Figure 4.20: Raman Zr0:5Hf0:5O2 and Ar PTM - Raman specrta of Zr0:5Hf0:5O2 and liquid
 Ar PTM before laser heating. Green dotted line shows the orthoI signature peak.
 76
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.21: Raman spectra at 15 GPa of Zr0:5Hf0:5O2 - Raman spectra at 15 GPa before
 and after 15 minutes of laser heating (LH). No discernible change in the spectra is observed. The
 green dotted line indicates the signature orthoI peak.
 The signature peaks are marked from (a) through (u) on  gures 4.22 and 4.23 for easy
 comparison. The XRD data was converted to d-spacing ( gure 4.23) so that it could be directly
 compared with the d-spacings found in the literature [1, 4]. The d-spacing is as de ned in
 equation 2.18.
 The 2 GPa and 8.3 GPa XRD data shows consistency with the previous MB-DAC data shown
 in  gure 4.13. The two monoclinic peaks are clearly discernible in the 2 GPa XRD pattern. The
 features of the monoclinic and orthoI are easily observed in  gure 4.22. Monoclinic peaks at
 12.5 and 14 are labelled by the (11-1) and (111) re ections in  gure 4.23. These re ections
 convert to the orthoI peak at 13.5 (the (211) re ection) and seen in the 8.3 GPa XRD pattern.
 In a similar way to the previous result ( gure 4.13) the two peaks at 12.5 and 14 dominate
 up to about 8 GPa after which the single orthoI peak appears. Two lower intensity monoclinic
 peaks at 15.5 (the combination (002) and (020) re ections) become one orthoI feature at 16 
(the (020) re ection). In the reference XRD patterns ( gure 4.13 (iii)) there are two peaks at
 22 and is labelled (n).
 The (hkl) values have been assigned from the literature as (b) (11-1) and (c) (111). Peak
 (d) is the combination of the (020), (002) and (400) peaks. The Ar peaks are not yet prominent
 77
4.3 High pressure and high temperature Raman and XRD experiments
 because the Ar has possibly only randomly crystallised. The gasket line is at peak (f).
 The next data point is 8.3 GPa and is a mixture of monoclinic and orthoI. The monoclinic
 peaks are observed at (i) and (j) (the (11-1) and (111) re ections, respectively, with the orthoI
 peak developing between them. The intensities are low due to the cell being skew in the holder
 and the gasket line the dominant feature at (m). The peaks (k) and (l) are thought to be the
 solidi ed Ar (solidi es at 1.2 GPa [26]) obscuring the signature orthoI re ections at (020), (002)
 and (400) labelled for ZrO2 and HfO2 in the literature [1, 4].
 As no phase transformation took place as the pressure was increased to 20 GPa after which
 the  nal laser heating experiment was performed. The laser was raster scanned over the sample
 over the course of an hour. In contrast to the previous laser heating attempt, the raster scan
 lasted longer, the laser power was increased and was focused to a smaller spot size.
 The laser heated 20 GPa XRD data (in  gure 4.22) shows the orthoI peak at (o) with two
 less intense shoulder features at (p) and (u) shown in  gure 4.22 (2 ) and  gure 4.23 in d-
 spacing. The peaks in  gure 4.22 at 16 , 14 and 19 represent re ections (102), (111) and
 (211) respectively. The (p) and (u) peaks are at a d-spacing of 2.80  A and 2.65  A. These
 peaks correspond to the (102) and (111) re ections, respectively, of the orthoII phase of the end
 members, as depicted in  gure 4.23. Peak (s) is also a feature of the orthoII phase and is at
 a d-spacing of 2.02  A corresponding to a re ection of (211). This (211) re ection is the most
 intense feature of the end member ZrO2 orthoII phase. There are orthoI features still present in
 the 20 GPa XRD laser heated data. The feature (o) is the (211) signature re ection that is seen
 to be emerging between peaks (i) and (j) in the 8.3 GPa data. Peak (a) and (h) are features
 of the monoclinic and orthoI phases, respectively (re ection not labelled) and is absent from
 the 20 GPa laser heated data. The orthoII (011) re ection is overlapped by the orthoI (211)
 re ection (o). The transition to orthoII accelerates with the application of heat in the ZrO2
 case. The Ar peaks are represented as blue arrow heads in both the  gures 4.22 and 4.23 and
 labelled (q) and (r) [55] and mask any orthoII peaks in that region. The peak at (t) (along with
 (f) and (m)) is the gasket line. The d-spacings are given in table 4.1.
 The 20 GPa XRD laser heated data consists of a mixture of orthoI and orthoII. The d-
 spacings and angles of the orthoI and the orthoI/orthoII mixture are re ected in tables 4.2 and
 4.3.
 The experimental data is shifted to the left of the literature values. The internal XRD cali-
 brant was di cult to track in initial tests and was not used in the XRD experiments. Systematic
 errors include the sample to detector distance, a conservative error is about 0.5 mm by adjusting
 78
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.22: XRD data of Zr0:5Hf0:5O2 with Ar PTM - 2 GPa, 8.3 GPa and 20 GPa XRD.
 Monoclinic to a mixture of orthoI and orthoII (see text for peak assignments). d-spacing of the Ar
 peaks are represented as blue arrow heads (from literature values) with the blue dotted lines showing
 the Ar as measured in the sample. The gasket line is indicated with the green dotted line.
 Table 4.1: Ar d-spacings at 2 GPa, 8.3 GPa and 20 GPa - For the (111) and (200) re ections
 Pressure (GPa) (111)( A) (200)( A)
 2 2.83 2.45
 8.3 2.62 2.27
 20 2.46 2.13
 79
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.23: d-spacing obtained from XRD data of Zr0:5Hf0:5O2 - 8.3 GPa (Ar PTM),
 20 GPa (Fl PTM) and 20 GPa (Ar PTM) XRD d-spacing plotted alongside d-spacing of the end
 members from literature. (i) Monoclinic orthoI mixture at 8.3 GPa. (ii) OrthoI phase at 20 GPa.
 (iii) OrthoI orthoII mixture. Details follow in text.
 80
4.3 High pressure and high temperature Raman and XRD experiments
 Table 4.2: 20 GPa d-spacing of fully converted high P orthoI phase - Fully converted high
 P orthoI phase of Zr0:5Hf0:5O2 at 20 GPa (Fluorinert PTM).
 (hkl) Angle ( ) d-spacing ( A)
 8.29 4.91
 11.5 3.54
 11-1 13.07 3.12
 111 14.65 2.78
 16.16 2.52
 20.31 2.01
 Table 4.3: 20 GPa d-spacing of high P orthoI and orthoII phase mixture after laser
 heating - Mixture of the high P orthoI and orthoII phase of Zr0:5Hf0:5O2 at 20 GPa (Ar PTM and
 laser heating).
 (hkl) Angle ( ) d-spacing ( A)
 011 13.87 2.93
 102 16.29 2.50
 111 18.7 2.18
 112 20.27 2.02
 211 21.05 1.94
 81
4.3 High pressure and high temperature Raman and XRD experiments
 the telescope in and out of focus, the error is  0.05  A in the re ections calculated using FIT2D
 which is a about 0.5 respective error in the angle. There is also an error associated with the
 beam centre position. The DAC needs to be removed from the XRD stage for pressure increases
 and heating. The monocapillary needs to be moved and therefore the beam centre is di erent for
 each measurement. The monocapillary optics also add a 10% error to peak widths [41]. Due to
 uniaxial pressure generation the pressed  ake and Ar PTM has a high potential for non-random
 crystallite orientation. This a ects the intensities of the di raction peaks and in some cases may
 suppress peaks quite signi cantly. There is also an error associated with pressure measurement
 using the ruby  uorescence method. Pressures can be accurately measured to 0.02-0.05 GPa
 due to di erential thermal expansion of the components of the DAC and to time dependent
 mechanical relaxation within the gasket [24].
 It is important to note that the one of the goals of the experiment was to realize (or initiate)
 the orthoII (cotunnite) phase at a low as possible pressure, even though the MB-DAC was stable
 and could have been pressurized to higher pressures.
 4.3.4 Comparison of the 20 GPa high pressure experiments as a function of
 temperature
 The following subsection compares the 20 GPa results from both the MB-DAC loaded with
 Fluorinert to the MB-DAC loaded with liquid Ar and laser heated to  1000 C. The XRD data
 and Raman spectra are directly compared to Mojaki?s [8] spectra and to the literature [50].
 The results lead to a formulation of what future experiments should be performed including
 experiments using synchrotron radiation.
 Experimental Raman spectra from this study (shown in  gure 4.24 (i) laser heated, and (ii)
 pressure only, at 20 GPa) are shown alongside literature values for (iii) HfO2 and (iv) ZrO2 [56].
 Figures (i) and (ii) have no major discrepancies, the Raman bands at 200 cm 1 and 550 cm 1
 are more intense in (ii) than in (i). Figure 4.24 (v) shows previous Raman data obtained by
 Mojaki, with her spectra displaying similar bands at 400 cm 1 and 500 cm 1. The three low
 intensity bands above 600 cm 1 are also present in all data, although slightly shifted. The band
 found at 200 cm 1 in  gures 4.24 (i), (ii), (iii) and (iv) is missing from the laser heated spectra
 in 4.24 (v). Spectra in (v) have a high signal to noise ratio which could account for the peak
 not being easily distinguished at 100-200 cm 1. Note that the signature orthoI Raman band is
 at 400 cm 1 and is absent from the orthoII Raman spectra of the end members in (iii) and (iv)
 and the laser heated Raman spectrum of Mojaki (red in (v)) which is indicated by the green
 82
4.3 High pressure and high temperature Raman and XRD experiments
 Figure 4.24: OrthoII Raman comparison at 20 GPa of Zr0:5Hf0:5O2 - (i) 20 GPa spectra
 with Ar PTM, (ii) 20 GPa spectra with Fluorinert PTM. (iii) OrthoII HfO2 Raman spectra. (iv)
 OrthoII ZrO2 Raman spectra. (v) Mojaki?s 26 GPa laser heated (L-H) Raman spectra [8]. Green
 dotted lines indicate the position of the signature orthoI peak at 400 cm 1.
 83
4.3 High pressure and high temperature Raman and XRD experiments
 dotted lines in  gure 4.24. Therefore,  gure 4.24 compares Mojaki?s Raman spectra involving
 laser-heating of the solid-solution at pressures bigger than 20 GPa with Raman spectra of orthoII
 phases of the end members, as well as our unambiguous identi cation of the Raman  nger-print
 of the orthoI phase of the Zr0:5Hf0:5O2 solid-solution. This inter-comparison, supported by our
 XRD results, suggests that Mojaki attained signi cant conversion to orthoII upon laser-heating
 at 26 GPa at temperatures not exceeding  1500 C.
 Figure 4.25 shows the comparison of the 20 GPa XRD data. The two Ar peaks are also
 clearly marked and are the (111) and (200) re ections in table 4.1. OrthoI is fully converted at
 20 GPa and there is no indication of the orthoII phase. This is con rmed by the absence of (p)
 and (u) and the orthoII signature peak at about 20 (the (211) re ection). The Ar PTM laser
 heated data shows the orthoI peak (211) at about 14 (marked (o) in  gures 4.22 and 4.25). To
 the right, are two shoulder features (p) and (u). These are the signature orthoII peak positions
 as seen in the end members labelled (102) and (111) re ections, respectively. Peak (s) at about
 20 (211) is also a signature peak of the orthoII phase. The presence of the two shoulder features
 and the (211) re ection indicate the onset of the orthoII phase.
 Figure 4.25: XRD comparison at 20 GPa of Zr0:5Hf0:5O2 - 20 GPa Ar PTM laser heated
 Zr0:5Hf0:5O2 and 20 GPa Zr0:5Hf0:5O2 with Fluorinert PTM.
 The high-pressure stabilized cotunnite phase can be recovered to ambient conditions (irre-
 versibility) in the case of the end members ZrO2 and HfO2, whereas orthoI phase of Zr0:5Hf0:5O2
 84
4.3 High pressure and high temperature Raman and XRD experiments
 cannot.
 This study incorporated modest cost-e ective tooling of the monocapillary compared with a
 synchrotron experiment. This work demonstrates that it is possible to give invaluable prelimi-
 nary indicators of the pressure response of materials using the SMART di ractometer coupled
 with the 0.1 mm mono capillary (laboratory based XRD system) ahead of synchrotron exper-
 iments. It is anticipated that a quantitative analysis of this system can now be performed
 using a synchrotron experiment (proposal submitted to the beam line 27 at the ESRF). These
 laboratory-based XRD experiments have con rmed that the monoclinic to orthoI phase tran-
 sition is sluggish at room temperature. The orthoI phase is fully stabilised at 20 GPa. Laser
 heating at 20 GPa initiates the conversion to orthoII. It is anticipated that protracted heating
 (hours) of the Zr0:5Hf0:5O2 solid-solution will fully convert the orthoI phase to the orthoII phase.
 Alternatively, the pressure can be increased in the range 20-30 GPa with concurrent laser heat-
 ing to accelerate the phase conversion to orthoII. This seems to have been the case in previous
 work of Mojaki (see  gure 4.24).
 85
5
 Discussion and Conclusions
 Low wave number lattice Raman modes at pressure may be tracked using the CZ-GAC. This is
 despite Zr0:5Hf0:5O2 being found to be a weak scatterer in the region where the background from
 the CZ anvil tip is at its most intense. The data con rms the quality of the manual polished
 CZ anvils. This is the  rst time that Raman bands Zr0:5Hf0:5O2 at pressure in the lattice
 mode region have been represented in MB-GAC experiments involving CZ. It is anticipated that
 the CZ-GAC may now be used quite routinely for Raman pressure measurements. It will su ce
 entirely for the case of strong to moderate scatterers or with Raman spectra very di erent to that
 of the spectra from the anvils. It may be possible to report valuable preliminary indicators of the
 pressure response of weak scatterers ahead of more suitable DAC experiments. The CZ-GAC is
 already known to be very suitable for IR spectroscopy work because of the low background in the
 high wave number regime, perhaps even superior to that of DAC work [18], taking into account
 that there may be a strong background contribution from Raman signals of the CZ anvil. This
 work demonstrates that complementary Raman data in the low wave number regime (lattice
 mode information) can also be derived from the same pressure cell, to appreciable pressures of
 10-15 GPa. Pressures of 10-15 GPa easily covers and even exceeds the pressure range in which
 the pressure response of bio- and macro- molecular systems are investigated [20]. These ideas
 may  nd similar applicability in moissanite gem anvils [20], which have a higher strength and
 stability compared to CZ, so that pressures up to  20 GPa for relatively convenient culet sizes
 (of  400  m) would be feasible. Emphasis must be placed on the modest cost-e ective tooling
 and polishing techniques used for the CZ anvil preparation. This methodology could in the
 least case be used as an educational tool to school novices in analogous DAC preparation and
 pressurization techniques, without having to resort to expensive acquisition of diamond-anvils at
 86
the outset. Furthermore, previous materials characterisation by optical spectroscopy at extreme
 high P-T conditions in excess of 10 GPa, (the almost exclusive domain of DACs) is now much
 more accessible with modest e ort, expertise and cost.
 The phase relations and compression behaviour of Zr0:5Hf0:5O2 (nano-sized powders of un-
 stabilized zirconia-hafnia solid-solution) under high pressure and high temperature have been
 investigated in-situ by observations using MB-DAC XRD and Raman spectroscopy. Two high
 pressure phases have been observed. OrthoI is present from about 5 GPa at room tempera-
 ture and has been found to adopt the same structures as the end members ZrO2 and HfO2
 (i.e. monoclinic to orthorhombic-I to cotunnite) at similar pressure ranges. The XRD data in
 conjunction with Raman spectra is a previously unreported result. The orthoI to orthoII (or
 cotunnite) phase is seemingly also sluggish as is the transition from monoclinic to orthoI seen
 in both MB-DAC high pressure experiments. This sluggishness may be overcome by heating
 at temperatures above 600 C [1]. The transition of the sample from monoclinic to orthoI is
 con rmed via Raman spectroscopy and XRD. The XRD data has been conclusively used to
 determine the set of Raman  ngerprints for the orthoI phase of Zr0:5Hf0:5O2 solid-solution. A
 further result shows that for the Zr0:5Hf0:5O2 solid solution, the orthoI phase is not quenchable
 to ambient conditions as in ZrO2.
 The laser heating setup was modi ed to produce routine laser heating of samples. The
 addition of an extra CCD camera and the replacement of the main camera resulted in better
 alignment of the cavity and sample. Zr0:5Hf0:5O2 begins the transformation from orthoI to the
 orthoII phase at 20 GPa and temperatures of between 500 and 1000 C. The transition from the
 orthoI phase to orthoII phase is evident by comparison to the end members. By comparing the
 Raman spectra and XRD data from the two separate 20 GPa pressure studies, it is clear that
 the laser heating is necessary at these pressures to assist the phase change to orthoII. It appears
 that the monoclinic to orthoI phase transition has to be complete (pressures of  20 GPa) before
 the orthoII transition can be initiated.
 Our unambiguous identi cation of the Raman signatures of the orthoI phase, from com-
 plementary Raman and XRD pressure experiments, clari es the previous higher P-T results of
 Mojaki [8]. The orthoI and orthoII Raman signatures of the Zr0:5Hf0:5O2 solid-solution have
 now been  rmly established experimentally.
 XRD data has been analysed qualitatively ahead of envisioned future synchrotron experi-
 ments. This study incorporated modest cost-e ective tooling of the monocapillary compared
 with a synchrotron experiment. This work demonstrates that it may be possible to obtain
 87
valuable preliminary indicators of the pressure response of materials using laboratory XRD
 equipment (i.e. the SMART di ractometer and 0.1 mm monocapillary) ahead of more sophis-
 ticated synchrotron experiments so that beam-time can be used optimally. It is anticipated
 that a quantitative analysis of this system can now be performed e ciently by a synchrotron
 experiment. There have been a number of recent studies on quenched cotunnite phases of ZrO2
 and HfO2 that have indicated very high bulk moduli of over 300 GPa [7, 56]. Our observations
 show that there are similar phase changes that are induced in the Zr0:5Hf0:5O2 solid-solution,
 following similar pressure and temperature excursions in P-T phase space as for ZrO2 and HfO2
 end-members. Therefore, it may be supposed that the high P-T phases of this solid-solution
 could well be new ultra-hard ultra-tough \superceramic" candidates.
 88
Bibliography
 [1] O. Ohtaka, H. Fukui, T. Kunisada, T. Fujisawa, K. Funakoshi, W. Utsumi, T. Irifune,
 K. Kuroda, and T. Kikegawa, \Phase relations and volume changes of hafnia under high
 pressure and high temperature," Journal of the American Ceramic Society, vol. 84, no. 6,
 pp. 1369{73, 2001. 1, 2, 3, 8, 10, 70, 73, 74, 77, 78, 87
 [2] J. M. Leger and J. Haines, \The search for superhard materials," Endeavor, vol. 21, no. 3,
 pp. 121{124, 1997. 1, 2, 7, 14
 [3] Y. Kamata, \High-k/Ge MOSFETs for future nanoelectronics," Materials today, vol. 11,
 no. 1-2, pp. 30{33, 2008. 1
 [4] O. Ohtaka, H. Fukui, T. Kunisada, T. Fujisawa, K. Funakoshi, W. Utsumi, T. Irifune,
 K. Kuroda, and T. Kikegawa, \Phase relations and equations of state of ZrO2 under high
 temperature and high pressure," Physical Review B, vol. 63, pp. 727{733, 2001. 1, 2, 3, 9,
 10, 70, 73, 74, 77, 78
 [5] O. Ohtaka, D. Andrault, P. Bouvier, E. Schultz, and M. Mezouar, \Phase relations and
 equation of state of ZrO2 to 100 GPa," Journal of Applied Crystallography, vol. 38,
 p. 174108, 2005. 2, 10
 [6] J. K. Dewhurst and J. E. Lowther, \Highly coordinated metal dioxides in the cotunnite
 structure," Physical Review B, vol. 64, p. 014104, 2001. 2, 10
 [7] J. Haines, J. M. Leger, S. Hull, J. P. Petitet, J. P. Pereira, C. A. Perottoni, and J. A. H. da
 Jordana, \Characterization of the cotunnite-type phases of zirconia and hafnia by neutron
 di raction and Raman spectroscopy," Journal of the American Ceramic Society, vol. 80,
 no. 1910, pp. 1910{1914, 1997. 2, 11, 88
 89
BIBLIOGRAPHY
 [8] E. Mojaki, Study of the zirconia-hafnia system and particularly its behaviour at high tem-
 peratures and pressures. PhD thesis, University of the Witwatersrand, 2004. 3, 6, 7, 8, 9,
 13, 28, 31, 39, 57, 58, 69, 75, 82, 83, 87
 [9] A. F. Goncharov and J. C. Crowhurst, \Raman spectroscopy under extreme conditions,"
 Journal of Low Temperature Physics, vol. 139, pp. 727{737, June 2005. 7, 31
 [10] A. F. Holleman and N. Wilberg, Inorganic Chemistry. Walter de Gruyter, 2001. 8
 [11] L. G. Liu, \New high pressure phases of ZrO2 and HfO2," Journal of Physics and Chemistry
 of Solids, vol. 41, no. 331, pp. 331{334, 1980. 10, 11
 [12] W. Yang, R. G. Parr, and L. Uytterhoeven, \New relation between hardness and compress-
 ibility of minerals," Physics and Chemistry of Minerals, vol. 15, no. 2, pp. 191{195, 1987.
 10
 [13] \International crystalographic structure database version: 1.4.6," 2007. 11, 12
 [14] L. Davison, Y. Horie, and T. Sekine, High-Pressure Shock Compression of Solids V.
 Springer, 2002. 11, 13
 [15] A. Jayaraman, \Diamond anvil cell and high-pressure physical investigations," Reviews of
 Modern Physics, vol. 55, pp. 65{108, January 1983. 14, 26, 27, 31
 [16] W. F. Wilkinson and G. Serman, Advances in Infrared and Raman Spectroscopy, vol. 6.
 Heyden and Son: London, 1980. 14
 [17] D. E. Patterson and J. L. Margrave, \Use of gem-cut cubic zirconia in the diamond anvil
 cell," Journal of Physical Chemistry, vol. 94, pp. 1094{1096, 1990. 14, 15, 16, 17, 18, 20,
 24
 [18] T. P. Russell and G. J. Piermarini, \A high pressure optical cell utilizing single crystal cubic
 zirconia anvil windows," Review of Scienti c Instruments, vol. 68, no. 4, pp. 1835{1840,
 1997. 14, 16, 17, 19, 20, 22, 23, 86
 [19] J. Xu, J. Yen, Y. Wang, and E. Huang, \Ultrahigh pressures in gem anvil cells," Gem anvil
 pressure cells, pp. 127{134, 1996. 15, 16, 59, 60
 90
BIBLIOGRAPHY
 [20] N. R. Jackson, R. Erasmus, and G. Hearne, \Simpli ed manual fabrication of cubic-zirconia
 gem anvils for extended energy-range spectroscopic studies to routine high pressures of 100-
 150 kbar (10-15 GPa)," Unpublished, 2009. 16, 58, 65, 67, 86
 [21] J. Xu and E. Huang, \Graphite-diamond transition in gem anvil cells," Review of Scienti c
 Instruments, vol. 65, no. 1, pp. 204{207, 1994. 17, 24
 [22] S. A. Moggach, D. R. Allan, S. Parsons, and J. E. Warren, \Incorporation of a new design
 of backing seat and anvil in a Merrill-Bassett diamond anvil cell," Journal of Applied
 Crystallography, vol. 41, pp. 249{251, 2008. 18, 19, 20, 37
 [23] E. Sterer and M. P. Pasternak, \A multipurpose miniature diamond anvil cell," Review of
 Scienti c Instruments, vol. 61, no. 3, pp. 1117{1119, 1990. 19, 41, 46
 [24] R. Miletich, D. R. Allan, and W. F. Kuhs, \High-pressure single-crystal techniques," Re-
 views in Mineralogy & Geochemistry, vol. 41, pp. 445{519, 2000. 19, 21, 22, 24, 36, 46,
 82
 [25] I. F. Silvera, A. D. Chijioke, W. J. Nellis, A. Soldatov, and J. Tempere, \Calibration of the
 ruby pressure scale to 150 GPa," Phisica Status Solidi, vol. b, no. 1, pp. 460{467, 2007. 20,
 21, 22
 [26] J. Witlinger, R. Fischer, S. Werner, J. Schneider, and H. Schultz, \High-pressure study of
 h:c:p:-argon," Acta Crystallographica Section B, vol. B53, pp. 745{749, 1997. 23, 48, 78
 [27] G. Hearne, A. Bibik, and J. Zhao, \CO2 laser-heated diamond-anvil cell methodology re-
 visited," Journal of Physics: Condensed Matter, vol. 14, pp. 11531{5, 2002. 23, 24, 25,
 26
 [28] R. Boehler, \New diamond cell for single-crystal x-ray di raction," Review of Scienti c
 Instruments, vol. 77, p. 115103, 2006. 24
 [29] Y. Akahama, H. Kawamura, and A. K. Singh, \Equation of state of bismuth to 222 GPa and
 comparison of gold and platinum pressure scales to 145 GPa," Journal of Applied Physics,
 vol. 92, no. 10, pp. 5892{7, 2002. 24
 [30] S. Ndiaye, \Introduction to the fundamentals of Raman scattering." Presented as a short
 lecture course in the School of Physics , WITS University Johannesburg, SA Nov-Dec 2004.
 27
 91
BIBLIOGRAPHY
 [31] P. Dhamelincourt, J. Barbillat, and M. Delhaye, \Laser confocal Raman microspectrome-
 try," Spectrometry Europe, pp. 16{27, 1993. 27, 30, 31
 [32] J. C. de Paula, \Raman spectroscopy of metalloporphyrins," Access:
 http://www.haverford.edu/chem/302/Raman.pdf, 10 December 2010. 27
 [33] D. A. Long, Raman Spectroscopy. McGraw-Hill, Inc., 1977. 27
 [34] I. Gregora, \Raman scattering," International Tables for Crystallography, vol. D, pp. 315{
 328, 2006. 27, 32
 [35] I. E. Wachs, \Raman and IR studies of surface metal oxide species on oxide supports:
 Supported metal oxide catalysts," Catalysis Today, vol. 27, pp. 437{455, 1996. 29
 [36] G. B. B. M. Sutherland, Infra-red and Raman Spectra. Methuen & Co. LTD., 1935. 30
 [37] M. Baraton and M. El-Shall, \Synthesis and characterization of nanoscale metal oxides
 and carbides: Ii. micro-Raman and FT-IR surface studies of a silicon carbide powder.,"
 NanoStractured Materials, vol. 6, 1995. 32
 [38] B. E. Warren, X-ray di raction. Dover publications, INC. New York, 1990. 33
 [39] M. M. Woolfson, An introduction to x-ray crystallography. Cambridge University Press,
 1997. 33
 [40] H. P. Klug and L. E. Alexander, X-ray Di raction Procedures For Polycrystalline And
 Amorphous Materials. Wiley New York, 1974. 35
 [41] S. Cornaby, T. Szebenyi, R. Huang, and D. H. Bilderback, \Design of single-bounce mono-
 capillary x-ray optics," JCPDS-International Centre for Di raction Data, pp. 194{200,
 2007. 34, 36, 82
 [42] \Mono-capillary optics," Access: http://www.horiba.com/us/en/scienti c/products/x-ray-
  uorescence-analysis/tutorial/mono-capillary-optics/, 10 December 2010. 36
 [43] K. Brister, \X-ray di raction and absorption at extreme pressures," Review of Scienti c
 Instruments, vol. 68, no. 4, pp. 1629{1647, 1997. 37
 [44] \Boehler almax anvils," access: http://www.almax-industries.com/Boehler-Almax.html, 10
 December 2009. 38, 44
 92
BIBLIOGRAPHY
 [45] A. Hammersly, S. Stevenson, M. Han and, A. Fitch, and D. Hausermann, \FIT2D," Access:
 http://www.esrf.eu/computing/scienti c/FIT2D/, 2 June 2008. 38, 54
 [46] \Apollo scienti c cc," Access: http://www.apollosci.co.za/, 8 November 2009. 41
 [47] Leica-Microsystems-Ltd, Leica L2. Business Unit SM CH-9435 Heerbrugg (Switzerland),
 english m1-288-0en ed., 2001. 53
 [48] G. R. Hearne, J. Zhao, A. M. Dawe, V. Pischedda, M. Maaza, M. K. Niewoudt, P. Kiba-
 somba, O. Nemraoui, J. D. Comins, and M. J. Witcomb, \E ect of grain size on structural
 transitions in anatase TiO2: A Raman spectroscopy study at high pressure," Physical Re-
 view B, vol. 70, p. 134102, 2004. 53, 65
 [49] Brucker-AXS-Inc., computer program SMART-NT. Bruker AXS Inc., Madison, Wisconsin,
 USA., 1998. 54
 [50] C. Carlone, \Raman spectrum of zirconia-hafnia mixed crystals," Physical Review B, vol. 45,
 no. 5, pp. 2079{2084, 1991. 58, 82
 [51] T. Ohsaka, F. Izumi, and Y. Fujiki, \Raman spectrum of anatase, TiO2," Journal of Raman
 Spectroscopy, vol. 7, no. 6, p. 321, 1978. 60, 65
 [52] K. Lagarec and S. Desgreniers, \Raman study of single crystal anatase TiO2, up to 70
 GPa," Solid state communications, vol. 94, no. 7, pp. 519{524, 1995. 65
 [53] S. Takele and G. R. Hearne, \Electrical transport, magnetism, and spin-state con gurations
 of high-pressure phases of FeS," Physical Review B, vol. 60, pp. 4401{4403, 1999. 67
 [54] S. Block, G. J. Piermarini, B. J. Hockey, B. R. Lawn, and R. G. Munro, \High-pressure
 transformation toughening: A case study on zirconia," Journal of the American Ceramic
 Society, vol. 69, no. 6, pp. 125{126, 1986. 74
 [55] S. Merkel, \Simulation of high pressure di raction pattern," Access:
 http://merkel.zoneo.net/HPDi /applet.html, 12 December 2007. 78
 [56] S. Desgreniers and K. Lagarec, \High-density ZrO2 and HfO2: Crystalline structures and
 equations of state," Physical Review B, vol. 59, no. 13, pp. 8467{8472, 1999. 82, 88
 1
 1Numbers after the date refer to the page numbers where the references appear in the text.
 93