This electronic thesis or dissertation has been downloaded from Explore Bristol Research, http://research-information.bristol.ac.uk
Author: Potticary, Jason L Title: Novel Approaches to Morphological and Polymorphological Control and Selectivity in Crystalline Systems
General rights Access to the thesis is subject to the Creative Commons Attribution - NonCommercial-No Derivatives 4.0 International Public License. A copy of this may be found at https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode This license sets out your rights and the restrictions that apply to your access to the thesis so it is important you read this before proceeding. Take down policy Some pages of this thesis may have been removed for copyright restrictions prior to having it been deposited in Explore Bristol Research. However, if you have discovered material within the thesis that you consider to be unlawful e.g. breaches of copyright (either yours or that of a third party) or any other law, including but not limited to those relating to patent, trademark, confidentiality, data protection, obscenity, defamation, libel, then please contact [email protected] and include the following information in your message:
• Your contact details • Bibliographic details for the item, including a URL • An outline nature of the complaint
Your claim will be investigated and, where appropriate, the item in question will be removed from public view as soon as possible. Novel Approaches to Morphological and Polymorphological Control and Selectivity in Crystalline Systems
By
JASON POTTICARY
School of Chemistry UNIVERSITYOF BRISTOL
A dissertation submitted to the University of Bristol in accordance with the requirements of the degree of DOCTOR OF PHILOSOPHY in the Faculty of Science.
MARCH 2018
Word count: Forty seven thousand
ABSTRACT
he physical properties and the potential applications of a material are defined exclusively by the way in which the matter packs together and how discrete portions of this matter Tform on a macroscopic level. For example, the mineral quartz is a commonly used material for optical experiments. The way in which silicon and oxygen atoms pack allow a fairly uniform transmission of light from 1400 nm to 200 nm and our ability to grow large single-crystals, allow for applications such as cuvettes and optical windows. Issues arise when developing materials if they possess the required physical properties but cannot be grown in the right morphology or they have the required morphology but need their properties tuned. Quartz would not be as useful, if it only grew as highly anisotropic, crystalline needles. This thesis investigates the control of crystal packing (polymorphism) of molecular crystals, specifically the polyaromatic hydrocarbon coronene, using magnetic fields applied during the crystal growing process. A previously unknown polymorph of coronene was observed and fully characterised. The polymorph is accessible when grown from solution under a field between 0.9 T to 1.0 T. When investigating why this selectivity may be occurring, it was found that not only is the molecular conformation of the new polymorph the most energetically favourable via simulation, but the new polymorph is also accessible at cryogenic temperatures. Previous data reported on low-temperature coronene is readdressed with the knowledge of the new polymorph. Further to polymorphism, the morphology of a complex oxide material, specifically the high- temperature superconductor yttrium barium copper oxide (YBCO), is addressed in attempts to grow nanowires by mimicking the mechanism of micro-crucible growth without the use of a carbonaceous bio-template. Nanowires of various lengths and widths were grown using different amounts of a sodium-based flux. The nanowires were identified as a range of materials (Y2BaCuO5 (Y211), YBa2Cu3O7−x (Y123), Ba2CuO) based on the temperature and sodium content. This method was easily translated to other, more complex, metal-oxide systems and promises to be a general route to the formation of inorganic nanowires.
i
DEDICATIONANDACKNOWLEDGEMENTS
The real heroes of this thesis and work contained herein are the people that have helped me to embark on this obsessive journey and so in this small way I wish to convey my Timmense gratitude. Without some of whom, this work would have been impossible and without the others, these past few years would not have been even remotely as fun. First to my wife Stasja, my infinite source of encouragement and inspiration and also my ground. Your patience and understanding have made all of this possible. You have always made me want to be better and for you, I will always keep trying. It doesn’t even need stating that I wouldn’t be here and this thesis wouldn’t exist without you, but I’m sure you agree, it is nice to have in writing. Also to the third member of our family, whom I cannot wait to meet! My supervisor Dr. Simon Hall, your boundless enthusiasm, encouragement and support have kept me not only steady but constantly interested in the work at hand. Thank you for granting me the opportunity to work in this field and within your group. May your innovative perspectives ever flow - Omnis caro et caseus omnis!. To my partner in death-defying experiments Dr. Lui Terry, as far as I’m concerned you haven’t lived if you have never manned a fumehood with an HF spill kit. You have only added positive things to my life and I know there is more to come. Dr. Ben and Mr. Sparticus Mills, never a dull moment, you guys were the best company one could hope for in a lab, a vast source of knowledge both scientific and otherwise. To all those in the Complex Functional Materials Group, past and present who have kept things interesting and been a constant stream of enthusiasm; Tash, Tom, Becky, Omar, Julia, Charlie, Torsten, Nicole, Ge, Pedr, George, Alex and all those I’ve missed. There is clearly not enough room to correctly articulate how valuable you have all been. Throughout the University of Bristol and School of Chemistry, to the people who have always been ready to offer assistance when needed. Dr. Chris Bell, thank you for always providing me with a new perspective. Mr. Jonathan Jones for help with the microscopes and keeping them ticking. Dr. Hazel Sparkes and Dr. Natalie Pridmore for endless amounts of support with anything related to X-ray diffraction. Professor Charl Faul for keeping the lab ship-shape and Bristol fashion. Ms. Ilerha Hassimova for keeping me optimistic and Jonesy Worrall for pleasant distractions from the past. To all collaborators with whom I have had the pleasure of working with specifically, Drs. Enrico Da Como and Simon Crampin from the University of Bath, Dr Hans Engelkamp from Radboud University, Dr. Yoichi Kamihara from Keio University and Dr. Liana Vella-Zarb from
iii the University of Malta. Also to Professors Yoshihiro Yamazaki and Sossina Haile for giving me a leg-up back into science. My family of whom there are clearly too many to name but I will do so anyway; Mum and Nick, Nana and Granddad, Matthew, Adam, Robin, Faye, Alicia, Lauren, Hannah, Ryan, Julia, Heidi, Lucy, Amelia, Cassie, Olivia, Paul, Vicky, Liam, Debbie, Nigel, Kat, Matt, Matt and Effie for never giving me a hard time for constantly being too busy and going months without contact, you were all one less stress to have to deal with and nothing I could ever do would make me deserving of being genetically stuck to such a fantastic group of people. My friend Mr. Gareth Rees, for great times at the Carrington institute and being one of the best people I know. Not sure where I’d be if I’d never met you. Finally, to the Bristol Centre for Functional Nanomaterials for giving me this opportunity and the seemingly bottomless funding, specifically for the support from Annela, Terry, Joanna, Becky and yes, even Andy. Thank you all.
iv AUTHOR’SDECLARATION
declare that the work in this dissertation was carried out in accordance with the requirements of the University’s Regulations and Code of Practice for Research IDegree Programmes and that it has not been submitted for any other academic award. Except where indicated by specific reference in the text, the work is the candidate’s own work. Work done in collaboration with, or with the assistance of, others, is indicated as such. Any views expressed in the dissertation are those of the author.
SIGNED:...... DATE:......
v
TABLEOF CONTENTS
Page
List of Figures xi
List of Tables xv
Acronyms xvii
1 Introduction 1 1.1 Rationale...... 1 1.2 Crystals and crystalline matter...... 2 1.2.1 Crystallisation ...... 2 1.2.2 Types of crystal...... 3 1.2.3 The unit cell...... 3 1.2.4 Naming conventions used in this thesis ...... 5 1.3 Crystal nucleation...... 7 1.3.1 Secondary nucleation ...... 7 1.3.2 Primary nucleation...... 7 1.4 Polymorphism ...... 10 1.4.1 Naming polymorphs...... 12 1.4.2 Metastable polymorphs...... 12 1.5 Morphology...... 13 1.5.1 Polymorph directed morphology...... 14 1.6 Characterisation of crystalline matter...... 14 1.6.1 X-ray diffraction ...... 14 1.6.1.1 Single crystal X-ray diffraction...... 16 1.6.1.2 Powder X-ray diffraction ...... 16 1.6.2 Rietveld refinement ...... 17 1.6.3 Multiphase Rietveld refinement...... 18 1.7 Summary...... 19
2 Materials and Methods 21
vii TABLEOFCONTENTS
2.1 Materials...... 21 2.1.1 Purification of crystalline precursors via sublimation...... 21 2.1.2 Purification of crystalline precursors via solution recrystallisation . . . . . 22 2.1.2.1 Coronene...... 22 2.1.2.2 Inorganic materials ...... 23 2.1.3 Solvents ...... 23 2.2 Methods...... 23 2.2.1 Glassware cleaning...... 23 2.2.2 Inductively coupled plasma...... 23 2.2.3 Dynamic light scattering ...... 24 2.2.4 Diffusion orientated NMR spectroscopy ...... 24 2.2.5 Crystal growth under a magnetic field...... 24 2.2.6 Coronene growth up to 1 T using permanent magnets...... 25 2.2.7 Crystal growth under higher magnetic fields of up to 25 T using Bitter electromagnets...... 26
2.2.8 Synthesis of Fe2O3 nanoparticles...... 26
2.2.9 Preparation of Fe2O3 polluted coronene systems...... 27 2.2.10 Synthesis of metal carbonate nanoparticles...... 27 2.2.11 Synthesis of complex oxide pellets...... 27 2.2.12 Bio-templated nano-wire synthesis...... 27 2.2.13 Computational methods...... 28 2.2.13.1 Unit cell calculations...... 28 2.2.13.2 Cluster calculation...... 28 2.3 Analysis...... 28 2.3.1 Single crystal X-ray diffraction ...... 28 2.3.2 Powder X-ray diffraction ...... 29 2.3.2.1 Powder refinement...... 29 2.3.3 Solution based UV-vis spectroscopy...... 29 2.3.4 Solid optical absorption...... 29 2.3.5 Microscopy...... 30 2.3.5.1 Transmission electron microscopy...... 30 2.3.5.2 Scanning electron microscopy ...... 30 2.3.5.3 Atomic force microscopy...... 30 2.3.6 Differential scanning calorimetry...... 30 2.4 Design and construction of a 2.5 T permanent magnet...... 31
3 Magnetically Induced Polymorphism in Coronene 33 3.1 Introduction ...... 33 3.1.1 Polyaromatic hydrocarbons...... 34
viii TABLEOFCONTENTS
3.1.2 Polyaromatic hydrocarbons as molecular crystals ...... 35 3.1.3 PAH crystal packing...... 37 3.1.3.1 Polymorphism in PAHs ...... 38 3.1.4 Coronene...... 39 3.2 Results and discussion...... 42 3.2.1 Growth of coronene under a magnetic field ...... 42 3.3 Characterisation...... 46 3.3.1 Structural stability...... 46 3.3.1.1 Calculated stability...... 46 3.3.1.2 Low temperature structural analysis...... 47 3.3.1.3 Size Specific cryogenic phase behaviour ...... 50 3.3.1.4 Differential scanning calorimetry...... 52 3.3.2 Optical behaviour ...... 53 3.4 Impact upon previous research...... 55 3.5 Summary...... 58
4 The Mechanism of Magnetic Crystal Growth 61 4.1 Introduction ...... 61 4.2 Magnetic purity concerns ...... 62 4.3 Molecular interaction calculations...... 64 4.3.1 Absorption prediction using TD-SCF calculations ...... 69 4.4 Prenucleation identity of coronene...... 71 4.4.1 Dynamic light scattering ...... 72 4.4.2 Diffusion-orientated NMR spectroscopy ...... 74 4.5 Coronene growth under high magnetic fields...... 76 4.5.1 In situ. measurement preparation...... 76 4.5.2 Crystal growth ...... 78 4.6 Summary...... 83
5 Growth of Complex-Oxide Nanowires via Synthetic Application of the Micro- Crucible Mechanism 89 5.1 Introduction ...... 89 5.2 Nanowires ...... 91 5.2.1 Nanowire growth...... 91 5.2.1.1 VLS growth ...... 91 5.2.1.2 Microcrucible growth...... 92 5.2.2 Superconductivity and superconducting materials...... 93 5.3 Results and discussion...... 94 5.3.1 Bio-templated nano-wire growth ...... 94
ix TABLEOFCONTENTS
5.3.2 YBCO growth...... 94 5.3.3 Non-templated micro-crucible growth ...... 97 5.4 The use of sodium as a flux ...... 99 5.4.1 Using additives as a flux ...... 99 5.4.1.1 2 % Na·X ...... 100 5.4.1.2 5 % Na·X ...... 109 5.4.1.3 10 % Na·X ...... 118 5.4.1.4 15 % Na·X ...... 127 5.4.2 Identification of nanowires...... 133 5.4.2.1 Selected area of electron diffraction of nanowires ...... 135 5.5 Brief attempts on another system ...... 139 5.6 Summary...... 141
6 Conclusions 145 6.1 Conclusions...... 145 6.1.1 Magnetically induced polymorphism in coronene ...... 145 6.1.2 Templated growth of metal-oxide nanowires ...... 147 6.2 Further work...... 148 6.2.1 Magnetically induced polymorphism in coronene ...... 148 6.2.2 Tempted growth of metal-oxide nanowires...... 149
Bibliography 151
A Appendix 173
B Appendix B 191 B.1 Design and construction of a 2.5 T permanent magnet ...... 191 B.1.1 Component design...... 192 B.1.2 Component construction ...... 194 B.1.3 Build Instructions ...... 196 B.1.3.1 Safety...... 196 B.1.3.2 Bringing two magnets into contact ...... 196 B.1.3.3 Magnet rig assembly...... 197 B.1.3.4 Bringing two magnet stacks into contact ...... 197 B.1.3.5 Mounting the dual magnet stacks in the aluminium magnet holder199 B.1.3.6 Mounting the dual magnet stacks onto the yoke base ...... 200 B.1.3.7 Mounting pole shoe...... 200 B.1.4 Operation Instructions ...... 200 B.1.5 Field Characterisation...... 201
x LISTOF FIGURES
FIGURE Page
1.1 Description of three types of crystal ...... 4
1.2 Definition of doff and ∠mol ...... 6 1.3 Graph explaining critical nucleus size...... 8 1.4 Representation of Polymorphism...... 11 1.5 Energy diagrams for enantiotropic and monotropic polymorphism...... 11 1.6 The energy of polymorphs during crystal growth ...... 13 1.7 Summary of Bragg diffraction...... 15
2.1 Experimental configuration for the growth of β-coronene ...... 25 2.2 Schematic of the sample stage for fields of up to 25 T...... 26
3.1 The four PAH polymorphs ...... 37 3.2 Representation of the 4 PAH crystal-packing motifs...... 38 3.3 A geological sample of natural coronene: karpatite ...... 39 3.4 Molecular structure of coronene...... 40 3.5 Optical images of coronene crystals grown under 0 T and 1 T ...... 42 3.6 SEM micrographs of coronene polymorphs ...... 43 3.7 Unit cell display of coronene polymorphs ...... 44 3.8 Graph characterising the nomenclature of γ- and β-coronene...... 45 3.9 A 2D representation of powder diffraction data collected as a function of temperature in coronene ...... 48 3.10 Comparison of coronene powder X-ray diffraction (pXRD) patterns ...... 49 3.11 Size specific emergence of β-coronene at cryogenic temperatures...... 51 3.12 Physical characterisations of coronene polymorphs...... 53 3.13 Optical images of coronene crystals grown under 0 T and 1 T under UV light . . . . . 54 3.14 Absorption of of both γ- and β-coronene single crystals...... 55 3.15 Comparison of magnetic and structural data from coronene at low temperatures . . . 57
4.1 Characterisation of Fe2O3 nanoparticles ...... 62
4.2 SEM micrographs of Fe2O3 contaminated coronene crystals...... 63
xi LISTOF FIGURES
4.3 Dispersion corrected, density-functional-theory (DFT-D) optimisation of small coronene aggregates...... 65
4.4 All measurable values of doff , πd and ∠mol for calculated clusters ...... 66
4.5 Relative energies of 2 coronene molecules locked at πd = 3.433 Å with different doff . 67
4.6 Energy map of 2 coronene molecules locked at πd = 3.433 Å...... 68 4.7 Schematic showing typical β and γ morphologies...... 69 4.8 Calculated absorption data for coronene clusters of size 1, 2, 3 and 4 molecules . . . . 70 4.9 UV-vis of a concentrated solution of coronene in toluene with sequential dilutions . . 71 4.10 DLS data from a cooling system of coronene in toluene...... 73 4.11 Description of a standard DOSY experiment...... 74 4.12 DOSY data for the coronene - toluene system...... 75 4.13 Experimental and calculated data for coronene ...... 77 4.14 Absorbance of coronene at 0 T as a function of temperature...... 79 4.15 Absorption intensity as a function of temperature at 0 T...... 80 4.16 Absorption intensity as a function of temperature for fields 0 T to 20 T ...... 81 4.17 Images of crystals grown under fields 0 T to 20 T ...... 83 4.18 Absorption intensity as a function of temperature for fields 0 T to 1.2 T...... 84
4.19 Comparison of temperature of onset of crystallisation (CT ) of both sets of experiments in high-field magnets ...... 85
5.1 Mechanism of micro-crucible nano-wire growth ...... 92 5.2 Powder diffraction and scanning electron microscopy (SEM) of YBCO grown using sodium alginate ...... 95 5.3 SEM and powder diffraction analysis of YBCO after solid state reaction ...... 96
5.4 Analysis of BaCO3 nanoparticles...... 97
5.5 SEM and powder diffraction analysis of BaCO3 nanoparticle-seeded YBCO after solid state reaction...... 98 5.6 Comparison of the solid state reaction of YBCO precursors calcined at various temperatures with 2 % Na salt ...... 101 5.7 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 800 ◦C ...... 102 5.8 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 820 ◦C ...... 103 5.9 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 840 ◦C ...... 104 5.10 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 860 ◦C ...... 105 5.11 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 880 ◦C ...... 106 5.12 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 900 ◦C ...... 107 5.13 SEM micrographs of YBCO pre-cursors containing 2 % Na held at 920 ◦C ...... 108 5.14 Comparison of the solid state reaction of YBCO precursors calcined at various temperatures with 5 % Na salt ...... 110 5.15 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 800 ◦C ...... 111
xii LISTOF FIGURES
5.16 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 820 ◦C ...... 112 5.17 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 840 ◦C ...... 113 5.18 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 860 ◦C ...... 114 5.19 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 880 ◦C ...... 115 5.20 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 900 ◦C ...... 116 5.21 SEM micrographs of YBCO pre-cursors containing 5 % Na held at 920 ◦C ...... 117 5.22 Comparison of the solid state reaction of YBCO precursors calcined at various temperatures with 10 % Na salt...... 119 5.23 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 800 ◦C ...... 120 5.24 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 820 ◦C ...... 121 5.25 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 840 ◦C ...... 122 5.26 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 860 ◦C ...... 123 5.27 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 880 ◦C ...... 124 5.28 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 900 ◦C ...... 125 5.29 SEM micrographs of YBCO pre-cursors containing 10 % Na held at 920 ◦C ...... 126 5.30 Comparison of the solid state reaction of YBCO precursors calcined at various temperatures with 15 % Na salt...... 128 5.31 SEM micrographs of YBCO pre-cursors containing 15 % Na held at 800 ◦C ...... 129 5.32 SEM micrographs of YBCO pre-cursors containing 15 % Na held at 820 ◦C ...... 130 5.33 SEM micrographs of YBCO pre-cursors containing 15 % Na held at 840 ◦C ...... 131 5.34 SEM micrographs of YBCO pre-cursors containing 15 % Na held at 860 ◦C ...... 132 5.35 SEM micrographs of YBCO pre-cursors containing 15 % Na held at 880 ◦C ...... 133 5.36 SEM micrographs of YBCO pre-cursors containing 15 % Na held at 900 ◦C and 920 ◦C 134 5.37 TEM and SAED of nanowires grown with 5 % NaCl at 820 ◦C and 880 ◦C ...... 135 ◦ ◦ ◦ 5.38 TEM and SAED of nanowires grown with 5 % Na2CO3 at 820 C, 840 C and 860 C . 137 ◦ 5.39 TEM and SAED of nanowires grown with 5 % Na2CO3 at 880 C ...... 138 ◦ 5.40 TEM and SAED of nanowires grown with 5 % Na2CO3 at 900 C ...... 138 ◦ 5.41 TEM and SAED of nanowires grown with 5 % Na2CO3 at 920 C ...... 139 5.42 Nanowires in the BSCCO system...... 140
A.1 A typical fort.34 geometry input file for CRYSTAL14...... 174 A.2 A typical input input file for CRYSTAL14...... 175 A.3 Unit cell parameters coronene calculated via Rietveld analysis ...... 178
A.4 Size distribution histograms for YBCO nanowires with 2 % Na2CO3 ...... 183
A.5 Size distribution histograms for YBCO nanowires with 5 % Na2CO3 ...... 184
A.6 Size distribution histograms for YBCO nanowires with 10 % Na2CO3 ...... 185
A.7 Size distribution histograms for YBCO nanowires with 15 % Na2CO3 ...... 186 A.8 Size distribution histograms for YBCO nanowires with 2 % NaCl ...... 186 A.9 Size distribution histograms for YBCO nanowires with 5 % NaCl ...... 187
xiii LISTOF FIGURES
A.10 Size distribution histograms for YBCO nanowires with 10 % NaCl ...... 188 A.11 Size distribution histograms for YBCO nanowires with 15 % NaCl ...... 189
B.1 Graphical representation of the structure of a 2.5 T permanent magnet ...... 193 B.2 Profile view 2D Magnetic field calculation produced using finite element method magnetics (FEMM) software...... 194 B.3 Profile schematic of the 2.5 T permanent magnet ...... 195 B.4 Top down schematic of the 2.5 T permanent magnet...... 195 B.5 Schematic of the end of the 2.5 T permanent magnet...... 196
B.6 Manipulation of loose neodymium iron boron (Nd2Fe14B) (NIB) block magnets . . . . 197 B.7 Visual documentation of magnet components and build process ...... 198 B.8 Formation of dual magnet stacks...... 199 B.9 Top view of mounting a dual magnet stack in aluminium magnet holder ...... 199 B.10 Stray-field map of a 2.5 T permanent magnet...... 201
xiv LISTOF TABLES
TABLE Page
1.1 Description of the 14 Bravais lattices ...... 5
3.1 US-EPAs list of priority polyaromatic hydrocarbons (PAHs) pollutants...... 36 3.2 Unit cell parameters of coronene grown under various field strengths...... 43 3.3 Calculated structural parameters for coronene in the two crystal forms γ and β . . . 47
4.1 Crystallisation temperature of coronene under high magnetic fields...... 82
5.1 Nanowire diameters observed with 2% Na content ...... 109 5.2 Nanowire diameters observed with 5% Na content ...... 118 5.3 Nanowire diameters observed with 10% Na content...... 127 5.4 Nanowire diameters observed with 15% Na content...... 135
A.1 Unit cell parameters of the two forms of coronene collected at various temperatures . 176 A.2 Table of full PBC calculations...... 177 A.3 The 32 PAHs used to identify packing motifs...... 179 A.4 Unit cell parameters phases used for identification and analysis of YBCO synthesis . 180 A.5 Relative amounts of phases in YBCO nanowire mixtures containing NaCl ...... 181
A.6 Relative amounts of phases in YBCO nanowire mixtures containing Na2CO3 . . . . . 182
xv
ACRONYMS
β-HB β-herringbone. NIB neodymium iron boron (Nd2Fe14B). γ-HB γ-herringbone. NMR nuclear magnetic resonance.
AFM atomic force microscopy. PAH polyaromatic hydrocarbon. API active pharmaceutical ingredient. PBC periodic boundary conditions. PCA pre-critical aggregations. B2212 Bi2Sr2CaCu2O8+x. PNA polynuclear aromatic hydrocarbon. BSCCO bismuth strontium calcium copper ox- PNC pre-nucleation cluster. ide. PTFE polytetrafluoroethylene.
CT temperature of onset of crystallisation. pXRD powder X-ray diffraction. CASTEP Cambridge serial total energy pack- age. SAED selected area electron diffraction. CNT classical nucleation theory. scXRD single crystal X-ray diffraction. SEM scanning electron microscopy. DFT density functional theory. SQUID superconducting quantum interference DFT-D dispersion corrected, density-functional- device. theory. SWHB sandwich herringbone. DLS dynamic light scattering. DOSY diffusion-ordered NMR spectroscopy. TD-SCF time dependant self-consistent field. DSC differential scanning calorometry. TEM transmission electron microscopy.
EDX energy-dispersive x-ray spectroscopy. UV ultraviolet.
FEMM finite element method magnetics. VASP Vienna ab initio simulation package. FWHM full width at half maximum. VLS vapour-liquid-solid growth.
HB herringbone. VSM vibrating sample magnetometer. HFML high magnetic field laboratory. XRD X-ray diffraction. ICP-AES inductively coupled plasma - atomic Y123 YBa2Cu3O7−x. emission spectroscopy. Y211 Y2BaCuO5. MCG microcrucible growth. YBCO yttrium barium copper oxide.
xvii
lotalpyia rpriso oi aeilaeafnto ftewyi hc h sub- the which in way the of function a are material solid a of properties physical all Almost nt ftesldcnitrc.Oraiiyt uepoete ycnrligteeinteractions these controlling by properties tune to ability Our interact. can solid the of units nnwrswtotteaddition the without nanowires YBCO temperature grow to high attempt mechanism an its the Here had detail. of recently in has reported growth additive, and biotempate quantified Nanowire a as limited. alginate using are options YBCO and superconductor morphologies nanoscale the desired conductive at a working forming where when use important for little particularly of is is This wire required. metal is surface a example For applications. investigated. for be materials will designing field magnetic a of influence the number the reasons, small under these a crystal For to hydrocarbon conclusions. restricted contradictory a topic draw of a these growth is cases, It some alternative. in explored and use less publications the much of growth, a crystal is directed field non-intrusive, magnetic and for a pressure apparatus of temperature, as As investigated formation. been crystal all of have nature light stochastic the reduce matter to which employed in be way the control to which by means novel these for For other. look self-assembles. each to around necessary sub-units is these arrange it to reasons, ability our to linked incontrovertibly is Rationale 1.1 diinly h opooyo rsaln tutr so infiatiprac when importance significant of is structure crystalline a of morphology the Additionally, can measures various capricious, is often and be, can solution from growth crystal Although Teeaen tagtlnso hr onr nnature.” in corners sharp or lines straight no are “There 1 .Gaudi A. – I NTRODUCTION
C HAPTER 1 CHAPTER 1. INTRODUCTION
of a biotemplate will be attempted, by replicating the system generated by the alginate in order to mimic the mechanism of growth. Any addition to the crystal growers tool kit, increases the potential avenues of investigation when designing a new material.
1.2 Crystals and crystalline matter
‘Crystal’ or ‘crystalline’ refers to, in its most basic sense, matter formed of an ordered arrangement of its most basic constituent parts. These constituent parts can be ions, atoms or molecules and with a few notable exceptions[1], extend this order in three dimensions. In contrast, an amorphous or non-crystalline solid has no structural order and is typified by plastics, gels and glasses. Crystalline solids in an ideal sense, are thought of as impurity and defect-free as well as infinitely repeating in all directions. In practical terms however this is obviously not the case. A ‘single crystal’ refers to a single piece of matter that is all part of the same crystal lattice, that is the direction and orientation of the repeating units are identical, regardless of which part of the solid is being observed. Although likely to contain small numbers of defects, impurities and inclusions, techniques for observing the overall crystal structure overlook these imperfections. Single crystals are identifiable in nature by straight lines and sharp corners, exemplified by crystalline minerals like sodium chloride salt crystals, snowflakes and quartz prisms. ‘Polycrystalline’ materials comprise of numerous, randomly orientated, crystallites (small crystals) These materials have no particular favoured direction in which the crystal lattice propagates; i.e., there is no preferred orientation. As they are made up of many small grains of crystalline material, polycrystalline materials also contain grain boundaries, the interface where two crystallites meet. These grain boundaries can have different properties to the bulk crystalline solid and so can change the overall properties of a material. For obvious reasons, the smaller the gains, the more the properties of the grain boundaries impact the material as a whole.
1.2.1 Crystallisation
Crystallization, the growth of crystalline material, is a central theme in the preparation of materials and has a dramatic impact in medicine[2,3], biology[4], and materials science[5] as well as the food and manufacturing industries[6]. The continued development of novel drugs, proteins, and advanced materials strongly rely on our ability to self-assemble molecules in solids with the most suitable structure in order to exhibit desired functionalities. Naturally, a greater understanding of how molecules self-assemble into crystals can only increase our ability to design and build novel materials in order to tune properties. Primarily for these reasons, research into crystallisation mechanisms and fundamentals, has consistently been an active area of investigation across disciplines in which crystallisation is important. Currently, there are no ways to directly observe the coming together of small single subunits during the first steps of
2 1.2. CRYSTALS AND CRYSTALLINE MATTER
crystallisation. To observe single molecule addition to the crystal surface in order to understand why one nucleation centre grows while another dissolves is yet to be comprehensively understood. The size range, coupled with the speed at which these interactions occur restrict investigation to monitoring these phenomena using indirect methods and extrapolation.
1.2.2 Types of crystal
As mentioned earlier in section 1.2 crystals can consist of a regular arrangement of atoms, ions or molecules which form covalent crystals, ionic crystals and molecular crystals respectively. Each crystal type is dominated by interactions typical of the particular particle, basic cartoons of each type can be seen in figure 1.1. Covalent crystals are materials that are effectively large molecules, that is, each atom present is covalently bonded to its nearest neighbours. The covalent character of these interactions between atoms give rise to the extraordinary physical strength typical to covalent crystals, as to break them, requires the breaking of these bonds. The obvious example of a hard, covalently bonded crystal would be diamond, an extended molecule of sp3 hybridised carbon. For these reasons, covalent crystals do not melt in the standard use of the word, but tend towards decomposition into constituent elements and compounds. Another hallmark of covalent crystals is the inability to conduct electricity due to the lack of charged species in the material. Ionic crystals, as its name would suggest, consist of ions that are held in the lattice by strong electrostatic interactions. Typically these salts have a large lattice energy due to the high energetic cost of the ions existing individually, which give ionic solids a characteristic high melting point. Perhaps the most well-known example of this type of crystal being sodium chloride. Molecular crystals are solids compromising of an ordered array of molecules stabilised by van der Waals interactions. The weak nature of these intermolecular forces tend to make molecular solids soft, fragile and volatile, indeed some smaller molecule crystals will vanish in hours if left open to the atmosphere, or within minutes in a vacuum. This also grants molecular crystals relatively low melting points and again, with no mobile charged particles, electrical insulators, at least in the classical sense[5]. These crystals are typically made up of light elements, covalently bonded into discrete units that can dissociate when the temperature is sufficient to cause the solid to melt. These units also have character of their own which when bonding allows, unlike the previous crystal types, can change their effective shape via bond rotation and contortion which can give rise to polymorphism which is described in section 1.4. Good examples of molecular crystals would be water ice, sugar or the fullerenes.
1.2.3 The unit cell
Because of the repeating nature of a crystal, a subdivision of the structure must exist that is stackable, that is, can be translated in three dimensions as to describe the overall crystal structure. The smallest possible fragment that fits this description is called the unit cell. The
3 CHAPTER 1. INTRODUCTION
(a) (b) (c)
FIGURE 1.1. Graphical description of (a) a covalent crystal in three dimensions where each atom is bound to eight of the other atom, (b) a ionic crystal in two dimensions where each ion has four nearest neighbours of the opposite ion and (c) a molecular crystal in two dimensions containing a molecule in two different orientations (pink and green), all other molecules are symmetrically equivalent to one of the two. Different coloured spheres represent different types of element and lines represent chemical bonds.
unit cell can be unambiguously described in terms size, shape and symmetry based on how the subunits of a crystal stack. As an example figure 1.1 (a) does not describe the unit cell of a crystal with two elements, if this cell copied and stacked on top of itself, the elements would not match at the cell boundary, therefore it is not translatable. The unit cell for this crystal would in fact need to be eight times larger packed in a cubic arrangement. The locations in the unit cell occupied by atoms (or vacancies) are called lattice points, and these can be used to classify the unit cell. It was shown in the 19th century by Auguste Bravais that all possible arrangements of repeating matter must fall into one of 14 lattice structures (Bravais lattices) within 7 crystal systems shown in table 1.2.3. As can be seen, all lattices outside of these 14 can be reduced to other smaller defined Bravais lattices for example, a base-centred cubic arrangement can be redefined as primitive tetragonal. All unit cells are parallelepipeds described using lattice parameters, a, b and c for the lengths and α,β and γ for the cell angles, all of which are shown in figure 1.2. Typically, when an angle is known to be 90° or if cell lengths are known to be the same, based on the crystal system, they are omitted. For example a cubic unit cell can be defined using a single lattice parameter (a) or a monoclinic unit cell can be described using a, b, c and β. The unit cell is chosen as to have the minimum volume whilst still containing a complete arrangement of asymmetric units. As will be addressed later, for characterisation of crystalline matter, lattice planes are used to measure distances between lattice points. A lattice plane is a two dimensional plane that cuts through the unit cell in such a way that if extended outside the unit cell, every unit cell in its path would be divided in the same way. These are described using miller indices, three
4 1.2. CRYSTALS AND CRYSTALLINE MATTER
Crystal Cell Cell Body Face Base Primitive (P) system lengths angles centred (I) centred (F) centred (C)
a, b, c Tetragonal Cubic α = β = γ = 90° a = b = c (P)
a, b, c Tetragonal Tetragonal Tetragonal α = β = γ = 90° a = b 6= c (I) (P)
a, b, c Orthorhombic α = β = γ = 90° a 6= b 6= c
a, b, c Trigonal Trigonal Triclinic Trigonal α = β = γ 6= 90° a = b = c (P) (P) (P)
Monoclinic Orthorhombic Monoclinic Hexagonal† a, c γ = 120° (C) (P) (C)
a, b, c α = γ = 90° Monoclinic Monoclinic Monoclinic a 6= c β 6= 90° (C) (C)
α 6= 90° Triclinic Triclinic Triclinic Triclinic a, b, c β 6= 90° (P) (P) (P) γ 6= 90°
† The hexagonal system specifically cannot be described by three lengths and three angles due to the symmetry of a hexagonal cylinder being formed of three rhombi with large angle 120°, length a, and a height c.
TABLE 1.1. Description of the cell lengths and angles in the 14 Bravais lattices. Where a lattice type does not exist, the equivalent reduced Bravais lattice is stated.
integers in the form hkl that define the plane as three points in which the unit cell parameters are intercepted. These integers are the reciprocal of the intercept of each cell axis, for example the lattice plane (132) crosses the unit cell at a, b/3 and c/2. When an axis is never intercepted, like a plane running parallel to a unit cell face, a 0 is used (e.g. the plane (110) will never intercept the c-axis).
1.2.4 Naming conventions used in this thesis
To identify some differences between the packing types and indeed to differentiate between some of the same type, some term conventions will here be defined:
5 CHAPTER 1. INTRODUCTION
Standard notation of miller indices denotes a single plane as (hkl), with the family of symmetry equivalent planes written as {hkl}. The direction through a lattice in real space is written [hkl], which is a direction normal to (hkl) (fig. 1.2 (a)).
(a)
(c) (b)
(d)
FIGURE 1.2. (a) Naming conventions for unit cell parameters. a,b and c axes are shown by red, blue and green lines respectively and location of angles α, β and γ are illustrated. (c) and (d) define ∠mol and doff , respectively as used in this thesis. Positions of the equivalent blue and green molecules in the unit cell are shown in (b).
6 1.3. CRYSTAL NUCLEATION
1.3 Crystal nucleation
Crystals do not form spontaneously as large ordered structures, but rather grow outwards from an initial point. Nucleation refers to such a growth centre, at which crystal sub-units can build upon. That is, the initial particle that acts as a growth centre for larger crystals to form. The ultimate state of a crystal system, (i.e. crystal shape, crystal polymorph, crystal number) is fundamentally linked to how the crystals nucleate and these early stages of crystal growth[6]. Unfortunately these critical moments of crystal conception, of the order of 100-1000 atoms[6], occur over short time periods, are too small and transient (being in a liquid medium) to easily allow the dynamics to be quantified. For these reasons, theoretical means in which to describe this stochastic process have been postulated and are useful tools when trying to rationalise empirical observations. Typically, when growing a crystal in a lab, there are various methods and mechanisms that are known to induce formation of these growth centres to propagate crystal growth. These crystal growth techniques are classified as either primary or secondary nucleation[7,8].
1.3.1 Secondary nucleation
As the least pertinent to this work, secondary nucleation will only be briefly described. Secondary nucleation refers to the process of introducing a small crystalline particle to a system containing homogeneous sub-units to ‘seed’ growth of larger crystals. This process is widely known as ‘seeding’. If used under the wrong conditions however, in a low supersaturation or heated solution for example, the seed itself can dissolve instead of inducing growth, losing molecular order and polymorphic identity. This form of nucleation becomes particularly important when investigating solution based crystal growth systems; as if a volatile solvent leaves too quickly from the liquid surface, a localised solute concentration spike can form small crystals near the solution surface. This can result in premature crystallisation within the system making repeatable and consistent crystal growth or data collection difficult.
1.3.2 Primary nucleation
Primary nucleation refers to systems where crystal growth occurs when no crystalline material is present or added. This can be due to foreign impurities like dust, presenting nucleation sites (heterogeneous nucleation) or can happen spontaneously (homogeneous nucleation) when conditions become favourable. Primary nucleation also commonly occurs around scratches on the growth vessel, agitation (e.g. shaking a solution)[9] or thermal shock. Although the growth of a crystal is something that can be observed in situ. and studied in detail[10], the initial nucleation is something significantly more difficult to empirically examine. Indeed theoretical modelling and experiments reporting evidence of nucleation mechanisms
7 CHAPTER 1. INTRODUCTION
are still frequently published[11,12]. The differences between the chemical character and growth environment of each separate system make a general rule of nucleation all but impossible. Classically, crystal formation was assumed to be a gathering of sub-units forming order via addition and is thus described as the tipping point when free energy generated during the clustering of molecules (∆Gv) exceeds that consumed by the formation of a growing surface area (∆Gs). Therefore, as crystal nucleation was considered analogous to droplet formation in a gas/liquid system[6,13] the total free energy of the system is the sum of each quantity:
(1.1) G = ∆Gv + ∆Gs
3 2 Based on spherical clusters, ∆Gv and ∆Gs have a dependency of r and r respectively.
4 3 2 (1.2) = πr ∆Gv + 4πr ∆Gs 3 As the volume and surface free energies have different signs, finding d∆G±dr = 0 allows a maximum to be found for ∆G as a function of cluster radius (rc).
∆ d G 2 (1.3) = 4πr ∆Gv + 8πr∆Gs = 0 dr
−2∆Gs (1.4) ∴ rc = ∆Gv
Once rc has been exceeded, crystallisation around that nucleus becomes favourable, as any further increase in size will only lower the total free energy of the system. This is represented graphically by the blue line in figure 1.3.
FIGURE 1.3. Classical representation of how crystal growth can propagate as a function of cluster size.
8 1.3. CRYSTAL NUCLEATION
As the probability of so many molecules simultaneously colliding into a cluster size with a radius > rc is low, the addition of single molecules, at a rate faster than re-dissolution, in isolated pockets of differing concentration was regarded as the most likely mechanism of cluster formation[6].
Mol + Mol Mol2
(1.5) Mol2 + Mol Mol3
Moln−1 + Mol Moln This has become known as classical nucleation theory (CNT). In CNT (equation 1.5) n is the number of molecules needed to form a cluster with radius rc. CNT is also known as ‘one-step nucleation’ as it only has a single, pre-critical, crystalline particle (or pre-critical aggregations (PCA)) phase before larger crystal growth occurs. However, as CNT is based on droplet formation, it is perhaps overly simplistic for the formation of ordered matter from solution as it, by definition, comes with a series of assumptions which do not necessarily transpose to the formation of ordered matter.
• Clusters exclusively form as spheres. Although this may be true to an approximation in some systems above a certain number of molecules, non-spherical PCAs have been observed experimentally[14]. The low numbers of molecules making the differences between surface and bulk material ill-defined.
• Clusters have a consistent density throughout which is indistinguishable from the bulk material (therefore ordered) and have no concentration or density gradients within the cluster. The internal ordering of PCA in such small sizes is difficult to measure. That being said, amorphous pre-nucleation clusters (PNCs) have been observed on certain systems[11]. This obviously will give the PCA a different internal structure than the final bulk material.
• Cluster formation has no effect on the concentration of feed material in its immediate environment. As a crystal grows in solution it necessarily contains a progressively larger number of particles. This leads to a zone of solute-depleted solution in the immediate vicinity of the crystal which can be directly observed in situ. [15]. These differences form concentration gradients between discrete pockets within the system.
• Aggregation or separation of pre-nucleation clusters does not occur and growth is restricted to the transport of monomer units only. Again, this has been observed to be inaccurate as a blanket rule of crystal nucleation. Aggregation of amorphous calcium phosphate has been shown to be a pre-cursor to growth of hydroxyapatite[16].
9 CHAPTER 1. INTRODUCTION
Although quantitatively, CNT has provided an accurate paradigm in which to work, more recently other non-classical pathways to crystal nucleation have emerged based on both simulation and empirical observations. Two-step nucleation is described as the formation of an ordered crystal nucleus via a liquid- liquid phase of solute and solvent. That is, the formation of small disordered liquid particles of solute suspended in the solvent. Once a high enough local concentration has aggregated, an ordered nucleus will form within after this PNC has reached a certain size, propagating crystal growth. It is fundamentally different from CNT insofar as it describes this liquid phase as a necessary mid-step between a homogeneous solution and a crystal nucleus of critical size. First proposed in 1997 using a simulation[17], it has been applied to numerous systems both organic[18] and inorganic[19]. It has since been established empirically as a pathway for protein crystallisation[20].
1.4 Polymorphism
Polymorphism refers to the ability of a crystalline material to pack in more than one way while remaining stoichiometrically identical. The change in packing can result in a difference of lattice parameters, symmetry, space group or even crystal system and crystal morphology. As a result, interactions between subunits will differ which, as mentioned earlier, alter the electronic or physical properties of the bulk material. In molecular crystals, the first instance of reported polymorphism was with the molecule benzamide[21]. It was stated ‘Langsam erkaltend und bei einer gewissen Concentration erstarrt die ganze Flüssigkeit Zu einer weissen Masse , die aus sehr feinen, seideartigen, dem Caffein ähnlichen, krystallnadeln betect.’ - roughly translated to ‘slowly cooling and at a certain concentration the whole liquid solidifies into a white mass, which is made of very fine, silky, caffeine-like, crystalline needles’, which was not the expected morphology of benzamide. Once left to stand it was noticed that the needles had transformed into rhombohedral crystals which, had the change been purely morphological would have been unlikely to occur. Figure 1.4 describes how polymorphism can manifest whilst using a single sub-unit, the red and black arrows indicate how a change in a physical parameter during growth can be used to encourage growth of a preferred polymorph. The forming of different polymorphs requires the altering of physical parameters either during crystallisation[22] (e.g. changing the solvent) or affecting prefabricated crystals (e.g. by applying pressure) [23]. Crystal systems that contain at least two polymorphs can be either enantiotropic or monotropic[24]. If two polymorphs are interchangeable whilst remaining a solid, either directly from polymorph A to polymorph B, or via an amorphous phase, they are known as enantiotropic. If the different polymorphs cannot inter-convert without adopting a disordered (gas, liquid or amorphous) intermediate phase, the polymorphs are known as monotropic. Figure
10 1.4. POLYMORPHISM
FIGURE 1.4. Description of polymorphism using a 2 dimensional arrangement of blue rectangular sub-units.
1.5 displays energies of a typical (a) enantiotropic and (b) monotropic polymorphic system. In both graphs, the red and blue lines represent the enthalpy and free energy (respectively) of different components of a two polymorph crystal system. Figure 1.5 (a) demonstrates how, by increasing the temperature, the lowest energy form for the system changes from polymorph I to II at point
‘Z’, which subsequently melts at point ‘X’. Polymorph I, being less stable above TpI/II, melting at point ‘Y’. In figure 1.5 (b) however, it can be seen that as polymorph II is, at no point, the most stable form before the system becomes liquid, polymorph I remains the lowest energy geometry until it melts at point ‘Y’.
(a) (b)
FIGURE 1.5. Energy diagrams for (a) enantiotropic and (b) monotropic polymorphism. Red and blue lines represent enthalpy and free energy respectively of polymorph I (dotted line) polymorph II (dashed line) and loss of crystal ordering as transition to liquid or gas (solid line). Green arrows represent the enthalpy of fusion of various transitions.
11 CHAPTER 1. INTRODUCTION
1.4.1 Naming polymorphs
Unfortunately, due to the extended time in which different polymorphs have been classified, there is some inconsistency regarding the naming of crystal polymorphs. This becomes confusing when attempting to compare structures throughout the literature as polymorph nomenclature is apparently in the hands of whichever author is publishing at the time. This is exemplified succinctly by Grzesiak et al. (table 1 in reference 25) referring to how the naming of four anhydrous polymorphs of carbamazepine have changed from paper to paper. The polymorph is sometimes identified by a prefix (e.g. α-, β-, γ-, δ-, -, ζ-, η-[26], polymorph A, polymorph B, polymorph C[27], HB-, SHB- [28]) or a suffix (e.g. (A), (B), (M)[29], form I, form II, form III, form IV[30],-α-, -δ[31]), these examples are not exhaustive and do not even consider calculated structures, solvates or co-crystals. Conventions are typically based on the order of discovery, the energy of the system or some physical consideration of the unit cell.
1.4.2 Metastable polymorphs
A saturated solution refers to a solution in equilibrium with a solid, at a given temperature. Typically the relationship requires higher temperatures in order to solvate an increasing amount of solute, i.e. a warmer system has a higher saturation. When a saturated solution is cooled below this temperature without nucleation sites or agitation, it is fairly common to obtain a solution with a higher solute content than that of the equilibration solution at this new temperature. These solutions are referred to as supersaturated, a state of which is necessary for crystallisation from solution to occur. In 1897 Wilhelm Ostwald suggested how the process of crystallisation from a supersaturated solution may not be as simple as matter arranging itself into the most thermodynamically stable conformation from the first sub-unit[32]. Rather, that between the first clustering of matter to the production of a large crystal, the system transitions through a variety of intermediate states or polymorphs, this has become known as Ostwald’s rule of stages:
“An unstable system does not necessarily transform directly into the most stable state, but into one which most closely resembles its own, i.e. into another transient state whose formation from the original is accompanied by the smallest loss of free energy.” [33]
A simplified depiction of this rule can be seen in figure 1.6. The experimental parameter can be any variable driving crystallisation (e.g. decrease in temperature, loss of solvent) and indicates how metastable phases MI and MII are formed as growth proceeds. Each blue asterisk represents a state of instability between stable and metastable forms, indicated by the energy wells. In this sense ‘metastable’ is defined as a crystalline state which is not the thermodynamically most stable form under at a given temperature and pressure. That is, there is some conformational energy-
12 1.5. MORPHOLOGY
FIGURE 1.6. The energy of polymorphs during crystal growth as described by the rule of stages. mI and mII indicate the positions of meta-stable polymorphs. Blue asterisks show the locations of instability. Double headed arrows show the energy barrier from a meta-stable polymorph to a lower energy form and the green hexagon shows the position of the thermodynamically stable polymorph. Experimental parameter could be any physical parameter used to grow crystals.
well in which the solid can sit before overcoming an activation energy (Ea) and transforming to more stable (lower energy) form. The stability of a metastable polymorph is defined by its ability [34] to overcome Ea, rather than the energy difference between two forms . In some cases, the metastable polymorph of a given material can posses properties more desirable than the thermodynamically most stable form, in which case these metastable forms can sometimes be harvested by a quench of the growth process via a drop in temperature or removal from the growth medium in order to make the Ea to a lower energy form insurmountable in the current conditions. It should also be noted that hydrates and solvates can also be involved in the rule of stages. This rule, although thought to have no generic proof[33] may have provided a rudimentary description of multi-step nucleation, as mentioned earlier in section 1.3.2. That is, during precipitation of a crystal from solution where multiple polymorphs are possible, it is not the thermodynamically most stable form that is formed first, but the one with the fastest rate, that is, the one kinetically favoured.
1.5 Morphology
Morphology refers only to the larger, three dimensional shape of the crystal. Unlike a difference in polymorph, a change in morphology does not necessarily describe a change in unit cell parameters or packing, although it can sometimes be an visual indicator of a polymorph change. The term ‘crystal habit’ is used exclusively to describe morphological changes with the same polymorph. The morphology of a crystal can be an important factor when developing materials for applications. If
13 CHAPTER 1. INTRODUCTION
a crystal has a plate-like or needle-like morphology, it means that certain crystal faces have grown anisotropically. Faces that propagate relatively slowly end up over-expressed in the morphology. In the pharmaceutical industry, if this particular over-expressed face has a tendency to stick to metals used in milling machines, this can cause a problem bringing an active pharmaceutical ingredient (API) to market. An API must not only be adequately soluble and bio-available when swallowed, but must also be amenable to being processed, that is, crushed, milled and if necessary, mixed with an excipient like calcium carbonate. In molecular crystals, anisotropic morphologies are common due to the variety of intermolecular interactions within a unit cell, for example the standard form of the API carbamazepine has hydrogen bonding in the crystal almost perpendicular to the direction of π-stacking.
1.5.1 Polymorph directed morphology
As mentioned in section 1.4, the polymorph dictates the packing within the unit cell which necessarily changes the ways in which molecules interact with their nearest neighbours. In some cases this can have a dramatic effect on the morphology of the crystal. A good example of this is in the pain and fever medication paracetamol which has two stable polymorphs[35] (form I and form II). The ubiquitous form I forms easily via crystallisation from ethanol and typically forms in large rhombohedral crystals with apparent similar growth rates in three dimensions. The more difficult to grow form II, known to crystallise from the melt, typically forms highly anisotropic fibre-like crystals with rapid growth in [001] relative to all other directions[36]. In this way, the polymorph has directed the morphology of the crystal. This is not an isolated case[21,37] and with known systems can be an indicator of growth evolution. Unfortunately, polymorphs that exclusively form highly anisotropic crystals can be hard to identify based on the lack of crystals large enough for structure solution. In these cases, morphological control is a boon.
1.6 Characterisation of crystalline matter
1.6.1 X-ray diffraction
Ultimately, when designing and building crystalline materials, identifying the position of atoms in the unit cell is essential for characterisation. Luckily, the repetitive nature of crystals allow for interrogation of these materials using the diffraction of radiation with wavelengths comparable to the unit size of the unit cell. After Wilhelm Röntgen first quantified X-rays in 1895, the use of X-rays in the following decades became more common place. In 1913 William Henry Bragg published ‘The reflection of X-rays by crystals’[38] with his father William Laurence Bragg, describing observed points of intensity over the arc of reflection from incident X-rays when using crystalline materials. The angle of these points of intensity were used to derive the Bragg relationship (equation 1.6)
14 1.6. CHARACTERISATION OF CRYSTALLINE MATTER
between the wavelength of incident radiation, the measured angle and a discrete set of parallel planes.
(1.6) nλ = 2d sinθ
Where n is any integer, λ is the wavelength of incident radiation, d is the spacing between equivalent reflection planes in the unit cell and θ is the angle of incident radiation towards the sample stage. This relationship works, as constructive interference between radiation elastically scattered by equivalent planes can be detected as increased intensity at the specular angle from the sample. Figure 1.7 attempts to give an overview of Bragg diffraction.
FIGURE 1.7. Graphical summary of Bragg diffraction in crystalline systems.
The angle θ is the angle at which the coherent incident radiation (red lines) approaches the material. All matter interacting with the radiation will cause it to scatter, if this scattering is elastic, there is no change in wavelength (λ). Equivalent points in the material (green spheres) that are can be symmetrically translated perpendicular to the reflection plane (black lines) scatter the radiation with a difference of the translation distance (d), or a multiple of. Using the distance d and the angle θ it can be shown that the radiation that travels to the first translation (d) will travel 2(d sinθ) further if scattered at a specular angle (θ), d sinθ more before reaching the point and d sinθ after being scattered. If this extra distance travelled is equal to a multiple of the wavelength of the incident radiation (nλ where n= any integer), then an increase in radiation intensity will be observed due to constructive interference at the angle θ from the reflection plane and 2θ from the incident radiation. Other common arrangements of the Bragg equation allow calculation of d-spacings from a given θ
15 CHAPTER 1. INTRODUCTION
µ nλ ¶ (1.7) d = 2sinθ
and an expected θ when looking for a specific d-spacing.
µ nλ¶ (1.8) θ = arcsin 2d
1.6.1.1 Single crystal X-ray diffraction
The standard means in which the structure of a crystal is solved is with single crystal X-ray diffraction (scXRD). This technique irradiates a single crystal of material with coherent X-rays, typically from a Cu or Mo source (λ=1.54056 Å and λ=0.71073 Å respectively[39]). For these reasons it is essential that the crystal be single and of sufficient quality, as crystal twinning and high numbers of defects will give ambiguous diffraction patterns. These X-rays are scattered as they pass through the ordered matter and cause regions of increased intensity when the Bragg relationship is satisfied in two dimensions, generating an array of spots. These spots are tracked across the 2D detector as the crystal is rotated in both position and intensity. Due to the data required to solve the structure, a complete data set can take a relatively large amount of time to collect. This information can then easily be converted by computer into areas of electron density in real space and used to generate a full picture of the unit cell.
1.6.1.2 Powder X-ray diffraction
When acquiring a sufficiently sized single crystal is not possible or when a quicker form of structural analysis is required, pXRD is commonly used. In contrast to scXRD, pXRD requires a polycrystalline sample for a complete data set as it relies on a random orientation of small crystallites in order to generate a diffraction pattern. A typical X-ray powder diffractometer uses a Cu X-ray source and a one dimensional detector, that is, a detector that only reads at a single angle 2θ. This angle is then swept across a 2θ range relevant for the material, allowing all potential reflection planes to generate an increase in intensity when 2θ satisfies this relationship. This random orientation allows for all possible reflection planes in a crystal structure to be in a position to satisfy the Bragg relationship at some angle of 2θ. Practically speaking, all possible reflection planes will be represented by the orientation of a percentage of the powdered sample. As each peak in a pXRD pattern represents a specific lattice plane, it can be indexed using miller indices (hkl). This can reveal much information about the unit cell, especially when being compared to a known pattern. For example, if on a powder pattern taken using Cu Kα radiation, the reflection indexed as (100) is found at 26.89° 2θ, then using equation 1.7, the d-spacing between these planes and thus the length of the a-axis is found to be 3.31 Å.
16 1.6. CHARACTERISATION OF CRYSTALLINE MATTER
This becomes more difficult when attempting the analysis of an unknown pXRD pattern. The random orientation of the sample can give no information on plane direction, that is, reflections in the family 100 and 010 are indistinguishable from each other if unknown. Interplanar spacings can be extracted whereas no information on reflection angles is present. Therefore, although solution from powder has been possible for some decades[40,41] it is a complex process requiring high-resolution data and relies on software-assisted allocation of peak position to the space group in order to index peaks. As pXRD data is effectively two dimensional data compressed into one, peak overlap complicates not only identifying peak positions but also extraction of peak intensities which is necessary for elucidating atomic positions. Problems with texturing can also cause a variation in reflection intensities as in practice a sufficiently random, isotropic sample is not always present. Ultimately, this is why pharmaceutical patents, to ring-fence as many potential polymorphs as possible, often find an unidentified pXRD pattern after a new preparation process and declare it a new polymorphic form with no knowledge of the structure of the unit cell.
1.6.2 Rietveld refinement
The diffraction pattern created upon collection of powder diffraction data, as mentioned in section 1.6.1.2 is routinely used as a comparison to reported or calculated structural data. Rietveld refinement facilitates the extraction of additional structural information from a powder pattern by thorough analysis of peak height, full width at half maximum (FWHM), shape and position which allows it to mitigate loss of information due to overlapping peaks. Originally conceived in 1967 by Hugo Rietveld[42], it was developed over the next 2 years and reported in 1969, intended for use with neutron diffraction[43]. Easily translatable to any diffraction technique based on a monochromatic radiation source, a theoretical profile is generated based on approximate initial input parameters which, are incrementally adjusted to fit an experimental pattern using least- squares regression analysis. The convergence criteria for refinement is minimisation of the function M when:
X n 1 o2 (1.9) M = wi yi(obs) − yi(calc) i c
Where wi is a statistical weighting parameter based on corrected intensity, yi(obs) and yi(calc) are the observed and calculated line profiles respectively and c is a scaling factor so [43] that yi(calc) = c.yi(obs) . Finally, i indicates an intensity measured at the angle 2θi. When using Rietveld refinement as part of computer software, there are 3 main parameters that need 2 reviewing before any further information is extracted: namely Rexp, Rwp and χ . The expected R- value (Rexp) describes the experimental data quality by determining the lowest R-value possible.
17 CHAPTER 1. INTRODUCTION
½ (n − p + c) ¾1/2 R = × 100 (1.10) exp P 2 wi yi(obs) i Where n is the number of data points collected, p the total number of refined parameters and c the number of constrained parameters. Therefore, as (n − p + c) is normally dominated by n, a higher resolution pattern will give a lower potential R-value. The weighted-profile R-value (Rwp) is related to the above equation 1.9 and compares how a model based on the input parameters resembles the experimental data. It is defined as:
P © − ª2 ½ wi yi(obs) yi(calc) ¾1/2 i (1.11) R = × 100 wp P 2 wi yi(obs) i
As Rwp is essentially attempting to mimic the Rexp value, the relationship Rwp ≥ Rexp should always be true. If Rwp < Rexp, it is an indication that convergence has failed. Finally, the parameter χ2 is a statistical measure of the goodness of fit and is often used to assess whether a fitting has been successful in fitting parameters. If the fitting seems unable to get below χ2 = 5 it is likely that there are incorrect parameters when fitting the model to the data, possibly space group or texturing. χ2 effectively describes the goodness of fit between the experimental and theoretical patterns line profiles.
2 2 (1.12) χ = (Rwp/Rexp)
2 Although lower Rwp and χ values normally indicate a better fitting of the model, unfortu- 2 nately there is no specific value of Rwp or χ that gives an official ‘thumbs-up’ for the refine- ment[44]. This means the fitting of pXRD data has drawbacks from pattern to pattern. If the signal to noise is low, i.e. there is a significant contribution to the total intensity of the pattern 2 from the background, the Rwp and therefore the χ can be artificially low when compared to a higher resolution pattern due to experimental imperfections having a larger impact. When using a fitting to correlate different data sets, using the same machine, sample amounts and collection parameters makes a more valid comparison.
1.6.3 Multiphase Rietveld refinement
When there is a sample with a number of phases involved, either an incomplete reaction, a synthesis with two resultant polymorphs, or a mixture that is difficult to separate, the pXRD pattern will contain a summation of both patterns. In its simplest case this manifests as peaks of phase I containing interstitial peaks of phase II. In this case each pattern can be indexed, with a fitted model that is the sum of a model for phase I, a model for phase II up to phase n.
18 1.7. SUMMARY
(1.13) clac = (I) + (II) + (III) + (n) yi yi yi yi ... yi
This is know as multi-phase refinement and can assign a scale factor to each of the present phases based on the reflection intensities, allowing for determination of the volume fraction of that phase (of crystalline material). Luckily if peaks are found to overlap then, as with single phase refinement the intensities of the stack peaks should be reflected in the model. Although small volumes of phases become less accurate due to the lack of statistical weights, it is a powerful technique for the identification and quantification of target and intermediate phases, not to mention weighting a mix of polymorphs within a single system.
1.7 Summary
When affecting the system of crystal growth and synthesis in various way, structural analysis of the product is essential for determining how effective an introduced variable has been. The following chapters describe a variety of crystalline products that have been structurally altered using two distinctly different ways of changing the system and attempts to describe how each phenomenon occurs. Based on the importance of fine-control during crystal growth, it is hoped that the information contained in this thesis will add, in some small way to the crystal growers ever- expanding tool-kit, allowing a higher degree of control over how crystalline material arranges.
19
ihu sn aumges nteisd ftegon ls ek odfigrwsinserted was finger cold a neck, glass ground the of inside the on grease vacuum using Without Approximately h r ftefls.Oc h aumwstre n h a nteamwsoee lwy as slowly, opened was to arm fitted the line on vacuum tap a the and on, finger turned bath cold was oil the vacuum an of the in tip Once submerged the flask. then with the was level of flask was arm The oil the seal. the a of create surface to the joint so glass up the of top the to applied between tip its have to as sublimation via precursors crystalline of Purification 2.1.1 Materials 2.1 methods following parameters. the using stated conducted the were employing thesis instruments this and in chapter any in contained techniques by publication. directly for collected up not written was and data supplied Where were herein. parameters contained instrument text myself, and figures all of preparation the 45–48 references 2018- HardwareX and 2017 Reports Scientific Nature nesohriesae,aleprmna rcdrs apepeaainadanalytical and preparation sample procedures, experimental all stated, otherwise Unless involving myself by submitted and written were chapter this of creation in used papers All arXiv), on preprint (with 2016 Communications Nature in published are chapter this of Parts Tk ehdadtyi.I tfis di tfaky n r nte.Btb all by But another. try and frankly, it admit fails, it If it. try and method a “Take en,tysomething.” try means, 0 . g 2 fognccytl eepae ttebto facen r cln flask. Schlenk dry clean, a of bottom the at placed were crystals organic of 1 . cm 5 rnlnRoosevelt Franklin – and cm 2 rmtesatn aeil aumges a then was grease Vacuum material. starting the from 21 M TRASAND ATERIALS
M C HAPTER ETHODS 2 CHAPTER 2. MATERIALS AND METHODS
to de-gas the material with as little disruption to the system as possible. The oil bath was then slowly brought up to temperature and left under vacuum to sublime. After a bulk of crystals had formed on the cold finger tip, the flask was lifted carefully from the oil bath and the vacuum relieved. The vacuum grease was wiped from the top of the joint between the cold finger and Schlenk tube and cold finger was lifted carefully out of the (now dry) glass joint and crystals harvested. The cold finger was replaced and the cycle repeated until the majority of the starting material was used up. The harvested crystals were again sublimed as above. The oil bath temperature used was specific to each material and kept as low as possible to avoid the carrying of impurities to the cold finger.
2.1.2 Purification of crystalline precursors via solution recrystallisation
Organic starting material was dissolved in a minimum of hot solvent and filtered using filter paper to remove any insoluble impurities. The filtered solution was then slowly cooled and crystals allowed to form overnight. The crystals were then harvested via filtration, again using filter paper and left to dry in air. This process was repeated a further two times before the crystals were used.
2.1.2.1 Coronene
Coronene crystals (Sigma Aldrich, purity 97 % (C84801)) were twice purified via sublimation as mentioned in section 2.1.1 at 180 ◦C. Ultimate purity was assayed using by nuclear magnetic resonance (13C and 1H). Structural analysis was conducted by, pXRD and scXRD. Different sizes of coronene crystals for the size dependant cryo-pXRD (section 3.3.1.3) analysis were all prepared using the purified coronene crystals dissolved in toluene. The largest crystals (fig. 3.11 (b)) were prepared by flash-cooling a vial containing a hot, saturated solution of coronene in toluene by submersion of the vial in a bath of water at ambient temperature. The resultant yellow (visibly acicular) crystals formed after approximately 30 s and were collected via filtration. Mid-sized crystals of coronene (fig. 3.11 (c)) were obtained using crystals collected from above placed, dry in a vial. The vial was then subjected to thermal-shock by an alternating submersion in liquid nitrogen and a warm water bath (50 ◦C) 25 times. This caused the crystals to shatter and resulted in a fine yellow powder. The finest crystals (fig. 3.11 (d)) were synthesised using a slightly modified method used to form rubrene micro-crystals[49] by injecting a hot, saturated solution of coronene dissolved in toluene quickly into a vial of cold, sonicating hexane anti-solvent. The solution was allowed to warm to room temperature and the solvent evaporate whilst still under sonication and the solid extracted from the anti-solvent via settling, decanting and drying.
22 2.2. METHODS
2.1.2.2 Inorganic materials
All inorganic materials were used as received from the supplier without further preparation.
Yttrium oxide (Y2O3, SigmaAldrich 205168), copper oxide (CuO, SigmaAldrich 544868), barium nitrate (Ba(NO3)2, SigmaAldrich 202754), ammonium hydroxide (NH4(OH), SigmaAldrich
09859), sodium chloride (NaCl, SigmaAldrich S7653), sodium carbonate (Na2CO3 , SigmaAldrich
S7795), bismuth oxide (Bi2O3, SigmaAldrich 202827), calcium nitrate tetrahydrate (Ca(NO3)2 ·
4H2O, SigmaAldrich C4955), strontium nitrate (Sr(NO3)2, SigmaAldrich 204498), iron chloride hexahydrate (FeCl3 · 6H2O, SigmaAldrich 31232).
2.1.3 Solvents
All solvents used, unless stated otherwise, were purchased laboratory grade and used as received with no further preparation.
2.2 Methods
2.2.1 Glassware cleaning
In every case involving crystal growth, all post-filter glassware was inspected to be scratch-free before being cleaned in seven steps. This has been included in order to impress the absolute need for a growth vessel with no potential sites for nucleation.
1. Scrubbing using a brush or pipe cleaner with an excess of soap and warm water, paying particular attention to any corners.
2. Filling the vessel with 0.7 M nitric acid for 30 min before rinsing with deionised water.
3. As above, repeated with 0.7 M sodium hydroxide.
4. Thorough rinse with deionised water, 15 to 20 times.
5. Rinsed with ethanol.
6. Rinsed with acetone.
7. Rinsed with the solvent used during crystallisation (no anti-solvent) before being dried, inverted in an oven.
2.2.2 Inductively coupled plasma
Trace metals that both could be an impurity and could have an impact magnetically (Fe, Co and Ni) were tested for via inductively coupled plasma - atomic emission spectroscopy (ICP-AES) using an Agilent 710 ICP-AES. 20 mg of coronene was digested in 5 ml of 1 % nitric acid (Sigma
23 CHAPTER 2. MATERIALS AND METHODS
Aldrich, ≥ 99.999 %, trace metal basis, 225711). The resultant solution was filtered through a 0.4 µm filter, analysed and compared to samples of Fe, Co and Ni prepared at 0.5 ppm (500 µgl−1) as a standard.
2.2.3 Dynamic light scattering
For particle size analysis using dynamic light scattering (DLS) a solution of coronene saturated at 93 ◦C for 7 d in toluene was transferred into quartz cuvette of path length 1 cm, kept at 93 ◦C by extrusion through a 0.22 µm syringe filter before being sealed. Using a ferry of freshly boiler water, the cuvette was transferred to the pre-heated (90 ◦C) sample space of a Malvern Zetasizer Nano S DLS instrument at which point it was allowed to stabilise for 30 min. The temperature program was then set to collect an averaged data set of ten passes at −5 ◦C increments from 90 ◦C to 20 ◦C with a equilibration time of 5 min at each temperature. The refractive index for coronene was set to 1.77[50].
2.2.4 Diffusion orientated NMR spectroscopy
Diffusion-ordered NMR spectroscopy (DOSY) was carried out on a Varian S500b 500 MHz nuclear magnetic resonance (NMR) spectrometer. Due to temperature gradient issues inside the NMR machine, only room temperature analysis was carried out. A saturated solution of coronene in deuterated toluene was prepared by dissolving an excess of coronene in 5 ml toluene-d8. This solution was left at 60 ◦C for 48 h and left to sit at room temperature for 1 h before being pipetted into an NMR tube through a filter paper plug. A second sample was prepared by diluting this room temperature solution by a half with an addition of toluene-d8 (1 mm toluene-d8 into 1 mm of saturated coronene solution). This second solution was then transferred to an NMR tube, again through a filter paper plug. The concentrations of both solutions were determined post-analysis via evaporation. Each set of DOSY data collected 15 sequential spectra with peak height decay fitted over all 15.
2.2.5 Crystal growth under a magnetic field
For successful and repeatable nucleation in solution based systems, homogeneity and consistency of the solution is essential. Although generally thought of as stochastic, variations in crystal growth can be minimised with careful attention to experimental parameters. The crystal growth systems must be introduced to the growth environment with minimal agitation, nucleation centres, solvent loss and as little temperature variation as possible. The precursor crystal powder was prepared using the methods highlighted in section 2.1.1 before being added in excess to the solvent and left for at least 7 d to saturate at the relevant temperature. The saturated solution was then extruded through a 0.22 µm polytetrafluoroethylene (PTFE) syringe filter pre-heated to the same temperature, into a clean growth vessel (section 2.2.1). Without letting the system
24 2.2. METHODS
cool, it was then gently transferred to a magnetic sample space and left to crystallise via either evaporation or via a cooling temperature ramp.
2.2.6 Coronene growth up to 1 T using permanent magnets
Figure 2.1 shows the experimental set up used to initially grow crystals under a magnetic field. A NIB magnet with dimensions 40 × 40 × 20 mm with a magnetic field strength of up to 0.63 T was carefully held at a distance of 1 cm from a magnetised iron pole-piece from an old vibrating sample magnetometer (VSM) with a magnetic field strength of up to 0.82 T using Al SEM sample stubs. The cavity formed between the two permanent magnets created a sample space with a magnetic field of 0.92 T to 1.09 T. A clean, dry quartz cuvette with 5 mm path length was inserted into this sample space and the whole apparatus was place in an oven set to 93 ◦C. A supersaturated solution of coronene (2.5 mgmL−1) in toluene was prepared and stored in the same oven at 93 ◦C. The solution was then passed through a 0.22 mm PTFE filter directly into the 5 mm path-length cuvette and sealed with a PTFE stopper to prevent solvent loss. Once sealed, the whole system was maintained at 93 ◦C for 4 h to allow the temperature to stabilise. The oven was then programmed to cool to 83 ◦C, 73 ◦C and 63 ◦C then finally 50 ◦C, with a 24 h isotherm at each temperature, before cooling to room temperature.
(a) (b)
FIGURE 2.1. (a) Schematic of the initial magnetic growth experiment and (b) a photo- graph of the experiment. A is the 1 T magnet; B the sample vial; C the magnetic holding plate. Field lines from the 1 T magnet through the experiment are indi- cated.
25 CHAPTER 2. MATERIALS AND METHODS
2.2.7 Crystal growth under higher magnetic fields of up to 25 T using Bitter electromagnets
In order to translate the above method from section 2.2.6 in order to allow for higher magnetic fields of up to 25 T large Bitter electromagnets at the high magnetic field laboratory (HFML) facility in Nijmegen were used. The solution was saturated and filtered as above before being ferried in a beaker of boiling water to minimise any dip in temperature to the sample rod pre-heated to 90 ◦C. The rod was then gently inserted into the bore of the magnet and the whole system was held at 90 ◦C until stable. Once stable, the magnetic field was applied before the temperature was decreased at a rate of 1.25 ◦Cmin−1. Figure 2.2 is a schematic of the sample holder used to deliver the cuvette into the bore of the magnets. In situ spectroscopy was used to record the concentration of coronene as a function of temperature using a Ocean Optics, DH-2000-BAL, balanced deuterium, halogen light source and Ocean Optics USB2000+ custom spectrometer. Optical cables were run down the sides of the sample rod and attached to lenses focussed on 45° mirrors, to carry the light from the source, to the spectrometer via the sample. The path of light through the sample was adjusted to be 1 cm from the bottom of the cuvette as to avoid scattering from crystals large enough to fall out of solution, therefore a drop in absorbance and thus concentration was used to identify the onset of crystallisation. The temperature of the sample space was controlled using an Julabo FP51 lab heater/chiller pumping water through a ‘tube within a tube’ fed from bottom to top.
2.2.8 Synthesis of Fe2O3 nanoparticles
Iron oxide nano-particles were prepared via the co-precipitation FIGURE 2.2. Schematic of method. A solution of 0.05 M FeCl · 6H O in oleylamine (Sig- the sample stage and 3 2 maAldrich O7805) was prepared (0.8110 g in 100 mm) and probe for spectroscopic placed in a sonicator with an overhead stirrer. 300 ml of 0.05 M measurements at fields NH OH was added drop-wise overnight whilst stirring and son- of up to 25 T for cell 4 icating. The resulting solution was left at 110 ◦C for 48 h to com- 5 at the HFML, Ni- pletely dehydrate before being heated to 450 ◦C at 1 ◦Cmin−1 to jmegen. remove any carbonatious material and NH4Cl. All solid remains
26 2.2. METHODS
were ground, collected and analysed using SEM and X-ray diffraction (XRD).
2.2.9 Preparation of Fe2O3 polluted coronene systems
A supersaturated solution of 3 mgml−1 coronene in toluene was filtered (pore size of 0.22 µm) before being mixed with a dilution of the suspension, giving effective concentrations of Fe2O3 from −1 −1 −9 −1 2.67 × 10 mgml to 4.17 × 10 mgml Fe2O3 in toluene. Typically, 250 µl coronene solution was added to 50 µm Fe2O3/toluene suspension was prepared in standard NMR tubes and placed under magnetic fields of 0 T, 1 T and 2 T over 24 hours. The temperature of the sample space was then lowered from 70 ◦C to 5 ◦C over 18 hours, after which the crystals were harvested.
2.2.10 Synthesis of metal carbonate nanoparticles
All metal carbonate nano-particles were synthesised using the co-precipitation method. Stock solutions of 0.05 M aqueous metal nitrates [X(NO3)2, X=Ba, Ca, Sr] and 0.025 M Na2CO3 were prepared. 100 ml of the former solution was placed in a sonicator and an overhead stirrer inserted. Under sonication and vigorous stirring 200 ml of the 0.025 Molar solution was added drop-wise over 8 h. The resultant milky solution was dried at 110 ◦C before being calcined at 450 ◦C. The resulting white powder was collected and analysed.
2.2.11 Synthesis of complex oxide pellets
Metal oxide and metal carbonate precursors for solid state reactions were mixed in a stoichio- metric ratio for target phases. These mixtures were ground by hand until the powder was homogeneous and particle size indeterminate by eye (≈ 30 min). Powders were loaded into a 7 mm pellet die, lubricated with a thin layer of oleic acid to avoid sticking and pressed at 2 t for a minimum of 15 min. The resultant pellets were gently broken (≈ 20-30 pieces) up to increase surface area for nanowire growth and calcined at the required temperature on a flat alumina tray.
2.2.12 Bio-templated nano-wire synthesis
Water soluble metal salts were dissolved in deionised water in a stoichiometric ratio in order to form the target phase to a concentration of 0.1 mol. A separate 1 % solution of sodium alginate (0.3 g in 30.0 ml) was prepared by adding the alginate to the water and mixing at 60 ◦C for 4 h until no solids are visible. When both solutions contained no visible solids, 3 ml of the metal salt solution was introduced to 30 ml of alginate solution before being vigorously shaken. The resulting gel was emptied into a Petri-dish, to maximise surface area, and left to dry in an oven at 60 ◦C overnight. Once dry and brittle, the solid was ground into a powder by hand and calcined in an alumina crucible.
27 CHAPTER 2. MATERIALS AND METHODS
2.2.13 Computational methods
2.2.13.1 Unit cell calculations
Initial computational calculations were performed by S. Crampin from the University of Bath, with density functional codes CASTEP[51] and VASP[52], using the Perdew-Burke-Enzerhof exchange correlation functional with semi-empirical dispersion corrections (DFT-D) to account for van der Waals interactions. CASTEP calculations (version 7.03) used in-built ultra-soft pseudopotentials for C and H atoms, a plane wave cutoff of 600 eV, and Monkhorst-Pack[53] k- point samplings of 3 × 4 × 2. Cell parameters and atomic coordinates were fully relaxed using the Broyden-Fletcher-Goldfarb-Shanno (BFGS) method, halting when residual forces fell below −1 1 meVÅ . Changing the cutoff energy from 450 eV to 600 eV caused structural parameters to change by < 0.001 nm or < 0.01°. Increasing k-point sampling to 6 × 8 × 4 changed energies by <1 meV. Robustness of results to choice of semi-empirical dispersion correction was assessed [54] [55] through the use of using the Grimme scheme with both default vdW radii (RH = 1.001, RC [56] = 1.452) and experimental (RH = 1.090, RC = 1.750), as previously employed by Fedorov et al. for γ-coronene. VASP calculations (version 5.3.3) use PAW potentials[57] with 500 eV energy cutoff, 3 × 4 × 2 k-point sampling and the Tkatchenko-Scheffler[58,59] dispersion correction. Unit cell geometry optimization was performed starting from experimental cell parameters and atom coordinates, both with and without symmetry constraint. Stability of optimized geometries was verified by re-optimizing after randomly displacing atoms by 0.05 Å in x, y and z. Further calculations to assess the suitability of a range of functionals at both the 621G and 631G levels (table A.2) were carried out using CRYSTAL14 software[60]. Experimental coordinates for γ- and β-coronene were used as a starting geometry for each calculation. Each calculation was given five attempts to converge with small changes in initial geometry before being labelled as did not converge (DNC). These data are summarised in A.2.
2.2.13.2 Cluster calculation
Cluster geometries were built with the computer program Gaussview (version 5.0.9)[61]. Calcula- tions were run using the software Gaussian[62] using parameters as indicated in the main text. For each number of molecules multiple calculations were run using different conformations of molecules and the optimised geometry in the text is representative of the final calculations.
2.3 Analysis
2.3.1 Single crystal X-ray diffraction
Intensity data for crystal structures were collected at various temperatures on either an Agilent SuperNova-E Dual diffractometer equipped with an Oxford Cryosystem, using copper Kα radiation (λ = 1.5418 Å) or a Bruker Kappa Apex II diffractometer, using molybdenum Kα
28 2.3. ANALYSIS
radiation (λ = 0.7136 Å). Data from the SuperNova were processed using the CrysAlisPro software (CrysAlisPro, Agilent Technologies, Version 1.171.37.35 (release August 2014 CrysAlis171 .NET; compiled on 13 Aug 2014). Data from the Apex II were processed using Bruker APEX2 v20 and Olex2 v1.2.7 software [63]. For all structures a symmetry-related (multi-scan) absorption correction was applied. Structure solution, followed by full-matrix least squares refinement was performed using the WINGXv2014.1 suite of programs throughout.
2.3.2 Powder X-ray diffraction
Due to small amounts of product, samples for powder X-ray diffraction were gently ground using a pestle and mortar before being pressed onto a silicon wafer on the surface of a low background sample holder using a glass slide, as to have a sample surface parallel to the wafer with a sample height not exceeding 0.2 mm. All powder patterns were collected using Cu-Kα (λ = 1.5418 Å) using a Bruker d8 Advance with PSD LynxEye Dectector. Low temperature pXRD patterns also used a Oxford Cryosystems pHeniX cryostat sample stage. Patterns were analysed using EVA software loaded with the ICDD PDF-2 2006 crystallography database.
2.3.2.1 Powder refinement
Powder fitting was used to extract both cell parameters and relative fractions in a multi-phase system using the open source software Profex version 3.10.2[64] a graphical user interface for the BGMN refinement engine. Experimental patterns were fitted against known patterns using the software and database mentioned earlier (section 2.3.2). Fitted known patterns were then used as structures for refinement. Refinement was considered complete when Rwp ≤ 1.5×Rexp.
2.3.3 Solution based UV-vis spectroscopy
Solution based UV-vis data was collected using a Perkin-Elmer Lambda 25 UV/Vis spectropho- tometer. Samples were prepared by pipetting 600 µl an undiluted solution saturated at 93 ◦C for 7 d into a low volume quartz cuvette with path length 1 cm. The cuvette was then inserted into the sample space after a background of pure toluene was recorded. Data was collected between 500 nm to 260 nm at a rate of 5 nms−1 with lamp and detector changes outside of the collection range. Subsequent dilutions were obtained by removing 200 µl of the previous solution before adding a further 200 µl of toluene. The concentration of the initial solution was determined retrospectively using evaporation and subsequent dilutions were extrapolated.
2.3.4 Solid optical absorption
Optical absorption spectra were recorded by E. Da Como at the University of Bath, with a ultraviolet-visible-near infrared spectrometer Agilent Cary 5000 measuring in transmission configuration. Single crystals of β- or γ-coronene were suspended in the sample beam path.
29 CHAPTER 2. MATERIALS AND METHODS
Unpolarized light was shone perpendicular to the a-b plane of the unit cells. High-quality crystals with flat surfaces were carefully selected to avoid light scattering effects. The beam size was narrowed to half the lateral size of crystals. All absorption spectra were recorded at room temperature in air.
2.3.5 Microscopy
2.3.5.1 Transmission electron microscopy
To prepare inorganic samples for transmission electron microscopy (TEM) analysis, a small amount of sample (< 500 µg) was transferred to an Eppendorf tube and 10 drops of bench ethanol was added using a Pasteur pipette. The tube was then sonicated for 30 s at which point the solution typically becomes slightly turbid. Using a pipette, an aliquot from just under the meniscus of the solution was withdrawn and a single drop, dropped onto a pre-coated copper grid, supported on a corrugated film surface. The grid was then covered and left to dry overnight. TEM images were collected with a JEOL TEM 2010 in image mode using a beam current of 110 µA and an accelerating voltage of 200 kV. Selected area electron diffraction (SAED) patterns were taken on the same microscope in diffraction mode, using a camera length of 15 cm.
2.3.5.2 Scanning electron microscopy
Samples for SEM analysis were prepared by adhering some sample to a sticky carbon pad mounted on a 7 mm aluminium stub. The mounted sample was then sputter-coated with a 15 nm layer of silver. SEM micrographs of the sputter-coated samples were obtained using either a JEOL IT300 SEM or a JEOL JSM 6330F high resolution SEM with a field emission gun. Images were acquired with a range of electron beam accelerating voltages and working distances.
2.3.5.3 Atomic force microscopy
Elastic moduli were determined using an Asylum Research MFP-3D Infinity AFM operating in AMFM viscoelastic imaging mode, using an AC160TS-R3 silicon tip (9§2 nm radius). Freshly cleaved mica was used as a calibration standard (measured at 178 GPa).
2.3.6 Differential scanning calorimetry
Melting point determination was done using a TA-Instruments Q100 DSC with temperature ramp of 1 ◦Cmin−1 between 35 ◦C to 460 ◦C with 2.2 mg of coronene hermetically sealed under a nitrogen atmosphere in an aluminium pan.
30 2.4. DESIGN AND CONSTRUCTION OF A 2.5 T PERMANENT MAGNET
2.4 Design and construction of a 2.5 T permanent magnet
In order to achieve magnetic fields as high as possible whilst still retaining a suitable crystal growth cavity, a 2.5 T static magnet was designed and built. This magnet has a 5 mm sample cavity with a homogeneous static field suitable for standard NMR tubes and 5 mm path length cuvettes as growth vessels. Full design and build instructions can be found in appendixB.
31
,teaiiyt oto oyopimi rai oeua crystals molecular organic in polymorphism control to ability the 1.4), (section earlier discussed As enrpre igetm n o otegoa cetfi omnt.Ti hpe describes chapter This community. scientific only global found, the be to can not as crystalline and far time the as single has in a polymorphs, changes reported new solidification, been create on to effect i.e. materials, an it molecular have While of can aggregates. is structure forces chiral solution magnetic or crystallizing crystals that the protein known of high-quality is convection of texturing growth controlled melt in where the example cases of for mainly in required, consists and and alloys explored and less metals much of is field magnetic external an of Introduction 3.1 S. by initiated were calculations cell unit VASP and CASTEP the Crampin spectra and absorption Como optical Da the referenced E. than been by other has collected myself myself by from collected used input was data minimal data Any only all Specifically, co-authors. with accordingly. data by analysed all publication and of the collected analysis of thesis, in behalf this on involved in and was myself I by herein. directly contained both text collected and figures all of preparation the 45–47 references - 2017 Reports Scientific Nature and a enrpre sn ierneo xenlparameters external of range wide a using reported been has l aesue nceto fti hpe eewitnadsbitdb yefinvolving myself by submitted and written were chapter this of creation in used arXiv) papers on All preprint (with 2016 Communications Nature in published are chapter this of Parts Ams vr usac a xs ntoo oepae rvddteexperimental the provided phases more or two in exist can substance every “Almost odtosaesuitable.” are conditions [46] . M AGNETICALLY ihl swl,1899 Ostwald, Wilhelm – I NDUCED 33 P LMRHS IN OLYMORPHISM . [65] hsfr rwhi h presence the in growth far, Thus . C
ORONENE C HAPTER 3 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE
polymorph selectivity in polyaromatic hydrocarbons (PAHs), using magnetic fields and attempts to explain the possible mechanisms involved. The ability to discover new phenomena and properties in materials depends on our ability to synthesize new structures. The discovery of new structures, however, should rely on variables that transcend the typical thermodynamic parameters of temperature and pressure. In the field of functional molecular materials, polymorphism, the presence of different crystal structures of the same molecular system, can be an opportunity to discover novel phenomena and tune properties [66,67]. For example, charge carrier mobility in organic semiconductors can be increased by crystallizing molecules under pressure, resulting in shorter intermolecular distances favourable for transport[68]. One successful tool at the scientist’s disposal to control crystal structure has been to perform crystallization in the presence of electric fields. This is a well- known process and is used extensively to prepare non-linear optical materials through electric poling. Crystal growth in the presence of an external magnetic field however, is much less explored, although it has been proven to be efficacious in the melt texturing of alloys [69] or in cases where controlled convection of the crystallizing solution is required, for example, in the growth of high-quality protein crystals[70], chiral aggregates [71] or in the alignment of liquid crystal and block copolymer arrays[72,73]. A magnetic field has even been demonstrated to be able to separate polymorphs of certain crystals post-synthetically[74] and in one case to preferentially nucleate a monoclinic form of terpyridine [75]. Although it is known that magnetic forces can have an effect on solidification and subsequent physical behaviour in crystals, their use to create previously unknown polymorphs in single crystals has never been reported.
3.1.1 Polyaromatic hydrocarbons
Polyaromatic hydrocarbons (PAHs) (also poly aromatic hydrocarbons and polycyclic hydrocarbons) are a a subset of hydrocarbons in which each molecule contains multiple aromatic rings. Not to be confused with polynuclear aromatic hydrocarbons (PNAs) which are a subset of PAHs containing exclusively fused aromatic rings, that is, adjacent rings that share two or more carbons. PAHs are continuously formed as a by-product of incomplete combustion of crude, carbonaceous fuel with sources both man-made (e.g the burning of coal or fractions of crude oil) and natural (e.g volcanic eruptions and forest fires). As the family of PAH molecules spans the smallest, two-ring member ‘Naphthalene’, to any size of hydrogen-terminated array or fused rings distinguishable from graphene the properties can vary wildly. As highly aromatic non-polar molecules, PAHs tend to be lipophilic and solid under standard conditions however the small two, three, four and five -ring members of the family can be volatile enough to exist in the gas phase and be marginally soluble in aqueous solution[76]. Therefore PAH molecules are both persistent and ubiquitous throughout our environment and thus are most thoroughly investigated from a toxicology perspective. The Environmental Protection Agency (EPA) in the United States, lists 16 priority PAHs when assessing environmental impact (table 3.1), all of which have six or fewer rings. Benzo[a]pyrene
34 3.1. INTRODUCTION
specifically, a five-ring PAH fraction of this incomplete combustion, which is present in high concentrations in cigarette smoke, is considered by medical research to have severe carcinogenic and tetragenic effects. As can be seen from table 3.1, the number and size of rings present in a molecule is not necessarily enough information to identify a PAH. For example, Benzo[a]anthracene and Pyrene, both having four, six-membered rings differ by two carbon atoms and two hydrogen atoms. Each PAH also has a unique set of properties apparently independent from ring number as evidenced by the two aforementioned molecules having an order of magnitude difference in regards to their solubility in water. From a more materials perspective, PAHs are still commonly researched due to their rigid planar structure, high stabilities and characteristic optical and electronic behaviour[75,77]. With few exceptions, PAHs consist exclusively of hydrogen-terminated sp2 hybridised carbon and deviate away from planarity for the following reasons; when the molecule contains ring sizes other than six-carbon, which forces parts of the molecule to become convex, as in corannulene or buckminsterfullerene or when the molecule contains rings bound by a single bond that can rotate, as in rubrene or p-terphenyl. Or when the molecule is large enough that steric interactions can contort the molecule away from planarity with a sufficiently low energy cost, as in Hexabenzo[a,d,g,j,m,p]coronene[78]. With this said, PAHs having multiple aromatic components, even non-planar examples will have electrons within a larger π-system that have unimpeded movement across sections of their host molecule. This gives each PAH a unique way in which it interacts with energy, other materials and other versions of itself.
3.1.2 Polyaromatic hydrocarbons as molecular crystals
Once refined to a sufficient level of purity, PAHs tend to form molecular crystals which allow this interesting electronic behaviour to propagate throughout the solid. These interesting properties observed in PAHs arise from the spatial arrangement of molecules within the crystal structure therefore necessarily determine the physical properties of the bulk crystalline solid[79]. Once above a certain size, PAH crystals are characteristically coloured due to their interactions with light being absorption in the visible region, from pentacene which presents as a deep purple colour, through reds and oranges[22] to the pale yellow of coronene[80], with a wide spectrum in between. The smaller PAHs are commonly found to have a pale yellow tint, but this is thought of as due to trace impurities[81]. This coloured nature of PAH crystals is indicative of band-gaps of interest to the fields of solar-farming[82] and plastic electronics [83]. Indeed, rubrene has been attributed the highest carrier mobility of any organic semi-conductor[84]. The high levels of anisotropy of PAH molecules lead to two main forms of molecular interaction within the crystal, π−π-stacking (C···C) and CH-π (C···H). In planar molecules, if one interaction builds faster during the growth of a crystal (C···C 6= C···H) the morphology of the crystal is affected, thus PAH crystals are commonly found as needles or plates.
35 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE
Name Formula Number Structure Molecular Solubility in of rings weight (g.mol−1) water (mg.cm−3)
Naphthalene C10H8 2 128.17 31
Acenaphthene C12H10 3 154.21 3.8
Acenaphthylene C12H8 3 152.20 16.1
Anthracene C14H10 3 178.23 0.045
Phenanthrene C14H10 3 178.23 1.1
Fluorene C13H10 3 166.22 1.9
Fluoranthene C16H10 4 202.26 0.26
† Benzo[a]anthracene C18H12 4 228.29 0.011
† Chrysene C18H12 4 228.29 0.0015
Pyrene C16H10 4 202.26 0.132
† Benzo[a]pyrene C20H12 5 252.32 0.0038
† Benzo[b]fluoranthene C20H12 5 252.32 0.0015
† Benzo[k]fluoranthene C20H12 5 252.32 0.0008
† Dibenz[a,h]anthracene C22H14 6 278.35 0.0005
Benzo[g,h,i]perylene C22H12 6 276.34 0.00026
† Indeno[1,2,3-cd]pyrene C22H12 6 276.34 0.062
† TABLE 3.1. The US-EPAs list of priority PAHs pollutants. denotes those classed as probable carcinogens to humans.
36 3.1. INTRODUCTION
3.1.3 PAH crystal packing
As molecular solids, most PAHs crystallize in four basic structure types according to well-defined geometric and energetic considerations[28]. These are the herringbone (HB) structure, the γ- herringbone (γ-HB) structure, the sandwich herringbone (SWHB) structure and β-herringbone (β-HB) structure (fig. 3.1). Comprehensive studies of PAHs have shown that the adoption of one of the four structure types depends ultimately on the relative strength of non-bonded C···C and C···H interactions[28,85,86]. The naming convention suggested by Desiraju and Gavezzoti, specifies using the nearest molecular neighbour angle and the unit cell short axis in which to assign the packing of a specific PAH to a named group.HB and SWHB structures containing mainly C···H (edge-face) interactions while the stability of the γ-HB and β-HB structures are dominated by C···C (face-face) interactions. It has also been observed that larger PAHs tend towards tighter packing with a smaller unit cell short axis[87].
(a) (b)
(c)
(d)
FIGURE 3.1. The four classes of crystal packing in PAHs. (a) is theHB structure, (b) the γ-HB structure, (c) the SWHB structure and (d) β-HB structure. Blue boxes show the unit cell as viewed parallel to the {010} planes, thus the short b-axis runs from the bottom to the top of the page. Red lines denote the angles measured to classify each structure.
37 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE
These packing motifs were classified in 1989 by Desiraju and Gavezzoti by collating the crystal structures of 32 well characterised PAHs listed in appendix A.3. Their work alludes to the ability to predict the probable packing family of any rigidly planar PAH based on the π-stacking. The four aforementioned packing families are easily identifiable when the nearest neighbour angle is plotted as a function of unit cell short axis as can be seen in figure 3.2.
FIGURE 3.2. Representation of the 32 PAH crystal structures (listed in appendix A.3), plotted short-axis vs. nearest-neighbour angle, adapted from Desiraju and Gavezzotti [1989]. Red, green, blue and black squares represent β-herringbone, γ-herringbone, herringbone (HB) and sandwich-herringbone (SHB) motifs, res- pectively. Dashed lines represent trends within a motif and the hashed grey bar indicates an implied region of unfavourable edge-face and face-face interactions, as suggested by Desiraju and Gavezzotti[28].
3.1.3.1 Polymorphism in PAHs
Polymorphism in PAHs has been demonstrated previously by using extremes of conditions (e.g. pressure[88], temperature[89]) or growing thin films on substrates, where variations in processing conditions direct the formation of different polymorphs[90,91]. However in single crystals at ambient pressure, polymorphism in PAHs is rare with perylene and pyrene being two notable cases exhibiting both herringbone and SHB polymorphs[92]. In terms of applications, single
38 3.1. INTRODUCTION
crystals of PAHs are to be desired however, as they have distinct physical advantages over thin-film versions of the same material. Most importantly, for conductive applications such as transistors, or optical applications in solar cells, single crystals typically have higher carrier mobilities than their thin-film analogues[93,94], which have to be defect free to achieve mobilities of the same order of magnitude.
3.1.4 Coronene
Coronene is a PAH composed of six aromatic rings arranged in a planar discoidal geometry
(fig. 3.4). The high molecular symmetry (D6h) and 24 electron π-system has made coronene an ideal model system for the study of graphene, due to it being large enough to display exotic electronic behaviour, but not so large that contortion becomes a complicating factor [95]. The crystal structure of coronene was first reported in 1944 by John Monteath Robertson and John
White and reported to have a space group P21/a with parameters a=16.10 Å, b=4.695 Å, c=10.15 Å, β=110.8°, γ-herringbone. 11 years later in 1955 a yellow hydrocarbon crystal was discovered in Zakarpattia Oblast in the Carpathian mountains, Ukraine[96] and named ‘Karpatite’ (also known as ‘Karpathite’, ‘Pendeltonite’ and ‘Carpathite’) which was later found to be naturally occurring coronene in a pure form shown in figure 3.3. Analysis of 13C isotopes in karpathite suggest that the organic source material for the mineral is fractionated when transported by hydrothermal vents near fault lines to crystallise in cavities, commonly between argillites and
FIGURE 3.3. Karpatite: coronene as found in natural geological formations (yellow shards). White crystals and red specks are quartz and cinnabar respectively. Sample from San Benito, California.
39 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE
diorite porphyry[97]. It is known to occur naturally in two further locations worldwide, San Benito County, in California, USA and Tamvatnei, far-east Russia and is known for it’s vivid blue fluorescence under UV-illumination. Containing no over-crowded peripheral hydrogens, the disc-shape, rigidity and planarity of coronene give it a point group of D6h as there is no conformational reason for the molecule to contort. Having a network of 24 sp2 bonded carbon atoms, coronene possesses a large 24-electron π-system which has put coronene under the ambiguous term ‘superaromatic’. This term has no agreed definition but occasionally surfaces when large PAHs are discussed. In 2003 it was defined by Aihara as “aromatic stabilisation due to cyclic motion of π-electrons along a macrocycle”[98]. Typically coronene is sourced through two commercial routes. The higher purity route (>99.5 %) is that of lab synthesis, a bulk process developed in order to synthesise high purity PAHs for studying biological activity and refining methods of spectroscopic detection. The second route tends to produce coronene of lesser purity (>95 %) and is refined from a build up inside industrial pipes within fossil fuel processing plants. As a product of ring-addition during incomplete organic fuel combustion, this deposit is up to 90 % coronene, and was first harvested from factories inside Germany, converting coal into liquid fuel during the oil embargo by the allies during world war II. This by-product can be refined to >95 % with little effort, impurities remaining tend to be planar molecules of a similar structure (e.g. ovalene)[99]. Synthetic routes to coronene have been reported steadily over the past 85 years with increasing efficiency and yield[100–104], the most recent of which achieves a 15 % to 20 % yield via benzannulation using a ruthenium catalyst[105].
A refining of the coronene unit cell in 1996 suggested a redefinition from P21/a to P21/n with a change in lattice parameters to a=10.122 Å, b=4.694 Å, c=15.718 Å and β=106.02°[106]. Although no real change in the cell volume, this cell defined by moving the lattice points is commonly used in literature. This unit cell has a short axis of 4.694 Å and ∠mol of 85.01° placing it comfortably
(a) (b)
FIGURE 3.4. The molecular structure of coronene. (a) Ball and stick representation where grey and white spheres represent carbon and hydrogen respectively. (b) Skeletal formula showing a single resonance structure.
40 3.1. INTRODUCTION
in the γ-herringbone region in reference to the naming convention mentioned in section 3.1.3. The simplicity of only sp2 carbon, terminal hydrogen and symmetry of the coronene structure has made it a model system for both theory and synthesis within in a few years of it being characterised and indexed. Described in 1948 as “...a new testing ground for the various theories of electronic structure of these aromatic systems.”[107] in reference to giving theoretical models empirical validation. Despite not being the smallest planar, high-symmetry hydrocarbon[108], the extended network of bonded carbon is large enough that electronic similarities to graphene and even graphite manifest yet small enough to be computationally feasible[109,110] and extrapolated to larger networks of two dimensioal carbon[95]. The use of coronene as the subject for theory and modelling regarding electronic structure[107], intermolecular interactions[111], current density[112] and π-stacking energy and bond length [113], spans over half a century and the synthetic control now available in providing high purity coronene[114] has even allowed experimental conformation of some of these effects[115,116]. More recently, it has been independently calculated that under an applied magnetic field, coronene has an overall diamagnetic susceptibility formed by the sum of a diatropic outer current located on the macrocycle and a paratropic inner current located on the central ring[117,118]. Using the same calculations, these counter-rotating currents are not observed in other similar PAHs, even when changed a small amount. As with other PAHs, coronene has a strong fluorescence which has also be thoroughly investigated[119–121] and has even been functionalised and used as a fluorescent marker[122–124]. Perhaps the most significant observed phenomena in coronene in the last decade, is that of its ability to superconduct when doped with potassium. Both coronene and the smaller PAH [125] picene were shown to be superconducting in the form KxPAH when 2.6 41 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE 3.2 Results and discussion 3.2.1 Growth of coronene under a magnetic field Coronene crystals grown from a supersaturated solution, as mentioned in section 2.2.5, were grown concurrently under magnetic fields of 0 T and 1 T with a 4 day cooling ramp from 93 ◦C to 50 ◦C to promote growth of large crystals. Although acicular crystals were harvested from both environments, their appearance differed significantly. Crystals grown under no magnetic field were 2 mm to 5 mm and pale yellow, as would be expected from literature. Structure solution using single crystal X-ray diffraction (scXRD) of these crystals indicated that the structure is the conventional γ polymorph; a =10.02 Å, b =4.67 Å, c =15.60 Å , β =106.7°, Z=2, space group P21/n, with an inter-planar distance (πd) of 3.43 Å, which is consistent with parallel π-stacking (fig. 3.7)[28,106]. In the presence of an external magnetic field of 1 T however, significantly longer (typically ≈2.5 cm) crystals were grown from, the same supersaturated solution and exhibit a different colour to normal coronene crystals (fig. 3.5). ScXRD indicates that the unit cell of the larger, orange crystals differ in a number of significant ways. While still monoclinic with space group P21/n, the 1 T crystals most noticeably, have a greatly shortened b-axis of 3.84 Å and monoclinic angle β of 96.24°. A full list of unit cell parameters are seen in table 3.2. Additional structural information was collected at a variety of temperatures and can be found in table A.1. FIGURE 3.5. Optical images of coronene crystals grown under 0 T and 1 T under daylight. Grid squares are 5 mm2 42 3.2. RESULTS AND DISCUSSION (a) (b) FIGURE 3.6. SEM micrographs of typical (a) β and (b) γ morphologies. Blue, green and red arrows indicate the a, b and c growth directions respectively. Literature[106] 0 Tesla 0.2 Tesla 0.5 Tesla 0.8 Tesla 1 Tesla (no field) Crystal Monoclinic Monoclinic Monoclinic Monoclinic Monoclinic Monoclinic system Herringbone γ γ γ γ β γ designation[28] Collection 200 200 200 200 200 293 temperature (K) a (Å) 10.0226(10) 10.0271(5) 10.0226(10) 10.0371(8) 10.3856(5) 10.122 b (Å) 4.6718(4) 4.6731(2) 4.6718(4) 4.7131(2) 3.7405(10) 4.69 c (Å) 15.6035(16) 15.5742(8) 15.5560(16) 15.6742(5) 17.2139(3) 15.72 β (◦) 106.395(11) 106.433(5) 106.305(9) 106.433(5) 96.842(10) 106.02 Volume (Å3) 700.904 699.960 699.093 711.193 664.017 717.80 Space group P21/n P21/n P21/n P21/n P21/n P21/n Z 2 2 2 2 2 2 Z0 4 4 4 4 4 4 πd (Å) 3.445 3.430 3.429 3.478 3.467 3.455 ◦ ∠mol ( ) 95.86 95.45 95.43 94.54 49.71 95.30 doff (Å) 3.18 3.17 3.17 3.17 1.59 3.17 TABLE 3.2. Unit cell parameters of coronene grown under various magnetic field strengths. πd is the stacking distance between coronene molecules, ∠mol is the nearest neighbour angle between molecules and doff is the distance between the molecule centres in the xy direction. 43 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE By comparing the two crystal forms from table 3.2 and displayed in figure 3.7, it can be seen that the short axis, b, is substantially decreased in length for coronene crystallised in the magnetic field compared with the that crystallised without field. This new polymorph also has a significantly smaller nearest neighbour herringbone angle of 49.71° compared with 95.86° in standard γ-coronene. Therefore the identification of the new crystal polymorph as a β structure can be made by reference to the plot of the inter-planar angle of nearest neighbour molecules versus the unit cell short axis, mentioned earlier in section 3.1.3 [28] (fig. 3.8). Crystal growth performed at weaker field strengths of 0.2 T, 0.5 T and 0.8 T resulted in crystals of the γ polymorph, suggesting that 1 T is close to a threshold for energetic selection between the two forms. β-coronene, as a previously unreported polymorph of coronene, the crystallographic data has been lodged with the Cambridge Crystallographic Data Centre (deposition number CCDC 1409823). (b) (a) (c) (d) FIGURE 3.7. Representation of the (a) γ and (b) β polymorphs of coronene. Differing perspectives of both unit cells (blue boxes) viewed slightly offset from along the a-axis (right) and along the b-axis (left). The relative shift of the molecules along the stacks are shown for (c) γ and (d) β. Red green and black arrows indicate the direction of the a-, b- and c-axis respectively. Figure adapted from reference 46. 44 3.2. RESULTS AND DISCUSSION FIGURE 3.8. Grouping of PAHs into structure types as with figure 3.2, updated to show the positions of the two polymorphs of coronene. The red circles show the positions of where both γ-(green centre) and β-(orange centre) coronene polymorphs fall on the graph. From this, it can be seen that the new polymorph sits squarely with other PAHs identified as having the β herringbone structure. The larger amount of π-overlap implies that this structure benefits from a higher amount of stabilisation from C···C, face-face, interactions whilst seeing a significant decrease in the use of C···H, edge-face interactions as a stabilising influence. The difference in size and colour suggest significant differences in both the growth mechanism and the inherent electronic structure. 45 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE 3.3 Characterisation 3.3.1 Structural stability As described in section 1.4.2, for multiple polymorphs to exist concurrently for a given crystal at a given temperature and pressure, one must be the most thermodynamically stable. That is, with two polymorphs of coronene being seemingly stable at room temperature and pressure, one of them must be meta-stable even with a negligible difference in lattice energy. With regards to the physical properties of molecular crystals, this would normally manifest as a difference in melting point and hardness[128]. In order to fully understand the structural stability of the new β polymorph of coronene relative to the γ form, both computational and empirical methods can be employed. 3.3.1.1 Calculated stability In an attempt to understand whether the new crystal structure is stable, meta-stable or perhaps a frustrated unit cell formed as domains in a disordered material, DFT-D calculations were carried out. The geometries of both the β and γ polymorphs were tabulated as a set of coordinates in separate text files (fig. A.1). The coordinate files were used as starting geometries for a number of calculation sets and allowed to optimise their geometry with no thermal factors (at 0 K). The output parameters for each calculation set were compared to experimental values to assess the accuracy of each set of calculation parameters, namely the basis set and functional. After which, the coordinates were displaced randomly by 0.05 Å and allowed to re-optimise to verify the stability of the output geometry. Table 3.3 shows a summary of some output parameters using Cambridge serial total energy package (CASTEP)[51] and Vienna ab initio simulation package (VASP) [52], both utilising plane wave basis sets, as well as various Van der Waals methods[55,57,58] and values for Van der Waals radii[56]. Table 3.3 shows that, regardless of the functional used, at 0 K the β polymorph is calculated to be at lower energy. This implies that, at 0 K, the β polymorph is the more stable of the two by 3.4 kJmol−1 to 4.3 kJmol−1 based on lattice energy differences. For completeness, a more exhaustive selection of functionals can be found in table A.2. Each optimisation was carried out at both the 621-G and 631-G levels, combinations that did not converge are labelled ‘DNC’. In this table it can be seen that the calculations returning the closest fitting parameters is PBE/631G returning only a −0.01 % and −3.06 % difference in unit cell volume for γ and β, respectively. The lattice energy difference in these calculations drops to 0.1547 kJmol−1, which is only slightly −1 −1 higher than kBT between 0 K to 10 K (0 kJmol to 0.0831 kJmol ). Although this gives no information as to the energy barrier required to alternate between the polymorphs, it suggests that β may not be seen exclusively at cryogenic temperatures but both forms may be present if coronene is enantiotropic. 46 3.3. CHARACTERISATION Eγ-Eβ Code vdW form a (Å) b (Å) c (Å) β (◦) d (Å) ∠ (◦) V (Å3) mol (kJ mol−1) γ 9.779 4.548 15.422 107.26 3.29 87.4 655 CASTEP ♣ 4.3 β 10.254 3.672 16.851 95.76 3.35 48.5 631 γ 9.924 4.665 15.915 105.53 3.52 82.1 710 CASTEP ♠ 3.8 β 10.336 3.871 17.264 95.04 3.58 44.9 688 γ 9.954 4.566 15.443 106.57 3.35 85.6 673 VASP 3.4 β 10.319 3.714 17.253 95.86 3.39 48.1 658 ♣ Grimme (2006) [55] ♠ Grimme (2006) with experimental van der Waal radii as suggested by Fedorov et al [55,56] Tkatchenko-Scheffler [57,58] TABLE 3.3. Constructed from DFT-D calculations using codes and dispersion corrections (vdW) as indicated. Lengths a, b, c are the lattice parameters and β the monoclinic angle; d is the interplanar distance between parallel coronene molecules. The columns labelled ∠mol and V report the herringbone angle between adjacent coronene molecules and the unit cell volume, respectively. The final column reports the lattice energy difference between the two structures; positive indicates β is the more stable. 3.3.1.2 Low temperature structural analysis Confirmation that β-coronene is more stable at low temperatures comes from powder XRD of polycrystalline γ-coronene recorded through a temperature cycle from room temperature to 12 K. On cooling through a temperature of 150 K, three new peaks are shown to emerge which cannot be indexed to the γ polymorph (JCPDS card number 12-1611)[45,47]. Figure 3.9 shows how the pXRD pattern, and thus crystal structure change as a function of temperature in γ-coronene. The top of the figure shows a single powder pattern from γ-coronene for clarity. In the box below, the peak height in counts is displayed in grey-scale (darker and lighter indicating a higher and lower number of counts respectively), and degrees 2θ from 8° to 32° is displayed from left to right. Temperature is shown decreasing from 300 K to 12 K and then warming back to 300 K (vertical axis). Extra reflections can be seen appearing between the (101) and (002) and the (301) and (113) reflections of γ-coronene as the temperature drops below 150 K. When the β-coronene unit cell is used to calculate a pXRD pattern, the identity of these new reflections become clear (fig. 3.10). As the β phase emerges from the baseline, assuming sufficient randomisation of crystal orientation, the most intense reflections would be the first expected to emerge. As seen in figure 3.10, it would be the (101) and (012) reflections of the β phase. In this case however, these reflections would be eclipsed by the, already present, (101) and (112) reflections of the γ phase at 2θ = 9.72° and 25.66°, respectively. Therefore, the next set of reflections expected to emerge would be those that have significant intensity but are located at 47 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE FIGURE 3.9. The top is an indexed, γ-coronene powder pattern to indicate peak positions. Temperature is decreased from 300 K to 12 K and then warmed back to 300 K (top to bottom). Green arrows indicate the emergence of the additional reflections due to β-coronene. Asterisks indicate the temperature at which the β phase becomes detectable (blue) and drops back into the noise (red). This data is adapted from that presented in reference 129 and used with permission. values of 2θ that are not shared by the γ pattern. These correspond to the (002), (101) and (112); (2θ = 10.55°, 10.67° and 27.72°, respectively) reflections of the new polymorph and sit precisely where the emergent peaks appear on the 12 K pattern. Thus β- and γ-coronene are enantiotropic polymorphs, with a critical temperature between 100 K and 150 K. The first appearance of reflections due to β-coronene; (002) at 100 K, (101) at 100 K and (112) 143 K during cooling, are marked on figure 3.9 with green arrows and the point at which they are no longer detectable, during warming, of (002) at 234 K, (101) at 234 K and (112) at 254 K marked with red asterisks. The blue asterisk denotes the highest temperature at which the β- coronene phase is detectable. At the lowest temperature (12.4 K) they have the relative intensity values (002) = 2.80 %, (101) = 2.39 % and (112) = 6.89 %. Unit cell expansion and contraction 48 3.3. CHARACTERISATION (a) (b) FIGURE 3.10. Comparison of coronene pXRD patterns. (a) shows calculated pXRD patterns for γ and β polymorphs as red and black lines respectively and (b) shows experimental patterns of polycrystalline coronene at 300 K and 12 K as green and orange lines, respectively. 49 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE appears to happen anisotropically and can be seen calculated in figure A.3. As temperature decreases, the γ unit cell contracts in a fairly linear fashion with the c-axis reaching a minimum at 74 K followed by the a- and b-axes reaching a minimum at 51 K. Upon warming, expansion also appears fairly linear, starting almost immediately on heating with no detectable thermal lag. Analysis of the unit cell over this temperature range shows an obvious hysteresis of the a, b, c and β parameters (fig. A.3). Interestingly, the resulting unit cell volume, on warming, is observed to actually exceed that of the initial room temperature value above 250 K before contracting back to the original value at 300 K. This behaviour is likely to be a direct result of the crystal structure accommodating the interchange from γ to β but will require further investigation. The β-coronene polymorph (approx. 4 % of total crystal volume at 12 K, calculated via multi- phase Rietveld analysis) can be seen forming as a result of cooling, with the emergence of three peaks at 2θ = 10.46°, 10.69° and 27.78° at low temperature, which represent the (002), (101) and (112) reflections of β-coronene, respectively. 3.3.1.3 Size Specific cryogenic phase behaviour The slow emergence of β as the temperature drops, raises the question as to why the enantiotropic transition in coronene appears to occur over a large range of temperatures as opposed to a seemingly small range as seen in systems like pyrene[92] and DMSO[130]. Initially, twice sublimed coronene crystals were cooled at different rates to the 12 K floor and the percentage of β was recorded as a function of time. Refinement of these pXRD patterns collected with cooling rates of 0.1 Kmin−1 and 2.4 Kmin−1 showed β fractions of 4.2 % and 4.6 %, respectively. The maximum amount of β being comparable regardless of cooling rate is suggestive of another variable controlling the volume fraction and temperature at which crystallites of coronene transition from one polymorph to he other. As the initial low-temperature structural measurements were taken using gently crushed needles directly from the sublimation apparatus, it is inevitable that a range of crystallite sizes would be present and can be seen in figure 3.11 (b). Thus if size is a factor in at which temperature coronene undergoes a phase transition, an emergence over a temperature range would in fact be expected. Numerous attempts were made to collect scXRD data at sub-liquid nitrogen temperatures but each crystal measured was seen to exhibit some small changes before shattering. Therefore it could in fact be just crystals above or below a certain size that have the requisite stability at a given temperature to resist the transition. In order to address this, three size ranges, needle diameters 12.47 µm (lnσsd = 0.689), (c) 4.83 µm (lnσsd = 0.542) and (d)2.04 µm (lnσsd = 0.363) of lengths 100 µm to 10 µm, 30 µm to 5 µm and 8 µm to 1 µm, coronene were prepared as mentioned in section 2.1.2.1 and measured at 12 K with a cooling rate of 2 Kmin−1 (fig. 3.11). Multi-phase refinement was carried out on each 12 K pattern in order to assess the percentage of each phase present. As can clearly be seen from these data, a smaller crystallite size has a significant effect on the amount of β-coronene formed at low temperatures. Multi-phase refinement suggests volume 50 3.3. CHARACTERISATION (a) (b) (c) (d) FIGURE 3.11. (a) pXRD patterns showing the size specific emergence of β-coronene at cryogenic temperatures and and (b) - (d) SEM micrographs displaying the sample size distributions. Black, red and blue patterns were collected at 12 K are of coronene needles with width means based on a log-normal distribution of (b) 12.47 µm (lnσsd = 0.689), (c) 4.83 µm (lnσsd = 0.542) and (d)2.04 µm (lnσsd = 0.363). Image boarders indicate the relevant pXRD pattern. Emergent reflections of the β polymorph are highlighted by green arrows. 51 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE fractions of 4 %, 18 % and 50 % for crystallite sizes of 100 µm to 10 µm, 30 µm to 5 µm and 8 µm to 1 µm respectively. The increase in β volume fraction at 12 K as a function of crystallite size is unambiguous, confirming that at least down to 1 µm, smaller crystals are more likely to change polymorph. In fact size-specific polymorphism has been reported before in 1:1 cocrystals of theophylline and benzamide[131]. In this example, the polymorph that is most stable in the bulk crystal, has unstable surfaces when compared to the polymorph that is meta-stable in the bulk. Thus, once the surface becomes a significant percentage of the crystal mass (i.e. at the nano-scale), the former polymorph becomes meta-stable and the latter is preferentially formed. In the case of coronene, this would suggest that the surfaces of coronene are less stable when in the γ form compared to that of the β form. This is seemingly intuitive when considering the long-edges of both polymorph needles are made up of C···H interactions and that γ coronene gains more of its total stability from this interaction. The lengthening of C···H bond length in β coronene relative to γ supports this hypothesis. Of course this does conversely imply that in the bulk, at 300 K, the C···H interactions dominate crystal stability making γ the most thermodynamically stable form. 3.3.1.4 Differential scanning calorimetry In general, in an enantiotropic pair of polymorphs, the more intrinsically stable member will have the greater specific enthalpy of fusion (fig. 1.5). As the differential scanning calorimetry (DSC) experiment to determine melting point (fig. 3.12) is conducted at far higher temperatures than 150 K, it is the γ-coronene polymorph which will be the more stable at these temperatures and therefore be expected to have the greater specific enthalpy of fusion. From figure 3.12, the values for the heat of fusion for each of the polymorphs were calculated as 20.78 kJ mol−1 and 25.26 kJ mol−1 for β- and γ-coronene, respectively, confirming that these are indeed enantiotropic polymorphs. The new β polymorph structure is notable for the large change in the herringbone angle. We can gain insights into the relative stability of the β-coronene structure when compared with the γ polymorph, by consideration of these changes in the CH···π hydrogen bonding motif. It is known that the stronger the hydrogen bond, the stronger the trend for linearity[87,132] depending on the strength of the proton donor. From scXRD, we can see that the CH···π angle in γ-coronene of 95.86° suggests a strong, almost linear CH···π hydrogen bond, whereas in β-coronene, this angle becomes 49.71° (viz. fig.3.7). This change in angle would suggest a weakening of the hydrogen bonding in β-coronene, which is concomitant with the change in estimated CH···π hydrogen bond length[133] that increases from 2.5 Å in the γ polymorph to 3.0 Å in the β-polymorph. This weakening of the CH···π hydrogen bonding should therefore manifest itself as a diminution of the physicomechanical properties of β-coronene. To confirm this, the melting point (TM) and elastic modulus (E) the modal value measured on the (1011 and 1011 crystal faces)) of both 52 3.3. CHARACTERISATION (a) (b) (c) FIGURE 3.12. Physical characterisations of coronene polymorphs. Isolation of the melting points (a), enthalpy of fusion (B) of γ- and β-coronene and (c) Elastic moduli measured via AFM measured on the 1011 and 1011 crystal faces crystal faces. On all graphs, γ, β and mica are green, orange and blue, respectively. ◦ polymorphs and find that in γ-coronene, TM = 436.14 § 0.01 C and E = 227 GPa, whereas for ◦ β-coronene TM = 435.48 § 0.01 C and E = 92 GPa. These data suggest weaker intermolecular forces in β-coronene at room temperature. 3.3.2 Optical behaviour Figure 3.13 shows optical images of the two different polymorphs taken under UV irradiation (λ = 365 nm). The remarkably different colours of the fluorescence of γ-coronene and β-coronene 53 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE FIGURE 3.13. Optical images of coronene crystals grown under 0 T and 1 T under UV illumination. crystals is further confirmation that the molecular packing in the crystal has been transformed, leading to altered electronic behaviour. To quantify this, the absorption spectrum of two single crystals of the two polymorphs was measured by shining unpolarised light perpendicular to the a- b plane of the crystals, that is, light propagation parallel to [001]. As shown in figure 3.14 (green plot), the γ-coronene single crystal is characterised by a first absorption resonance at 468 nm assigned to the free exciton in coronene, in agreement with previous studies[134]. The β-coronene spectrum is by stark contrast almost featureless, with an absorption onset at 780 nm and a maximum at ≈ 500 nm (fig. 3.14, orange plot). This is a major change in the optical properties of this material in the γ form. From these spectra, it can be seen clearly that β-coronene is a more strongly absorbing polymorph than γ-coronene. Organic materials tend to be sensitive to a particular range of wavelengths, which can be seen in the features of the γ-coronene absorption spectrum and the lack of absorption of wavelengths > 583 nm. In β-coronene, however, the crystal strongly absorbs over a wide band of radiation, from the UV into the near-IR (320 nm to 847 nm). DFT-D calculations correctly describe the existence of the new stable polymorph and indicate that no appreciable difference in the indirect bandgap (fig. 4 b-c from reference 46) exists between the two polymorphs, suggesting that the large shift in the light absorption onset is instead related to a change in the fundamental photoexcitations in the two structures[135,136]. In β-coronene it is difficult to identify sharp resonances from Frenkel excitons as observed for the γ polymorph. Instead, the structureless absorption band of the β polymorph originates from charge transfer excitons that are favoured by the larger overlap between the molecular π orbitals. Moreover, the transition dipole moment of a charge transfer exciton is likely to be oriented parallel to the a-b plane, which would ensure optimal coupling with the electromagnetic radiation in the experiment. 54 3.4. IMPACT UPON PREVIOUS RESEARCH FIGURE 3.14. Normalised absorbance of crystalline β- (orange) and γ- (green) coronene. 3.4 Impact upon previous research The discoidal nature of coronene coupled with a large π-system, make the molecule exquisitely susceptible to counter-rotating ring currents under an applied magnetic field[117]. Evidence of this phenomenon can be seen with an abnormally large chemical shift of the deshielded peripheral hydrogens, during 1H-NMR spectroscopic measurements[137]. This gives the molecule a high magnetic anisotropy under an applied field which allows SQUID magnetometry to elucidate the difference in magnetic susceptibility of the two polymorphs as a function of the molecular rearrangement. Coronene has been a well researched system since its discovery. Throughout this research, various low temperature studies have reported unexpected phenomena at temperatures below 150 K. Through quantification of the temperature dependant structure of coronene, namely that the β polymorph emerges at temperatures below 100 K, though novel to this work, can be used retroactively to help rationalise this previously reported anomalous behaviour for low temperature coronene. I have previously collected superconducting quantum interference device (SQUID) data for polycrystalline coronene between 300 K to 12 K and identified a hyeteresis in the diamagnetic susceptibility between 300 K to 150 K[129]. The bulk magnetic susceptibility of the sample was measured as a function of temperature and therefore of crystal structure. Upon increasing 55 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE the temperature after applying a field, the sample became more diamagnetic and follows −6 −1 −3 a curve that begins to stabilise at χv = −1.82 × 10 emuOe cm . A sharp increase in diamagnetism is observed at 212 K until 267 K where it remains mostly constant until 286 K, −6 −1 −3 before reducing to χv = −1.88 × 10 emuOe cm at 300 K. With the field still applied and the temperature decreasing, there is no change in susceptibility until 164 K where the signal decreases in diamagnetism until it intercepts the zero field cooled (ZFC) line at χv = −1.80 × 10−6 emuOe−1 cm−3 at 98 K. The magnetic response of coronene at applied fields of H0 = 0.01 kOe, 10 kOe, 20 kOe and 50 kOe, shows that the magnetisation of coronene scales linearly with the applied field. When comparing these results, it is clear that the crystallographic structural hysteresis has strikingly similar features to the thermal hysteresis observed in the SQUID measurements, and at similar temperatures. The peaks appearing in the X-ray pattern at 150 K upon cooling (marked with a blue asterisk), clearly coincide with a marked decrease in diamagnetic susceptibility as seen in the magnetometry results. Upon warming, there are three points of note, 212 K, 267 K and 286 K. As the X-ray peaks begin to vanish at 212 K, SQUID magnetometry shows an increase in diamagnetic susceptibility until 267 K where susceptibility stabilises as the X-ray peaks diminish to below the level of noise in the pattern. Interestingly, the diamagnetic susceptibility is observed to surpass that of the initial cooling value significantly. When compared to the structural data, particularly the b-axis, (shown in appendix A.3) this larger susceptibility coincides with the unexpected increase in the unit cell parameters upon warming, suggesting that a larger unit cell results in a larger diamagnetic susceptibility. The third point at ≈286 K shows a sharp decrease in diamagnetic susceptibility before reaching the maximum temperature of 300 K. Upon re-cooling, the susceptibility does not decrease but remains relatively constant before an upturn towards the paramagnetic region at approximately 150 K. Figure 3.15 (a) shows the magnetometry data with the positions of relevant peak activity marked with asterisks and the direction of thermal cycling marked with arrows. It would appear that the emergence and waning of the β phase, sign-posted by the emerging reflections in the X-ray experiment, is having a direct effect on the magnetic behaviour of coronene. As there is no applied magnetic field present for the pXRD measurements, this shows unambiguously that at higher temperatures, the magnetic susceptibility thermal hysteresis is due to the structural transition from the γ to the β polymorph. As the hysteresis in both the SQUID data and the powder diffraction correlate, it is reasonable to postulate that the new polymorph is having a discernible effect on how the coronene molecules respond to an applied magnetic field. When crystallographic reflections due to β-coronene are present, the overall bulk diamagnetic susceptibility is diminished. As this magnetic response scales with applied field to at least 50 kOe, this implies that the new phase is enhancing the paratropic ring current. In PAHs, a deviation from planarity in individual molecules reduces aromaticity[138], so the reduction in thermal undulation at low temperature would cause an increase in ring currents under a constant field. It is therefore reasonable to assume that the minor reorganisation in 56 3.4. IMPACT UPON PREVIOUS RESEARCH (a) (b) FIGURE 3.15. Comparison of magnetic and structural data from coronene at low temperatures. (a) Shows the magnetic susceptibility of 22 mg coronene as a function of temperature. (b) Displays the fitting of the structural powder data for coronene to the γ-coronene pattern. The blue and red asterisks represent the highest detectable temperature in which β-coronene reflections are visible on cooling and warming respectively. Adapted from reference[47] crystal structure involving a unit cell contraction and deviation from planarity in the formation of β-coronene is the cause of this magnetic behaviour. In 1983, Byrn et al. reported solid-state NMR on crystalline coronene at temperatures as low 13 as 5 K and observed an unexpected deviation of the δ33 tensor component of the C signal from the axis perpendicular to the plane of the molecule[139]. This 13.4° offset at 5 K, not observed at room temperature, was suggested by the authors to be due to an unexpected axis of freedom for the molecules in the solid state at low temperatures. However, reassessment of this data with knowledge of the β polymorph of coronene reveals that a partial transformation of the powder into β-coronene at low temperatures would necessarily mean that two phases would be detected via solid-state NMR[139]. It is therefore clear that as γ-coronene and β-coronene differ only in molecular overlap and nearest neighbour angle, a mixture of the two polymorphs in the solid state could indeed be calculated as an average 13.4° shift in δ33 tensor. Quantitatively, based on a 53° difference in the nearest neighbour angle in the new polymorph it can thus be resolved that of the total mass of coronene tested in this previous study, approximately 38 % assumed the β form. As mentioned in section 3.3.1.3 crystallite size is also a factor governing the proportion of crystals undergoing the phase transition. Higher percentages are observed in smaller crystallites which are likely to be found in fast-cooled (shattered) or milled mixtures of γ-coronene and β-coronene[140]. Regarding the optical response of coronene, section 3.3.2 reports that the way in which each 57 CHAPTER 3. MAGNETICALLY INDUCED POLYMORPHISM IN CORONENE polymorph interacts with light differs significantly [46]. It would therefore be likely that this difference would be observed in spectroscopic studies of coronene at low temperatures. The low temperature optical measurements of coronene has been reported numerous times and in all studies, anomalous behaviour was reported to be observed below 90 K. Two papers reported on the low temperature optical behaviour in 1994 (references 134 and 120) and made reference to a possible electronic or structural phase transition. These studies showed that changes in spectra from 298 K to 2 K indicated the emergence of an absorbance band (band C, fig. 2 from reference 141) at 2 K that is a third the intensity of the expected 0-phonon band for the γ polymorph. This band sits to the lower energy side, which would be expected from a greater overlap of the π- orbitals; precisely what is seen in the crystal structure of β-coronene. In addition to this, the increase in charge transfer excitons would lead to a drop in those self-trapped, especially within 1 1 the Lb (separating from La) band which is parallel to the b-axis, exhibiting the greatest overlap. Band C’s intensity at 2 K also suggests >30 % conversion into the β polymorph at the lowest temperature. Perhaps the most important sphere in which knowledge of the precise physical behaviour of coronene is essential, is in the field of computational chemistry. As stated earlier, coronene is a key molecule used for understanding behaviour and prediction of extended sp2 carbon systems including molecular crystals and graphene. In preparation of an analytical or predictive simulation using quantum chemical calculations, it is necessary to bench-mark all input parameters against experimental data. This bench-marking optimises the balance between quantum chemical calculations and the use of approximations, especially necessary when examining larger systems of periodic boundary conditions. The reason this is so critical is that the correct correlation of input to experimental data minimises computational load whilst validating further calculation within the same system. As has been shown however, all semi-empirical calculations (and all subsequent research conducted using these methods), parameterised to adopt the assumed γ-coronene structure at cryogenic temperatures will likely be erroneous, due to the appearance of the β-coronene polymorph. More specifically, as has been shown crystallographically, the stacked molecular overlap in β-coronene (a geometry which most dispersion-corrected methods agree to be the most thermodynamically favourable) leads to a weakening of the C-H···π interaction, thus it follows that calculations on orbital interactions within crystalline coronene must take into account the appearance of the new polymorph in order to be considered to be accurate. 3.5 Summary Through application of a permanent magnetic field higher than 0.8 T, it is demonstrated that the well-studied PAH coronene can be grown as a new “β” polymorph from solution. This polymorph has been confirmed to have a range of physical properties have been significantly altered from the 58 3.5. SUMMARY ubiquitous γ form. The optical properties of the β-coronene crystal have been transformed to such an extent that it absorbs a much wider range of wavelengths from the UV into the near-IR. This new single crystal may therefore be of great interest in organic optoelectronics and photovoltaics. The β polymorph has also been shown to be less hard with a slightly depressed melting point indicative of a metastable polymorph. Density functional theory (DFT) calculations investigating energetic and geometrical differences between the two polymorphs have provided evidence that β-coronene is more stable at 0 K with the slightly lower lattice energy. This is also evidenced by the spontaneous appearance of β-coronene when cooled below 150 K and disappearance of, when warming. Which in turn implies that β-coronene is metastable at room temperature and pressure. As the electronic behaviour of the material is intimately linked to the crystal structure as it changes from one polymorph to another under thermal cycling, under which a partial polymorph transition has been show to occur, it can and has been inadvertently observed in data collected involving low temperature coronene. In addition, it can be shown that the previously unaccountable behaviour of coronene in spectroscopic studies (absorption, fluorescence and luminescence), along with 13C solid-state NMR spectroscopy and computational modelling is due to the appearance of the β-coronene polymorph at temperatures below 110 K and the concomitant change in physical properties below that temperature. From the work presented here, it is essential that in the future, the physical properties of the new β-coronene polymorph are taken into account in order to have confidence that low-temperature physical characterisations and optimised geometry calculations are correct. 59 h tat ntecrnn-oun ytm hscatratmt oucvrseic bu how about specifics uncover to attempts chapter This system. coronene-toluene the in acts it why st h usino h napidmgei edo > of field magnetic applied an why of question the to As h antcfil safcigtesse nodrt rdc ieyscesu addtst which to candidates successful likely predict applied. to be order can in technique system this the affecting is field of magnetic fundamentals the the understand to important is it systems, crystal molecular all to translatable counteract to size given a of clusters clustering, for pre required trimers a field or at dimers magnetic saturation of maximum formation number freedom, interactions, daunting conformational solvent-molecule a of temperature, of (degrees given summation system a each a to be specific could variables field of under actions systems each technique, same growth. crystal mediated magnetic-field for impractical them making growth crystal for required higher a with molecules diamagnetic Planar anisotropy. high magnetic relatively the the to has linked coronene ( intimately that noting be susceptibility worth to is likely It is molecule. it the strength, of field susceptibility this at molecules other in not Introduction 4.1 − 346 nei a la htmgei-edmdae oyopsslciiywsnteasily not was selectivity polymorphs mediated magnetic-field that clear was it Once the using when systems other in observed phenomena interesting of variety the on Based . 0 × So n okaon,i’ l sonig ae,fie i n dirt...” and air fire, Water, astounding. all it’s around, look and “Stop 10 − 6 emuOe χ f= of ) − T 1 − cm HE 243 − [143] 3 × M 10 CAIMOF ECHANISM − r aeadtn ob noul oayual concentration usable any to insoluble be to tend and rare are ) 6 emumol χ − [117,142] 1 o ntnehexabenzo[ instance for , 61 M o yrcro,wt ihmagnetic high a with hydrocarbon, a for 0 AGNETIC . T 8 shvn hsefc ncrnn and coronene in effect this having is k B T isle a otn,etc.). content, gas dissolved , C bc,ef,kl,no,qr RYSTAL C,2009 ICP, – crnn ( ]coronene C HAPTER G ROWTH 4 χ = CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH 4.2 Magnetic purity concerns Previous work reporting interesting magnetic behaviour of both carbon allotropes[144] and hydrocarbons[125] has been either retracted[145] or steeped in controversy[146,147], mainly due to magnetic impurities. To try and decouple any effect these impurities might have on the crystal systems is therefore essential. Initially, to evaluate for the presence of any magnetically active impurities in the crystalline starting material, twice sublimed coronene was digested in trace metal grade nitric acid (Sigma 225711) before being analysed via ICP-AES. The purified coronene used for the growth of these crystals, show the level of magnetically active impurities such as cobalt, iron and nickel are at the parts per billion level (Co = 0.138 ppm 0.98 ppb; Fe = 33.81 § 0.56 ppb; and Ni = 13.46 § 0.65 ppb), as no reports of doping systems at these concentrations to facilitate polymorph selection are forthcoming in the literature, it is unlikely that these metals play a role. In the wider system however, impurities become increasingly difficult to monitor in situ. Attempting to simulate any possible magnetic impurities that may have inadvertently found their way into the solution, ferromagnetic nanoparticles were introduced to coronene dissolved in toluene at concentrations far beyond that possible through atmospheric delivery alone[148], in order assess its effect on crystal growth. Nanoparticles of Fe2O3 were synthesised as mentioned in section 2.2.8 and dispersed in toluene via sonication at a concentration of 32 mg in 20 ml. Performing six sequential dilutions of 1:20 to give an ultimate concentration of 25 µgm−3 (2.5 × 10−8 µgml−1), a concentration still approximately 50× higher than any found in the atmosphere. (a) (b) FIGURE 4.1. Characterisation of Fe2O3 nanoparticles used to intentionally introduce magnetic impurities. (a) pXRD showing pure phase α-Fe2O3 and (b) TEM of Fe2O3 nanoparticles demonstrating the sub-micron size distribution. 62 4.2. MAGNETIC PURITY CONCERNS Solutions of Fe2O3 polluted coronene in toluene solvent systems were prepared as mentioned in section 2.2.9 and the crystals harvested. Regardless of the concentration of Fe2O3 suspension used, both the polymorph and morphology of the crystals appears to be unaffected. Crystals grew consistently at a range of sizes indistinguishable from those grown with no added Fe2O3 nanoparticles. A general trend was observed in the relative sizes of the crystals grown under the different magnetic fields upon repeats, however both polluted and control solutions showed the same trend. Crystals grown above 1 T were shorter with a more ribbon-like morphology implying either suppression of growth along the b-axis or promoted growth along the a or c-axis. Both scXRD and pXRD showed all crystals to be the γ polymorph with the secondary growth direction along the c-axis under 2 T. The lack of the β phase is assumed to be a consequence of the inability to finely control the magnetic field strength and faster cooling ramp. A majority of the crystals grown at 0 T and 1 T had a higher aspect ratio than analogous crystals grown at 2 T, which were shorter and more numerous. Thus these results are in line with the previously observed suppression of crystallisation by the field, peaking between 0.9 T. (a) (b) (c) −1 −1 FIGURE 4.2. SEM micrographs of 2.67 × 10 µgml Fe2O3 contaminated coronene crystals grown at (a) 0 T, (b) 1 T and (c) 2 T. Red circles indicate EDX data showing exclusively C. Green circles indicate points with detectable Fe content via EDX analysis; Fe=24.8 %, O=42.8 % and Fe=34.1 %, O=29.2 % for 0 T and 1 T respectively. Crystals obtained in the experiments examined using SEM at 4.17 × 10−9 µgml−1, 8.33 × 10−8 µgml−1, −6 −1 −5 −1 −4 −1 1.67 × 10 µgml , 3.33 × 10 µgml and 6.67 × 10 µgml had no Fe2O3 on or around the crystals detectable visually or using EDX, indicating that nanoparticle inclusion in the crystal is either rare or absent completely. This is not totally unexpected as coronene crystallisation has been reported to be unimpeded by anything foreign to the system, even other PAHs[97]. At 63 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH −2 −1 −1 −1 the highest concentrations however, 1.33 × 10 µgml and 2.67 × 10 µgml , Fe2O3 particles were identifiable on the the crystal surface of those grown at 0 T and 1 T using EDX (figures 4.2, (a) and (b)). It should be noted that the surface adhesion of these particles are likely a function of removing the crystals from solution as they appear adhered and not embedded. Analysis for the crystals grown under fields of 0 T, 1 T and 2 T showed no observable differences. As the particles are only on the surfaces of the crystals, not within them or at the obvious nucleation centres, it is reasonable to conclude that the inclusion of magnetic nanoparticles does not interfere with coronene crystal growth. 4.3 Molecular interaction calculations Over the past three decades computational chemistry has become an increasingly powerful tool in the research chemists tool box. An increase in the power of computational hardware coupled with an extensive range of user friendly software, has made computational modelling of things like reaction pathways and molecular optimisation accessible to those outside the field. Performing chemistry on a computer rather than in a wet-lab, while fraught with drawbacks (e.g. novel experiments rely on assumptions and may not correctly mimic reality) has significant advantages. The ability to view in atomic detail, how a known reaction likely occurs down to bond rotation and radical formation, not only lends valuable insight to why and how these reactions occur but can be used to predict reaction bottle-necks and intermediates in novel reactions before any glassware has even been washed. As subunits come together to form a crystal in solution (equation 1.5) mentioned in section 1.4.2, Ostwald’s rule of stages suggests that the least stable polymorph is formed first followed by a host of unstable configurations, finishing in the most stable for the environment in which it forms. As direct observation of such a small, dynamic system is not currently possible, it becomes prudent to use geometry optimisation calculations in which to study the likely interaction of small numbers of molecules and design strategies in which to interrogate nucleation. Indeed, computational modelling of cluster formation can offer insight into crystal nucleation currently unattainable through experimental techniques[149]. In fact, various working models for nucleation currently employed include two-step model mentioned in section 1.3.2 originate from computational methods[6,17]. Having ascertained in chapter3 that both the γ and β forms of coronene sit in an apparent energy minima and that β coronene is the most stable form at lower temperatures, the question of how small numbers of molecules interact can shed light of the crystallisation mechanism of coronene. DFT-D optimisation calculations were conducted using the dispersion corrected functional b97d with the basis set 6-31(d), configured to calculate up to 5 d-orbitals. Molecules were set >10 Å apart and orientated randomly before initiating optimisation. 64 4.3. MOLECULAR INTERACTION CALCULATIONS (b) (a) (d) (c) (f) (e) FIGURE 4.3. DFT-D optimisation of small coronene aggregates of size (a) 2 (b) 3 (c) 4 (d) 6 (e) 8 and (f) 10 coronene molecules. Calculations were run in Gaussian09 software after building molecules and setting coor- dinates on GaussView05. Figure 4.3 shows the optimised conformation of 2, 3, 4, 6, 8 and 10 molecules of coronene. To initially benchmark the calculation the data obtained from figure 4.3 [113,150] (a), πd=3.343 Å and doff =1.611 Å was compared and found to be a match with literature . Subsequently, as shown in figure 4.3, regardless of the number of molecules optimised here, 65 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH FIGURE 4.4. All measurable values of (A) doff , πd and B ∠mol from calculated clusters, some of which are displayed in figure 4.3. As guides for the eye, blue and red markers indicate even and odd numbers of molecules in the clusters. Orange and green lines show values for β and γ-coronene respectively. where measurable, doff lies between 1.59 Å to 1.62 Å with πd between 3.35 Å to 3.60 Å and ∠mol 90° to 170°. These data are summarised in figure 4.4. It should be noted that the orientations graphically displayed in figure 4.3 are a single converged arrangement of the number of molecules stated, for example three molecules could arrange as displayed in figure 4.3 (b) or a single stack of three molecules. Subsequently, as expected, the number of converged geometries increased as the number of molecules increased, however these variations did not differ by more than 0.1 kJmol−1. It can be seen from figure 4.4, values of doff , πd and ∠mol measured from the small clusters, greatly favour those found in a bulk β-coronene crystal over γ. All values of doff from the calculated geometries were within 4 % of those with the β polymorph and almost half that found in the γ polymorph. Values of ∠mol are distributed over a wide angle range and clearly do not favour a specific, nearest neighbour, angle to the same degree as doff . This may in fact give some qualitative indication as to the magnitude of how much effect each 66 4.3. MOLECULAR INTERACTION CALCULATIONS FIGURE 4.5. Relative energy data of 2 coronene molecules locked at πd = 3.433 Å calculated using a dispersion corrected B3LYP functional with a TZVP basis set. Energy is given as relative from coronene molecules set 12 Å apart. Inset images show the geometry of the molecules at the specified positions; perfect overlap (at 0 Å and β overlap at 1.59 Å. The red arrow indicates doff = 3.17 Å, where γ is expected to sit. interaction has on the final polymorph, that is, the appearance that coronene molecules will adopt almost any angle with its neighbour in order to conform to doff ≈ 1.59 Å with another molecule. As can be seen in figure 4.5, there is a clear minima at doff = 1.5 Å in which to expect the molecules to orientate. This minima closely matches the doff = 1.59 Å seen in β-coronene. However, the limitations of such a single pass restrict any information based on the rotation of the molecules relative to each other. For a clearer picture, two molecules were once again pinned to a stack distance of 3.433 Å and a glide of 0.1 Å increments between 0 Å to 4 Å additionally, a rotation of 30° at increments of 2° was introduced. Although multiple attempts at optimising the geometry of two coronene molecules was attempted, all of which gave doff = 1.61 Å, to search for any additional, low energy conformations of a coronene dimer an energy profile was calculated and can be seen in figure 4.6. Initially, coronene molecules were restricted to a stack distance of 3.433 Å, the distance in which a coronene 67 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH FIGURE 4.6. An energy map of 2 coronene molecules locked at πd = 3.433 Å. Energy is given as a difference from the lowest value. Red arrows indicate the locations of molecular geometry found in the polymorphs of coronene. dimer relaxes and shifted by increments of 0.1 Å and the energies recorded. Figure 4.6 gives a more complete picture of the energy landscape of a coronene dimer restricted to πd = 3.433 Å. The colour map describes the relative energies from the lowest calculated value at doff = 1.5 Å and −1.5 Å and an angle offset of 6° (or symmetrically, 54 degree). Energy wells can clearly be seen at this and any symmetrically equivalent point, as well as doff = 0 Å and an angle offset of 30°. Labelled arrows indicate the geometry of any adjacent stacked molecules for both polymorphs. As can be seen, molecules in β-coronene are sat within a local energy minima while those of γ sit at a higher energy, with a difference of ≈ 19 kJmol−1 relative to β. Figures 4.3 and 4.6 illustrate the stark contrast of how the ratio between C···C and C···H interactions contribute to the stability of each polymorph and how this is observed throughout PAH crystals[28]. Namely that the favoured growth along [010] would likely be exaggerated in β-coronene over γ-coronene whilst growth along [001], specifically the T-shaped interactions between the stacks, would likely show the opposite trend. This perhaps explains why crystals of β tend to grow in a ribbon-like morphology (b > c > a, fig.4.7 (a)) and γ with a more typical, 68 4.3. MOLECULAR INTERACTION CALCULATIONS (a) (b) FIGURE 4.7. Schematic version of figure 3.6 for ease of visualisation showing typical (a) β and (b) γ morphologies. Blue, green and red arrows indicate the a, b and c growth directions respectively. needle-like morphology (b > c ' a, fig.4.7 (b), also 3.6). Based on the clear preference of β overlap when clustering in small numbers, it stands to reason that there must be a critical size, in which the γ geometry becomes favourable for a given temperature. The data presented in section 3.3.1.3 also suggests crystallite size is a factor in directing polymorph selection. Although this present work has dealt with coronene clusters of up to 10 molecules, the preference for β overlap can still be seen in calculated clusters of up to 500 molecules giving a particle radius of 7.512 nm, although polymorphism was not the focus of the study[151]. 4.3.1 Absorption prediction using TD-SCF calculations As small clusters of coronene molecules are preferentially adopting a similar doff as the β form and that β-coronene has a absorption spectrum significantly different from γ-coronene, a difference in absorption may be detectable using spectroscopy. As mentioned earlier, time dependant self-consistent field (TD-SCF) calculations can be used to estimate the electronic transitions a molecular system allowing for prediction of optical properties including absroption. The first 15 electronic transitions for all optimised geometries from figure 4.3 were calculated at the TD-B97D/TZVP level used by Li[113]. These calculations were bench-marked against those of a single molecule of coronene and can be seen in figure 4.8. It can be seen from figure 4.8 that the larger the cluster, the lower the wavelength absorbed and the lower the absorbance intensity. The absorption profile is different for at least one, two and three molecule clusters. A single molecule has been given a single peak λmax =338.6 nm, whereas the two molecule calculation has been given 3 peaks of similar intensity, λmax =364.0 nm with 2 69 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH FIGURE 4.8. Calculated, TD-SCF absorption data at relative intensity, for coronene clusters of size 1, 2, 3 and 4 molecules coloured black, red, green and blue respectively. Tick marks under the graph indicate positions of calculated electronic transitions of non-zero intensity. secondary peaks at λ =386.5 nm and λ =449.5 nm. The three molecule profile has again, only a single peak of notable intensity, λmax =448.0 nm. Four molecules have a lower energy absorption but of only negligible intensity. Most importantly, the difference in absorption profiles suggests that identifying single coronene molecules and clusters of β overlap concomitantly, is plausible. Indeed, tracking the concentrations of π-stacked dimers using ultraviolet (UV) spectroscopy is well established[152]. If the identification of peaks allows the monitoring of different cluster concentrations in solution, using in situ UV-vis spectroscopy, how a magnetic field affects peak intensity will be suggestive as to a mechanism for the observed polymorph selectivity. 70 4.4. PRENUCLEATION IDENTITY OF CORONENE 4.4 Prenucleation identity of coronene In order to find qualitatively, if absorption peaks due to clusters in solution can be measured in a coronene-toluene solution concentrated enough,UV-vis spectroscopy was used to analyse a solution saturated at room temperature. Figure 4.9 shows the UV-vis absorbance of a saturated solution of coronene in toluene after two weeks of saturating at 93°. The trace with the most intense absorbance was collected from the neat coronene solution once allowed to cool and was − subsequently measured at 1.7 mgml 1 (5.66 mM) after 4 h at room temperature. Each subsequent spectrum is collected from a one in three dilution of the previous concentration for 30 spectra − from 5.66 mM to 4.43 × 10 5 mM. FIGURE 4.9. UV-vis spectroscopy of concentrated coronene in toluene. The initial concentration was found to be 5.66 mM with sequential dilutions are collected at one in three of the previous concentration. All colours are a guide for the eye. At the lower concentrations, the absorbtion profile is in good agreement with literature[153,154]. However this profile is starkly different from those at the highest concentrations with the appearance of peaks at 366 nm, 387 nm, 402 nm, 410 nm, 427 nm and 450 nm that have not been previously ascribed to coronene or any known breakdown product or coronene oligimers. Discrete peaks in absorbance indicate electronic transitions of a particular energy, that is generally not the result of scattered light which appears as a broad and noisy peak as is seen below 350 nm under the highest concentration. 71 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH When compared to the positions of calculated absorbance peaks in section 4.3.1 for clusters of 2, 3 and 4 molecules, similarities can be seen. For the β stacked dimer, peaks at 364.0 nm, 386.5 nm and 449.5 nm match convincingly with the experimental peaks of 366 nm, 387 nm and 450 nm respectively. The calculated absorbance for the trimer at 448.0 nm also fits well with the 450 nm peak. Further peaks observed at 402 nm, 410 nm and 427 nm remain unaccounted for using this method and may be the result of higher order clusters or particle-particle interactions, i.e solute- solute, cluster-cluster, solute-cluster interactions due to the higher concentrations[155]. With no literature on the absorptivity of clusters of different sizes and indeed if they behave differently than molecular coronene, not knowing the concentration of these clusters in solution, this is hard to quantify. These data however, do describe a species in solution other than molecular coronene absorbing at a lower energy. 4.4.1 Dynamic light scattering DLS is a commonly used technique to size particles in solution using the way in which light that is scattered through a sample changes as a function of time. Moving by Brownian motion, the speed of which is determined by solvent viscosity, temperature and particle size, the faster the scattering intensity changes at a given temperature, the smaller the particles. The change in intensity is analysed using a correlation function which will, due to greater changes, decay over time thus a long decay is indicative of larger particles. Attempting to measure coronene clusters of a pre-critical size, a hot, filtered solution of coronene in toluene was tested for particle size as mentioned in section 2.2.3. Data collected at −5 ◦C increments from 90 ◦C to 20 ◦C can be seen in figure 4.10. Analysis of the data collected reveals a number of inconsistencies that throw the reliability of the data in to question. First and foremost, the initial collection at 90 ◦C suggests the presence of particles of diameter 1.106 µm which would be substantially larger than any particles remaining from the filtration with a pore size of 0.22 µm. It is possible that the laser light used during measurements itself, induced the formation of micro crystals after forming an electric field across amorphous aggregations as has been reported in other systems such as urea and glycine[156–158] but there is no further data to substantiate this in the coronene - toluene system. Secondary to this, the particle size measurements are not stable with regards to themselves. Variations in repeat measurements suggested that either the particle size distribution was dynamic, even after 30 min equilibration time, or there is a further issue with the system regarding DLS measurements. This is likely due to a number of factors which would make this particular system unsuitable for DLS. Solubilised coronene is a coloured, intensely fluorescent solution, both of which can make DLS measurements difficult due to light absorption and differentiating between scattered light and the fluorescence of a sample. Another complicating factor regarding DLS measurements is the particle dispersity (Ð) and particle shape due to the use of the Stokes- 72 4.4. PRENUCLEATION IDENTITY OF CORONENE FIGURE 4.10. DLS data from a cooling system of coronene in toluene. Data was taken in −5 ◦C increments cooling from 90 ◦C to 20 ◦C. Einstein equation (eq. 4.1) to determine the diffusion coefficient (D) used when determining size: kT (4.1) D = 6πηrs Where k is the Boltzmann constant and T is the temperature of the system, η is the solvent viscosity and rs is the hydrodynamic radius of the particles. Therefore, in an idealised sample, a solution has spherical particles with Ð=1, that is little to no variation in particle size. As the data out measures a decay of a correlation function, a high Ð can cause peak broadening and an averaged decay can give completely the wrong values. There are indeed various mathematical ways around the fitting of a wide range of particle sizes but without suitable data to plug into the functions, measuring a solution with a high Ð is difficult. There is no reason to think that once coronene starts to crystallise, there will be a spontaneous formation of spherical PNC of a similar size. Based on the data, the truth is likely closer to an array of anisotropic, needle shaped particles of a wide range of sizes making DLS apparently unsuitable for particle sizing coronene PNC. Interestingly however, the only time figure 4.10 shows any particle size data, other than that with distributions with means between 0.712 µm to 2.669 µm is the appearance of distributions 73 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH centring on particle sizes of 68 nm and 58 nm for analysis at 40 ◦C and 35 ◦C respectively. Due to the intensity generated by particles with such significantly reduced volumes (1.02 × 10−4 µm3 and 4.85 µm3 for spherical particles with diameters 0.058 µm and 2.1 µm respectively) the numbers of these smaller particles are likely several orders of magnitude more numerous. However, as this data is not particularly reliable, it remains a qualitative curio. 4.4.2 Diffusion-orientated NMR spectroscopy Another sensitive method for solution based particle size analysis is DOSY. This technique is also based on the Stokes-Einstein equation (equation 4.1) and therefore relies on larger particles moving at a different rate through a solvent of known temperature and viscosity than smaller particles. In a standard DOSY experiment, a magnetic field gradient is applied to the sample, contrasting to a standard NMR experiment which uses a uniform field over the whole sample. Once the initial pulse of the magnetic field has occurred and nuclei have been dephased by different amounts along the length of the sample tube, a second counter-gradient is pulsed after a known amount of time (∆T). If all affected nuclei in the sample remain stationary during ∆T, between the two opposing pulses, the sample is refocussed and acquisition of re-emitted radiation can begin as with a standard NMR sample with full intensity. Realistically however, a solution is in motion and particles will be in different positions between the two pulses. The practical consequence of this is a reduction in the re-emission intensity for each nucleus that is not refocussed by a perfectly counter-acting magnetic field, which would only be the case if the particle was in the same vertical position after ∆T. If the gradient of the field is increased in magnitude, the signal loss becomes more pronounced over the same distance because there will be a higher field difference over distance d with a steeper gradient. Therefore if the pulsing is repeated while the magnitude of the magnetic field gradient is increased sequentially, information about the rate at which particles can move through (a) (b) (c) FIGURE 4.11. Basic description of a standard DOSY experiment. (a) and (b) show the positions of two different particle sizes during both gradient pulses. Dotted lines represent a change in the gradient magnitude. (c) describes the change in intensity as the magnitude of the gradient increases. 74 4.4. PRENUCLEATION IDENTITY OF CORONENE FIGURE 4.12. DOSY data collected for two concentrations of coronene in toluene, with peaks labelled. (a) and (b) show data for concentrations of 1.65 mM and 3.30 mM respectively. On both graphs, the top trace is a single spectra taken from the lowest field gradient of each and the left trace is the peak integral at each value of D. 75 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH solution can be obtained. Figure 4.11 shows a brief description of how DOSY works. Figures 4.11 (a) and (b) show two particles of different sizes during the first and second pulses respectively. The dotted lines represent field gradients of varying magnitudes from B to -B along the length of the sample tube. The peak intensity drop for each particle as a function of field gradient magnitude is described in 4.11 (c) with the dotted lines indicating the decay fitting used to calculate the diffusion coefficient. Samples of coronene in deuterated toluene were prepared as described in section 2.2.4 at concentrations of 3.30 mM and 1.65 mM. Each DOSY spectra were calculated by fitting the decay from 15 proton spectra taken sequentially and are displayed in figure 4.12. It can be seen from this data that there is clear separation of peaks due to both the solvent and the solute. The single peak due to coronene protons has a chemical shift of 8.60 ppm in both spectra which is in good agreement with literature[137,159]. The shifts for toluene in both data sets are 2.06, 6.96, 7.00 and 7.07ppm which is again in good agreement with literature[160]. The diffusion co-efficient for coronene however does change slightly from 9.67 × 10−6 cm2 s−1 and − − − − 1.06 × 10 5 cm2 s 1 for 9.67 × 10 6 mM and 1.06 × 10 5 mM respectively, which when applied to equation 4.1 with the experimental temperature (295 K) and the reported viscosity for toluene at this temperature[161] give hydrodynamic radii of 5.79 Å and 6.35 Å, respectively. Although the more concentrated solution does give a larger hydrodynamic radius than the less concentrated solution by 0.56 Å, both of these values are smaller than the known values for coronene, which has a diameter of 7.48 Å. Possibly due to the anisotropic shape of the molecule and the averaged signal providing a length smaller than the actual diameter of the coronene disc. Ultimately as clustering molecules of coronene appear to have no detectable effect on the magnetic shielding and thus the chemical shift of the molecules, extracting only an averaged value of coronene particle size is possible. While this may help with understanding how magnetic fields can affect the distribution of coronene cluster sizes, the physical limitations of the DOSY experiment make it impractical at present for any magnetic growth experiments. 4.5 Coronene growth under high magnetic fields 4.5.1 In situ. measurement preparation After the failure to obtain reliable in situ particle size data using diffusion based techniques, attempting to find clusters spectroscopically was attempted at the HFML as described in section 2.2.7. Figure 4.13 shows absorbance spectra obtained in 3 separate ways. The yellow plot shows the absorbance for a diluted solution of coronene in toluene collected on a standard bench-top lab UV-vis spectrometer as described in section 2.3.3. The blue line shows the absorbance of a saturated solution of coronene in toluene at 80 ◦C collected in situ. inside the bore of the cell 5 magnet at the HFML. The data is truncated at 320 nm due to the restrictions of the optical equipment available. The black, red and green plots are calculated 76 4.5. CORONENE GROWTH UNDER HIGH MAGNETIC FIELDS FIGURE 4.13. Experimental and calculated data for coronene clusters. The yellow trace is the absorption spectra for a dilute coronene solution collected on a bench top UV-vis spectrophotometer. The black, red and green dotted traces are calculated absorption of 1, 2 and 3 clusters of coronene molecules respectively, in the lowest energy β form. The blue trace is experimental absorption data collected of a saturated solution in situ. under an applied magnetic field, truncated due to the limitations of the light source, spectrometer and optical cables available. absorbance spectra of the 1, 2 and 3 molecule clusters from section 4.3 that adopted a β like structure. In experimental data for a dilute sample obtained locally, reported absorbance data for solution based coronene and in the solid state (as crystals) no peaks at 366 nm have previously been observed. This is likely due to the high concentrations used in this experiment being far higher than those normally used when using a standard UV-vis spectrophotometer. Evidence for this can be seen in the dilution, absorption data in figure 4.9, where a peak at 366 nm can be seen above concentrations of 0.33 mM along with peaks at 387 nm, 402 nm, 410 nm, 427 nm and 445 nm that do not appear in the experiential data here. When compared to the calculated data for a two-molecule cluster, a small peak at 366 nm is indeed observed which is to be expected as the π-stacking direction has been reported as the lowest energy electronic transition in planar PAHs[162] which suggests a change in π-stack overlap would have a significant effect on the electronic properties of the clusters. Unfortunately, 77 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH it cannot be conclusively stated that this peak is due to β like clusters due to the lack of observing other peaks calculated and this being the highest energy of the three. This, coupled with the uncertainty based around properties predicted via computational chemistry in more complex systems, means the matching of the peaks can be used only as a potential candidate regarding to the identity of the peak. The lack of the lower energy peaks here is likely a function of the age of the saturated solution as time restrictions at the HFML facility did not allow for 7 d of saturation preparation. If the larger clusters have lower energy transitions, it is probable that time is required for these to form to a sufficient size and number to be detectable. 4.5.2 Crystal growth Given the profound effect of magnetic fields of 0 T to 1 T have on the ultimate polymorph of coronene, the effect of fields >1 T were investigated. The HFML situated at Radboud University in Nijmegen, NL is a facility for carrying out experiments at high sustained fields of up to 37.5 T[163]. Each cell at the HFML contains a large Bitter electromagnet with bores of up to 50 mm requiring 150 dm3 s−1 of chilled water in order to function. This high reliance on large volumes of chilled water coupled with power consumption of up to 20 MW, places obvious restrictions on the duration of experiments. Thus far, as β-coronene has only been grown over time scales that would be impossible to replicate at such a facility (4 d), attempting to grow this specific polymorph at higher fields would be difficult. In lieu of looking for β-coronene at higher fields, assessing how high magnetic fields affect the crystallisation process can be attempted. As mentioned in section 2.2.7, a saturated solution of coronene in toluene was left at 80 ◦C for 24 h before being filtered in to a quartz cuvette and inserted into the pre-heated bore of the magnet. After being held at 90 K to stabilise, the magnetic field was applied before the bore was cooled to 5 ◦C over 68 mins (1.25 ◦Cmin−1). As a control, a spectrum was recorded over the temperature range to assess signal quality and can be seen in figure 4.14. When comparing the absorbance to that in figure 4.8 it can be seen that clustered forms of coronene appear to be present in solution. The high amount of light scattering at concentrations at or near saturation makes obtaining a reasonable line profile from 500 nm down to 280 nm difficult. Therefore transmission of light from a 300 nm to 1100 nm light source was detected in situ., so the concentration of the solution could be monitored in real time and the spectrum recorded every 10 s. For this reason the spectrometer was tuned to read the single peak at approximately 366 nm attributed to an overlap of various sizes of β stacked clusters. Unfortunately, as the growth of β-coronene in a cuvette typically results in 1-5 crystals, witnessing the inception of a single crystal is unlikely, due primarily to the detected area of collected light covering only a small fraction of the total volume of solution. As coronene grows in toluene, the crystals typically fall to the bottom of the growth vessel leaving less mobile solvated 78 4.5. CORONENE GROWTH UNDER HIGH MAGNETIC FIELDS FIGURE 4.14. Absorbance of coronene at 0 T as a function of temperature. Each spectra plotted represents a snapshot taken at one minute intervals. The black line and symbols indicates how the temperature program changes over time with the red triangle indicating theC T , that is the intensity of the peaks sequentially declines. coronene resulting in a more dilute solution. Therefore, as the spectroscopic monitoring of the solution was the only practical option, a sequential decrease, that is a continuous decrease in a 3 point average, in absorbance was used to determine the onset of crystallisation. From here on known asC T . Data from a single temperature sweep at 0 T can be seen in 4.14. The line profile of the measured absorbance can be seen in figure at the front of the graph, the peak at 366 nm can clearly be seen. The temperature program is indicated by the black line and symbols, withC T marked with a red tetrahedron. As the temperature program was the same for each sweep, both time and temperature can be used when referring to the point of nucleation when analysing this data. For consistency in this thesis, temperature will be used. Figure 4.14, displays how the concentration of the saturated 79 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH coronene solution remains constant until a certain temperature is reached before slowly starting to decrease, in this case at 42 ◦C. The rate of absorption intensity decreases, then becomes more linear as the system approaches 5 ◦C. For ease of visualisation and comparison between data sets, peak height can be plotted as a function of temperature. FIGURE 4.15. Absorption intensity at 366 nm as a function of temperature at 0 T. The temperature at which crystallisation begins is indicated with a black arrow. The red arrow indicates an anomaly seen in various sweeps. Figure 4.15 shows the absorption intensity of the peak at 366 nm during the temperature sweep at 0 T. The black arrow indicates the point at which crystallisation was taken to occur ◦ CT , which in this case is 42 C. The red arrow indicates a commonly observed anomaly in these experiments, this temporarily breaks the trend of the decline, manifesting as either a sudden increase of absorption of a short plateau. As this anomaly appeared seemingly at random during the declining section of the data and was attributed to crystals forming in or falling through the light path due to its irreproducibility during repeats. Once a stable method of transferring the sample to the magnet bore, giving reproducible results at 0 T was cultivated, absorption was measured as a function of temperature for field strengths of 0 T, 0.25 T, 0.5 T, 1.0 T, 2.0 T, 5.0 T, 10.0 T and 20.0 T. The crystallisation temperatures for field strengths of 0 T to 20 T shown in figure 4.16 are summarised in table 4.1 with images of the resulting crystals in figure 4.17. 80 4.5. CORONENE GROWTH UNDER HIGH MAGNETIC FIELDS FIGURE 4.16. Absorption intensity at 359 nm as a function of temperature for fields of 0 T, 0.25 T, 0.5 T, 1 T, 5 T, 10 T and 20 T. Initiation of crystallisation temperature is indicated by a black arrow. From looking at the morphology of the crystals obtained from the high-field experiments, no quantifiable differences can be seen from a standard growth without any magnetic field present. Although slight changes can be seen, without sufficient repetitions to spot trends in morphology, no conclusions can be drawn. Interestingly, during the initial set up and calibration of the equipment at the HFML facility, after a 0.7 T cooling run, a small crystal of β-coronene was discovered upon post-analysis. This crystal was used as a seed, attempting to grow bigger β-coronene crystals without the application of a magnetic field, but was unsuccessful. Although all measured coronene from these experiments was the γ polymorph, analysis of these data, reveal a stark difference between theC T of coronene crystals under various applied field strengths and compared to the control. It appears that in general, the application of a magnetic field at any strength up to 20 T causes a suppression ofC T within the coronene - toluene system when compared to that of no field present. At the lowest field tested, 0.25 T,C T was delayed to 20 ◦C, a suppression of 22 ◦C from that of the control. This trend continued as the field was increased to 0.5 T which began to noticeably crystallise at 6 ◦C, a suppression of 38 ◦C from 81 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH Table 4.1: Crystallisation temperature of coronene under high magnetic fields Field Crystallisation Powder ◦ strength (T) temperature (CT )( C) diffraction 0 42 §0.5 γ 0.25 20 §0.5 γ 0.5 6 §0.5 γ 1 11 §0.5 γ 5 19 §0.5 γ 10 32 §0.5 γ 20 34 §0.5 γ no applied field. This initial, seemingly linear trend is broken however when the field is increased to 1 T, which begins to crystallise at 11 ◦C, a suppression of 31 ◦C from the control but 5 ◦C sooner than 0.5 T. From here, any increase of field strength appears to reduce the suppression delta from the control experiment. Applied fields of 5 T, 10 T and 20 T reduced the delta to 23 ◦C, 10 ◦C and 8 ◦C, respectively. Analysis of these data suggest that there is more than one physical attribute of crystallisation being affected by the magnetic field with the most profound effect occurring between 0.5 T and 1.0 T. For a finer look at crystallisation around these field strengths, further experiments using the same experimental parameters were carried out at 0.7 T, 0.8 T, 0.9 T, 1.0 T, 1.1 T and 1.2 T. As the temperature of crystallisation of a saturated solution of coronene in toluene was depressed to 6 ◦C, a single degree higher than the lowest temperature of which the equipment was capable, the solvent was modified to make coronene less soluble. In order to get a detailed picture around these lower field strengths, hexane was added to the saturated solution as an anti-solvent at a ratio of 1:10. Figure 4.18 shows this data in the same format as the previous field strengths (fig. 4.16). As can be seen, there is a clear rise in the crystallisation temperature of coronene when hexane is added. The control, 0 T run started crystallising at 63 ◦C, a difference from the 100 % toluene of 21 ◦C. However, the largest temperature gap in the initiation of crystallisation from experiments carried out between 0 T to 20 T was much higher than the experiments containing hexane. The maximum difference in the 100 % toluene experiments was 36 ◦C, between 0 T and 0.5 T, whereas the maximum difference in the current experiments is 15 ◦C between 0 T and 0.8 T, ◦ ◦ 21 C smaller. The trend however looks remarkably similar. 0 T has the highestC T of 62.5 C ◦ §0.5 which, when the field starts to increase, has a fairly linear decrease inC T to 47.4 C §0.5 at ◦ 0.8 T, passing through 0.7 T withC T =49.2 C. The change between 0.8 T and 0.9 T, is the smallest with just a 0.1 ◦C difference up to 47.5 ◦C, which is within the error. Once the field increases past 0.9 T, only an increase inC T is seen with the 1.0 T, 1.1 T and 1.2 T experiments havingC T s of 49.9 ◦C, 52.2 ◦C and 58.3 ◦C §0.5 respectively. 82 4.6. SUMMARY FIGURE 4.17. Images of crystals grown under fields 0 T to 20 T. Blue squares are 1 mm. Bottom centre image shows a small β-coronene crystal grown during a calibration run under 0.7 T. The data from both sets of experiments are plotted here in figure 4.19. When looking at both sets of data together, a maximum suppression ofC T can be seen somewhere between 0.8 T and 0.9 T with crystallisation under both higher and lower magnetic fields increasingC T . 4.6 Summary Attempting to understand the mechanisms involved with the appearance of the β-coronene polymorph when crystallising under an applied field, the possibility of the polymorph being affected by magnetic impurities has been addressed. Any amount of added Fe2O3 nanoparticles introduced to the solution had no apparent effect on the polymorph or morphology of the resultant crystals. With no suggestion of heteronucleation this is good evidence for the lack of potential impurities contributing to the crystal growing system. Computational optimisation of small clusters of coronene suggest fairly convincingly that 83 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH FIGURE 4.18. Absorption intensity at 359 nm as a function of temperature for fields of 0 T, 0.7 T, 0.8 T, 0.9 T, 1.0 T, 1.1 T and 1.2 T. Initiation of crystallisation temperature is indicated by a black arrow. free molecules prefer to sit with doff =1.611 Å to lower the energy of a face-face interaction. This behaviour extends at least up to 10 molecules and is shown to be within 0.02 Å of doff within the β crystal. Interestingly, the lack of a consistent ∠mol within the same geometries convey the importance of C···C interactions when these crystals grow. Additionally, that the initial clustering of molecules at the inception of a nucleus is likely dominated by doff ∼1.6 Å interacting molecules. If this interaction were being somehow stabilised or re-enforced by the magnetic field, it would be expected to see crystals with doff ∼1.6 Å at sizes beyond those observed without a field applied. Understanding this, characterising the identity of solubilised coronene at high concentrations becomes key. Using UV-vis spectroscopy, higher concentrations of coronene do indeed appear to exist, in at least some form, in small β stacked arrangements when compared to calculated absorption. Although far from conclusive, the presence of sharp peaks and therefore discrete electronic transitions, suggest the presence of a uncharacterised species in solution bearing a marked resemblance to the calculated absorption peaks for these particles. Determining the size of these clustered particles was less successful. Both DOSY and DLS gave frustratingly ambiguous results due to restrictions imposed by the coronene/toluene system 84 4.6. SUMMARY FIGURE 4.19. Comparison ofC T of both sets of experiments in high-field magnets. Each graph summarises the data collected between (a) 0 T to 20 T and (b) 0 T to 1.2 T taken from figures 4.16 and 4.18 respectively. 85 CHAPTER 4. THE MECHANISM OF MAGNETIC CRYSTAL GROWTH and the techniques themselves. DOSY implied that higher concentrations of coronene does involve larger particles than the less concentrated analogue, but the low concentrations used in order to retain a solubilised sample at ambient temperatures do not allow for any parallels to be drawn with the higher concentrations data. DLS data provided a scattered impression of particle sizes in solution, indeed even a solution filtered through a 0.22 µm filter showed signs of micron-sized particles instantly. The appearance of new peaks at 42 ◦C may indicate the formation of new crystal nuclei but this cannot be substantiated from this data. The suppression of crystallisation from solution when under an applied magnetic field is an under-reported phenomena. Here a delay in the onset of crystal growth has been observed in two different solvent systems. Most noticeably, coronene in toluene saw the crystallisation temperature drop by 37° just between the applied fields of 0 T to 0.9 T, with 42 ◦C at 0 T matching nicely with the emergent peaks in the DLS. Supression of nucleation is entirely consistent with reported systems in literature where fewer, higher quality crystals are grown in the presence of a magnetic field [164–166]. At higher levels of supersaturation field-induced suppression of post-critical nuclei, the solute feedstock surrounding a nucleus above critical size will be increased, supplying more material for filling defects during growth. As molecular crystals are known to readily form different polymorphs under different conditions[3] one of which would be concentration, for example the epilepsy drug carbamazepine has five known forms, one of these, form I, is currently achievable from the melt. This suggests that form I is accessible through higher concentrations than that achievable in solution, that is, an increase of the solute concentration can directly affect polymorph selectivity and in this regard carbamazepine is by no means an isolated case within molecular crystals[67,88,167,168]. By reaching lower temperatures at the same concentration, that is a higher supersaturation, it is possible that one of the conditions required to produce β-coronene is being satisfied. The experimental data therefore suggests that the magnetic field is in fact, achieving polymorphic selectivity through an increase of supersaturation via suppression of nucleation and that this effect will be the most pronounced between 0.8 T and 0.9 T in the coronene - toluene system. Perhaps most interestingly, these data have presented a suppression of nucleation seemingly affected by two counteracting effects, the steep suspension of 37 ◦C within just 1 T and then the gradual decrease in this effect as we approach high-fields. It has been reported that concentration plumes can be inverted under high magnetic fields[15] but the stark differences between the apparent forces at play, one steeply linear with a negative gradient and the other reminiscent of growth to saturation, would not support this idea. The under-studied nature of how magnetic fields affect a dynamic system like the growing of crystals from solution make it easy to speculate but hard to substantiate what is actually occurring. As the energetic barrier for a change in polymorph apparently becomes higher as crystal size increases, it seems plausible that whatever effect the magnetic field is having, it is in the initial stages of molecules clustering. Previous studies on how magnetic fields effect water have produced contradicting results reporting a 86 4.6. SUMMARY strengthening and weakening of H-bonding networks making it abnormally difficult to draw any conclusions. The two forces holding a coronene crystal together, namely C···H and π-stacking do help reduce the possibilities however. If, for instance the applied field weakened C···H and strengthened π-stacking through molecular charge distribution by induced ring-currents, causing the reflexive behaviour observed. 87 a nwihtesld r omd oi rsascnas egonfo ouinlk molecular like solution from grown be also can crystals Ionic formed. are solids the which in way ,ms otc n otiigaBpaei ai f11i h rsneo .Ti becomes This O. of presence the in 1:1 of ratio a in phase B a containing one contact must A, the in crystals molecular from greatly differ crystals covalent or ionic 1 , chapter in mentioned As hsde o nyrqiehg eprtr odtosbtdet h ako oiiyi the in mobility of lack the to due but conditions temperature high require only not does This oi tt,tesocimtyo ri nefcsms lob da.Freape uigthe during example, For ideal. be also must interfaces grain of stoichiometry the state, solid fcus ti ohpsil n ieyta nemdaetrir hsswl form will phase. phases target tertiary the intermediate of that formation likely successful and for possible ratio both correct is the it in course O Of of presence ScVTe the like in oxide meet quaternary must complex more However off-stoichiometries. a or of intermediates formation stable for are there when important more increasingly ABO like oxide metal tertiary a of formation state solid materials. starting solid the the within in atoms formed between crystals bonds ionic of contrast, formation In and breaking form. the or require break state, not will bonds temperatures covalent requiring specifically which become crystals at to Molecular interactions formation. sub-unit crystal the for for enough temperatures both favourable lower of require capable structures rotation, anisotropic bond single highly and Both with contortion controlled. nanometres, diffusion in necessarily measured is molecules that process and a ions salts, of solubility the to due crystals, Introduction 5.1 ..tmgttl smc fgetitrs bu h tag hnmn htoccur that phenomena strange the about interest great of much us tell might “...it ncmlxstain.Frhroe on hti otiprati htit that applications.” is technical of important number most enormous is an that have would point a Furthermore, situations. complex in G OT OF ROWTH A PIAINO THE OF PPLICATION C OMPLEX -O 89 XIDE M ICRO N 3 ri otiigapaecontaining phase a containing grain a , NWRSVIA ANOWIRES 2 -C O 8 hsscnann c n Te and V Sc, containing phases , RUCIBLE .Fymn 1959 Feynman, R. – M S [169] ECHANISM YNTHETIC C HAPTER effectively , 5 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM becoming a stoichiometrically correct feed material for the target phase, but again this requires mobility and is typically why solid state reactions routinely require repeated re-grinding between calcinations[170]. Beyond using magnetic fields to control both crystal polymorph and morphology, growth control of ceramic crystal systems is also essential for application design. The network of bonding throughout the crystal not only forms hard and thermally stable materials but the nature of the bonding allows for precise tuning of electrical and thermal properties superior to that in molecular crystals. So much so that these complex functional materials group properties from insulators like quartz to superconductors[171]. For instance, the ability to grow single-crystal thin films of indium tin oxide allows for use in solar cells as a high transmittance, conductive layer. The epitaxial growth of a N-type material on a P-type material forms the basis for a P-N junction used throughout electrical engineering. In recent years, a number of strategies to reduce both the time and energy required to form complex metal oxides have been reported [172,173]. The introduction of a flux like Na into the reaction material is a strategy that increases the mobility of atoms in the solid state by providing a medium in which the metals are less restricted. This increases the variety of atomic interactions required, increasing the probability of the formation of the lowest energy phase at a given temperature. Other methods using bio-templating and ionic liquids are designed to maximise favourable interactions by homogeneously distributing a stoichiometric ratio of metal ions, thus dramatically reducing the need for constant regrinding and long dwell times at high temperatures. At the nanoscale, growth control of these materials becomes more specific, leading either to the use of expensive equipment making scale-up difficult[174] or requireing a large amount of scaffold templates[175], which is unsuitable for production of carbon and oxygen sensitive materials. Bio-templating in particular, although other theories exist[176], is believed to work by chelating metal ions in a regular arrangement[177]. Long chain biopolymers bind to metals holding them in a regular arrangement during heating, effectively lowering the need for ion transport through the solid. Once the carbonatious material is removed above 400 ◦C, the ions are distributed ideally to form the target phase. This method in particular has been used to synthesise high-temperature superconductors as nanowires[178], sponges[179], mesoporus structures[180], plates[181] and even a skull and crossbones[175]. In this chapter, a method for the production of complex-oxide nanowires is explored and developed as to provide both a scalable and transferable synthesis. 90 5.2. NANOWIRES 5.2 Nanowires With the physical limit of shrinking technologies being reached, resulting in a breakdown of Moore’s law, the rapid acceleration of emergent technologies seen over the past half century risks stagnation. Especially with the potential emergence of quantum computation consistently dogged with both engineering[182] and physical challenges[183], the demand for ever shrinking devices becomes increasingly important. Improving the control of growth at the nano-scale, is both inexorably linked and ideally positioned to help tackle some of these engineering problems. Nanowires are high-aspect ratio, single crystals with widths described in nm and lengths in µm often referred to as one-dimensional structures. Ubiquitous in the field of materials science, they are of interest for both their integration into this shrinking technology[184] and as curios themselves due to interesting properties as a function of their size[174] like quantum confinement. For these reasons control over the width, length and even crystallographic growth direction is important and each new method reported represents a higher degree of control available to the scientific community. 5.2.1 Nanowire growth Nanowires have been created through a range of methods both top-down, through methods like etching and lithography,[185] and bottom up[186], through methods like vapour-liquid-solid growth (VLS) or biotemplating[177]. The applicability of multiple techniques to a single material is rare with each lending a preference to a particular technique and some indeed not able to adopt any[187]. In regards to applicability, all methods must be thoroughly characterised as greater understanding of growth mechanisms are leading to the design and creation of more and more complex materials. 5.2.1.1 VLS growth Perhaps the most well-documented bottom-up growth of nanowires is the VLS mechanism. This mechanism, first proposed in 1964[188] involves droplets of liquid-phase material, either a catalytic eutectic mixture of the wire feed material and non-crystallising counter material[189] or one consisting of the crystallising material[190]. The former method receives new material through diffusion through the vapour-liquid interface as it deposits crystalline layers at the solid- liquid interface forming parallel sided single crystal nanowires. This method necessarily requires surrounding the growing droplet with the material in the vapour-phase for continued growth. If the rate of deposition is greater than the rate of diffusion, the melting point of eutectic mixture will begin to rise, changing the contact angle which alters the width of the wire until it eventually solidifies halting growth. The latter method is limited in the amount of available feed material from the start as there is no mechanism to replenish the droplet. Nanowires grown in this way 91 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM typically taper towards the end, as a result of the reduction of droplet radius due to the ever reducing volume of material. 5.2.1.2 Microcrucible growth Although the most common method of nanowire growth, VLS is not a panacea method transferable to all materials. In 2006 Hall reported nanowire growth of the superconductor YBCO using a chitosan template[178]. These nano-wires were 50 nm§5 nm in width and between 1 µm to 2 µm in length and were reported as having thinning widths and tapered ends, suggestive of growth involving a molten reservoir of material gradually depleting as the wire lengthens. In 2014 Boston et al. expanded upon this work where it was noted that using the same method of growth, some of the wires had uniform widths and ends that appeared truncated, looking much more like one would expect from VLS catalysed growth[191]. Through the use of high temperature TEM, it was observed that the uniform wire widths were in fact a product of growth out from a nano-sized, molten particle embedded in the surface of the reaction mixture (fig. 5.1). After heating to 500 ◦C, most of the organic material has been removed leaving a ‘current-bun’ motif (fig. 5.1 inset) consisting of Ba-rich, nano-sized, particles embedded in a matrix of various Y and Cu species. Upon further heating, the nano-particles move within the matrix before reaching an interface. After which, the Ba-rich particle is ‘fed’ from the surrounding phases of material as it produces a single crystal nano-wire, effectively pushed out from the surface. If the particle diameter at the surface remains consistent throughout, so does the width of the wire however, if FIGURE 5.1. Representation of the mechanism of micro-crucible nano-wire growth demonstrated via the formation of Y211. 1 shows the movement of the Ba- rich nano-particle towards a surface, 2 the influx of Y and Cu species and the beginnings of Y211 formation at the interface. 3 the resulting nano-wire. 92 5.2. NANOWIRES the matrix breaks down, the wire will of course grow thicker. This mechanism of growth, though at a different magnitude, was hypothesised by Matsubara in 1994 when observing the growth of bismuth strontium calcium copper oxide (BSCCO) micro- whiskers and dubbed ‘microcrucible’ growth[192]. This method has obvious drawbacks, limiting its general applicability even amongst complex oxide materials. Nano-particles, rich in a particular metal are by no means guaranteed to form and it may be a phenomena specific to Ba within Y and Cu. Indeed, this method has been attempted in the formation of the superconductor BSCCO so far with no success. The use of organic templates also excludes any materials sensitive to C, O or N, for example the recently discovered iron arsenide superconductors[170] which will not form at all in the presence of any excess O. The use of sodium alginate therefore must be treated as a potentially reactive additive as well as being a scaffold in which a material can grow around. 5.2.2 Superconductivity and superconducting materials Superconductors are materials that exhibit zero, detectable, electrical resistance. That is, a closed loop made from a superconducting material will exhibit a perpetual flow of electrical current, once it has been set in motion. Currently, all superconducting materials only demonstrate this property below a certain temperature, known as a superconducting transition temperature or critical temperature (Tc), intrinsic to the bulk material. Since the discovery of superconductivity in Hg at 4.19 K in 1911[193], efforts have been made to design and identify materials that exhibit this phenomena at increasingly higher temperatures. Design and discovery of superconductors with a higher Tc improves the potential applications of these zero-resistance materials, the obvious goal being that of a superconductor that requires no cooling. With the discovery of high-temperature superconductivity in BaxLa5−xCu5O5 that functions as a superconductor above 30 K by Bednorz and Müller in 1986[194], considered a Nobel Prize worthy developmental milestone insofar that a device made containing these materials could be cooled using cheap, safe and abundant cryogens, the practicality of integrating superconductors into technology increased significantly. The following decade, numerous similar complex oxide systems were found to exhibit superconductivity at or above this key temperature of 77 K[195], a temperature reachable by liquid nitrogen, with the highest Tc at the time of writing, of a single material at ambient pressure still at 133.5 K with Hg-doped BSCCO[196]. YBCO is a complex oxide first reported as a superconductor in 1987 and was the first to be [171] above this 77 K threshold with a Tc of “between 80 K and 93 K” . The acronym YBCO describes a family of compounds with a varying stoichiometry of the elements yttrium, barium, copper and oxygen and are often referred to by their ratios, for example the compound YBa2Cu3O7−x is commonly referred to as Y123. A well studied family of materials, YBCO has a number of synthetic routes, is not overly sensitive to oxygen and is stable once fabricated which make it an ideal candidate for growth templating. As mentioned in section 5.2.1.2, growth mechanisms of Y123 93 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM nanowires via biotemplating have been rigorously characterised. Due to chemical sensitivities to elements like oxygen however, biotemplating is not the solution to growing nanowires for all materials[170]. What is currently lacking in the synthesis of anisotropic high temperature superconductors however is a general, scalable method to produce abundant nanowires. In this chapter, such a synthesis is presented and the general applicability shown. Throughout this chapter, starting refinement parameters for phases present can be found in appendix A.4. 5.3 Results and discussion 5.3.1 Bio-templated nano-wire growth For the sake of comparison, growth of YBCO nano-wires using sodium alginate was carried out as mentioned in section 2.2.12 using the precursors Y(NO3)3·6H2O, Cu(NO3)3·2.5H2O and Ba(NO3)2 in a stoichiometric ratio for Y123( 0.1915 g, 0.3490 g and 0.2610 g, respectively). The resulting powder was calcined initially at 500 ◦C for 2 h before being reground by hand and calcined again at 920 ◦C for 2 h. The pXRD pattern and SEM micrographs of the resulting black powder are shown in figure 5.2. The reflections present suggest the formation of Y123 as the sole YBCO phase containing small amounts of unreacted CuO and BaCuO2. The broadened peak width is suggestive of nano- structuring. SEM analysis shows nano-wires of width 40 nm to 110 nm and lengths up to 1.5 µm covering most of the surface of the powder. Interestingly a higher density of wires are observed within crevices in the pellet surface, that is, wires are more likely to be found growing on surfaces facing other surfaces. Although surfaces that were facing open space did have small nodules which could indicate the remains of wires that have grown and since melted. As there is unlikely to be any significant variance of temperature in the furnace, it is probable that a difference in oxygen content at the powder-air interface would cause the two environments to grow wires at different times during the synthesis. High surface area structures are likely to suffer oxygen depletion from surface layers, transforming the stability of the material, in this case lowering the melting point. Elemental composition of the wires was determined as Y=13.22 %, Ba=41.48 %, Cu=28.22 % and O=17.08 % by mass, using single point EDX. These values are all within 1.6 % of the expected values for Y123. 5.3.2 YBCO growth By means of a control regarding the synthesis of YBCO via a solid state reaction, oxides of yttrium (Y2O3), copper (CuO) and barium carbonate (BaCO3) were hand ground and pressed into a pellet as mentioned in section 2.2.11 and heated to 920 ◦C for 2 h. These precursors were chosen based on the previously reported phases present during nanowire growth[191] and mixed 94 5.3. RESULTS AND DISCUSSION (a) (b) (c) FIGURE 5.2. (a)pXRD showing Y123 as a major phase (orange circles) with minor phases CuO (green stars) and BaCuO2 (red squares) and (b) and (c)SEM of YBCO, grown using sodium alginate. in a stoichiometric ratio to form Y123. Figure 5.3 shows a pXRD pattern and SEM micrograph of this pellet after a single 2 h hold at 920 ◦C. Figure 5.3 (a) shows phase analysis via multi-phase refinement of the powder diffraction pattern of the solid state reaction of Y2O3, CuO and BaCO3. Specifically, the black and light blue lines represent the experimental data and calculated fitting respectively with a Rwp of 2.76%. Detectable phases imply a largely incomplete reaction, showing a relatively small proportion of Y123 and Y211 phases, 37.8 % and 8.0 % respectively. Other detectable crystalline phases are Y2Cu2O5 = 1.3 %, BaCuO2 = 11.2 %,Y2O3 = 3.8 %, BaCO3 = 16.8 % and CuO = 20.8 %. As solid state reactions for the formation of complex oxides are known to take a significant time with >50 h in a furnace being common, the detection of unreacted precursor materials and binary oxide minor phases is expected. SEM analysis of the pellet surface shows a granular morphology 95 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) FIGURE 5.3. (a) Analysis of a pXRD pattern of YBCO precursors after a reaction time of 2 h at 920 ◦C. Tick marks under the baseline show reflection positions of target phases used to refine the pattern and colour coded as per the figure legend. Rwp=2.76%. (b) and (c) show the surface of the pellet at different magnifications. 96 5.3. RESULTS AND DISCUSSION with no visible anisotropic features. 5.3.3 Non-templated micro-crucible growth In order to head towards a non-biotemplated synthetic route to YBCO nanowires, BaCO3 nano- particles were synthesised as stated in section 2.2.10 and used in place of the standard hand- ground BaCO3 in an attempt to mimic the ‘current bun’ phase of the alginate bio-templated nano-wire growth. Analysis of the BaCO3 nanoparticles can be seen in figure 5.4. FIGURE 5.4. Analysis of BaCO3 nanoparticles synthesised via sol-gel combustion method. pXRD showing phase pure BaCO3 ‘witherite’ structure. Insets show TEM micrographs of the nanoparticles As this system does not need the initial 2 h hold at 500 ◦C for nano-particles to form in a matrix of various Y and Cu phases, once pressed, the pellet was calcined only at 920 ◦C with a heating rate of 5 ◦Cmin−1 for 2 h. After 2 h, the pellet was allowed to cool to room temperature. Figure 5.5 shows a pXRD pattern of the subsequent material. Multiphase Rietveld analysis of the resultant material showed detectable crystalline phases of Y123=44.1 %, Y211=0.2 %,Y2Cu2O5 = 10.6 %, BaCuO2 = 1.3 %,Y2O3 = 1.5 %, BaCO3 = 21.4 % and CuO = 20.7 %. The introduction of BaCO3 nano-powder has increased the relative amount of Y123, from 37.8 % to 44.1 % which is to be expected due to the larger surface area of starting materials available for reaction. There is a large difference in the amount of Y211, which has dropped to 0.2 % from 8.0 %, implying that a smaller BaCO3 particles favour the formation 97 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) FIGURE 5.5. (a) Analysis of a pXRD pattern of BaCO3 np seeded YBCO precursors after a reaction time of 2 h at 920 ◦C. (b) and (c) show SEM micrographs of the observed anisotropic structures at different magnifications. 98 5.4. THE USE OF SODIUM AS A FLUX of Y123 over Y211 possibly due to the higher surface area restricting oxygen from thermal decomposition escaping without reacting. The fraction of CuO appears to be the only detectable crystalline phase that is consistent between the two samples at 20.7 % down from 20.8 %. Phases that decreased in volume between the two samples were BaCuO2 and Y2O3 decreasing by 9.9 % and 2.3 % respectively. Interestingly both BaCO3 and Y2Cu2O5 had a larger presence in the nanoparticle-containing reaction pellet. Y2Cu2O5 is 6.8 % more abundant and BaCO3 is increased by 4.6 %. In order to inspect the system for the presence of possible wires, the sample was analysed using SEM (fig. 5.5). As can be seen in figure 5.5, distinct regions of rod-like structures are scattered across the surface. Although their growth direction does not appear to follow any particular pattern regarding the powder surfaces, they do indeed appear to be affected by their neighbour’s growth direction and found in clusters. These anisotropic structures measure between 87.2 nm to 474.2 nm in width, with a mean of 237.9 nm, which is, at the smallest end of the distribution, at least twice the width of the nanoparticles used for seeding. The rods are clearly different from the wires seen after using an alginate bio-template as they have a lower aspect ratio, are thicker and have a more uniform growth direction. Attempts at relocating these structures onto a TEM grid for phase analysis using SAED were repeatedly unsuccessful. 5.4 The use of sodium as a flux 5.4.1 Using additives as a flux In metallurgical terms, a ‘flux’ is an additive used to perform one or more of a variety of tasks, most commonly to purify, clean or reduce the effective melting point of a metal based system. Used commonly to affect metal joining, fluxes are used to impede the formation of and to dissolve metal oxides typically detrimental to the joining process and to allow the metals to flow with less beading, generally caused by oxide layers formed at elevated temperatures. During the industrial Hall-Héroult process, smelting of aluminium from bauxite the ore is facilitated by mineral Cryolite (Na3AlF6). The aluminium oxide is dissolved in molten cryolite which drastically lowers the temperature at which the aluminium can undergo electrolysis. During this process, the ratio of NaF to AlF3 is adjusted in order to further reduce the melting point of the system and extra AlF3 is added as a flux. During the bio-templating of complex oxides, the templates have commonly contained alkali- metals, both Na[197] and K[198]. Although not previously thought of as required for the formation of nanowires through bio-templating, Na has not only been present but its requirement for metal [197] oxide anisotropy has been speculated on . As a source of sodium, Na2CO3 or NaCl were added to the reaction mixture prior to being pressed into a pellet and heated for 2 h regardless of Na content or isotherm temperature. In order to find the optimum amount of sodium and 99 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM temperature for YBCO formation and wire growth, various amounts of sodium and calcination temperatures were used. In contrast to a standard quench experiment where a reaction is stopped during the ramp, each following reaction was left at the reported temperature to react for 2 h in order to search for wire formation. The following show a comparison of the data collected from these experiments. All numbers showing the relative amounts calculated via multiphase Rietveld analysis, displayed graphically in figures 5.6(d), 5.6(c), 5.14(d), 5.14(c), 5.22(d), 5.22(c), 5.30(d) and 5.30(c), can be found in tables A.5 and A.6. 5.4.1.1 2 % Na·X After 2 h at 920 ◦C the sample containing 2 % NaCl has a similar fraction of Y123 than that of the BaCO3 nanoparticle seeded sample at 45 %. However there is a drastic increase in the fraction of Y211 compared to the same sample, from 0.2 % to 19.1 % bringing the total YBCO content up to 64.1 %. Notably, there is a lower fraction of BaCO3 and a higher fraction of BaCuO2 implying a higher amount of conversion of the Ba starting material into Y211 and BaCuO2, a mixture [169] known to be precursors to Y123 . The Na2CO3 analogue has a higher fraction of Y123 and lower fractions of Y211 and BaCuO2 suggesting that the release of oxygen upon breakdown of the CO3 is promoting the formation of the more oxygen rich Y123. At 2 % Na·X content Y123 becomes the major phase at the higher temperatures, NaCl ◦ ◦ ◦ ◦ ◦ between 860 C to 880 C and Na2CO3 between 820 C to 840 C, approximately 40 C cooler. CuO is present at all temperatures, appearing to be the most resilient of the starting materials with 23.8 % and 16.9 % total volume fraction after 2 h at 920 ◦C. In the NaCl mixture there is an increase of both Y123 and Y211 as a function of calcination temperature, with a small drop in ◦ Y211 from 20.9 % to 19.1 % within the last 20 C. In contrast, the Na2CO3 mixture reaches its maximum volume fraction of Y211 with 20.3 % at 860 ◦C, with a fairly linear decrease over the last 60 ◦C to 14.9 % at 920 ◦C. In both cases, a decrease in volume fraction of Y211 appears to coincide with an increase in Y123 implying a conversion from one 211 to 123. In both cases, ◦ BaCO3 is still present at 800 C but remains in higher concentrations for longer in the Na2CO3 mixture. This is likely a result of a higher amount of CO2 generation as a result of Na2CO3 decomposition. If CO2 plays a role in reactions between species then the extra provided by the Na2CO3 would mean less is required from the BaCO3. In this case, it would appear that the total amount of amorphous Y, Ba and Na containing material has increased due to the total amount ◦ of CuO present at 800 C. The relative amount of Y2O3 is similar between both mixtures but ◦ ◦ decreases to noise by 880 C in the Na2CO3 mixture, 20 C lower than the NaCl mixture. BaCuO2 forms regardless of the Na species used and decreases as the temperature increases implying direct conversion to YBCO at higher temperatures. In both cases, no Y2Cu2O5 was seen at any temperature above the detection limit of pXRD but has been kept in for comparative purposes. 100 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.6. Comparison of the solid state reaction of YBCO precursors calcined at various temperatures. All reactions contain 2 % Na by mass as either NaCl ((a) and (c)) or Na2CO3 ((b) and (d)). (a) and (b) display pXRD patterns calcined for 2 h at the labelled temperature. (b) and (d) show the relative amounts of phases in each powder pattern calculated using multi-phase Rietveld refinement. The colours for each phase (Y123, Y211,Y 2Cu2O5, BaCuO2 ,Y2O3 ,BaCO3, CuO) are consistent across all four graphs. 101 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.7. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 800 C for 2 h. At 2 % Na·X, Y123 is the dominant crystalline phase when left to calcine at 920 ◦C for 2 h. However, Y123 formation appears to be favoured by the addition of Na2CO3 over NaCl. When compared to a solid state reaction without the addition of sodium, the total fraction of YBCO (both Y123 and Y211) formed after 2 h at 920 ◦C is increased by 19.8%. This implies a lowering of reaction temperature when using Na·X, an effect that is amplified using a carbonate over a chloride. In order to track the evolution of growth of any nanowires, SEM micrographs of each, post- calcine, reaction mixture were carried out. At 2 % NaCl, as expected from a mix of crystalline phases, a mix of morphologies were present and no wires or evidence of wires were visible (fig. 5.7). Small regions of layered, plate-like structures were observed on the same order of magnitude as nanowires but do not appear to be any kind of deformation or merging of wires that had perhaps formed and subsequently melted [fig. 5.7 (b)]. In the analogous Na2CO3 sample, the 102 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.8. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 820 C for 2 h. same mixed morphology due to a mixture of precursor phases is visible. In isolated pockets of the Na2CO3 sample, wires were found and are shown in figures 5.7 (c) and (d). The wires have a mean diameter of 172.9 nm, ln standard deviation of 0.339 nm and up to 12 µm in length, have parallel sides and truncated ends, implying microcrucible growth (MCG). The identity of the wires was undetermined due to their scarcity. At 820 ◦C, the NaCl sample looks much as it did at 800 ◦C [figures 5.8 (a) and (b)] with a mixture of seemingly isotropic structures of different morphologies. In contrast to this, the Na2CO3 sample from the same temperature shows a greater abundance of anisotropic structures as can be seen in figures 5.8 (c) and (d). As with the nanowires seen at 800 ◦C, the wires appear to have parallel sides and truncated ends, as would be expected of MCG. On this sample however, a slightly larger size distribution of wires were observed with a similar mean diameter of 179.0 nm but with a larger ln standard deviation of 0.493 nm and lengths of up to 40 µm. Although with 103 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.9. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 840 C for 2 h. a larger sample size due the larger abundance, a broadening of the size distribution would be expected, there is evidence of wire merging due to the presence of stepped ends as can be seen on the central wire in figure 5.8 (d). EDX of small regions of these low temperature wires however, indicate a significant lack of yttrium, suggesting that, along with the pXRD these wires are likely BaCuO2. At 840 ◦C the morphologies for both Na systems begin to resemble each other and are more homogeneous when magnified 1000 times. Under higher magnifications however, the NaCl mixture appears slightly more amorphous in texture than the Na2CO3 analogue, with rounded edges 1 µm to 6 µm in size. The higher magnification of the Na2CO3 mixture reveals more crystalline looking particles with the remains of nanowires evident as can be seen in figure 5.9 (d). As all complex phases in the system undergo incongruent melting[169,199,200] at atmospheric pressure, the melting of the initial wires cannot be said to be indicative of any particular phase. 104 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.10. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 860 C for 2 h. In both cases, there are still high amounts of crystalline CuO present in the mixture but in the sample containing Na2CO3, Y123 has already become the dominant phase. The featureless morphology of the NaCl mixture is continued through 860 ◦C. Particle size ◦ has increased possibly due to further reaction and sintering. The Na2CO3 mixture at 860 C has begun to show signs of a second ‘wave’ of nanowire growth. Interestingly, these wires appear thinner in diameter with a mean of 109.6 nm, are fairly uniform in distribution (ln σSD=0.310) and this time only up to 1 µm in length. Based on the pXRD data, any new material forming at this temperature is likely YBCO, either Y211 or Y123. At 880 ◦C both 2 % mixtures produce Y123 as the majority phase however morphological differences between the samples are again becoming more apparent. The particle size in the NaCl mixture continues to increase whilst keeping an isotropic, granular appearance. The mixture containing Na2CO3 has continued to grow nanowires, although not ubiquitously. Small regions 105 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.11. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 880 C for 2 h. of the sample surface have pockets of wires as can be seen in figure 5.11 (d). These wires have parallel sides, truncated ends and a mean width of 178.4 nm, surprisingly close to that of the ◦ same sample at 820 C but with a tighter distribution (ln σSD=0.376) and remain shorter than those at lower temperatures with lengths of up to 4 µm. Stepped ends can also be seen, again implying a merging of adjacent wires during growth. The rest of the sample is granular with particle sizes smaller and with a lower dispersity than the NaCl sample at the same temperature. By 900 ◦C, both 2 % samples are showing the presence of nanowires. This is the first sign of nanowires in the NaCl sample and are thin, practically below the detection limit of the microscope, with a mean of 76.5 nm and short with wires longer than 1 µm a rarity. Although these structures appear to have parallel sides the length of the wire, the resolution of the microscope prevent conformation of this. These wires were also in isolated pockets and by no means covering more than ≈5 % to 7 % of the surface. The areas surrounding the wires have nano-sized texturing that 106 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.12. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 900 C for 2 h. was not present at any lower temperature implying either the beginnings of wire growth or the remains of where wires have melted or reacted. Evidence of wires on the Na2CO3 mixture is shown by what appear to be melted or reacted structures as no complete clusters of nanowires were observed although the structures are similar in size to the NaCl analogue with a mean width of 77.2 nm, ≈ 100 nm thinner than that observed 20 ◦C lower. On a wider scale, the general morphology of the sample now resembles Y123 as formed by a standard solid state reaction[201]. After 2 h at 920 ◦C the wires in the NaCl sample have persisted [fig. 5.13 (b)], although at only 46 % Y123 at this temperature, the actual stoichiometry of the wires cannot be determined via pXRD and SEM alone. The wires remain thin at 98.6 nm but can be found with greater lengths of upto 4 µm implying a greater rate of growth at the higher temperature. As can be seen in figure 5.13 (c), the wires are found on the surface in small balls of ≈ 1 µm in diameter. ◦ The Na2CO3 sample at 920 C has become much more uniform in morphology, likely Y123 based 107 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.13. SEM micrographs of YBCO precursors containing 2 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 920 C for 2 h. on 64 % volume fraction in the mixture and similarity to the reported YBCO morphology[201]. Some very isolated pockets of nano texturing can be found as seen in figure 5.13 (d) but no clear evidence of nanowires can be found. At 2 % Na·X content between 800 ◦C to 920 ◦C nanowires are seen at various stages. Using NaCl as an additive, the first sign of any anisotropic, high aspect ratio growth is not until 900 ◦C but they persist after 2 h at 920 ◦C. The wires grown using NaCl at these temperatures are too thin to confirm whether or not they have truncated ends on SEM and due to their sparsity on the surface, no wires were able to be found when transferring the sample to a TEM grid. This also made crystallographic identification difficult. Apart from that, the NaCl sample at 2 % retained an unremarkable morphology as might be expected from a solid state reaction. In contrast to this, the sample with Na2CO3 as an additive had wires or evidence of wires visible at all temperatures tested. Localised to small areas across the sample surface the wires present at 800 ◦C and 820 ◦C 108 5.4. THE USE OF SODIUM AS A FLUX Table 5.1: Numerical analysis of observed nanowire diameters of samples containing 2% Na content obtained through image analysis of SEM micrographs from figures 5.7- 5.13. Absent numbers indicate samples that had no observable nanowires. Count indicates the total number of wires measured in the image, mean is calculated using a log-normal distribution fitting of the data and lnσSD is the log of the standard deviation. 2% NaCl Na2CO3 Mean Mean Temperature (◦C) Count ln σ (nm) Count ln σ (nm) diameter (nm) SD diameter (nm) SD 800 - - - 16 172 0.33 820 - - - 32 178 0.49 840 - - - 10 105 0.43 860 - - - 79 109 0.31 880 - - - 79 178 0.37 900 101 76 0.25 60 77 0.24 920 69 98 0.19 17 87 0.29 appeared to have both tapered and truncated ends, suggestive of both VLS and MCG mechanisms. These initial wires also have a lack of yttrium as indicated by EDX and have almost vanished by ◦ 840 C which implies that these initial wires are in fact BaCuO2, a phase which not only forms at temperatures lower than YBCO, but can form a eutectic mixture with CuO, which can facilitate the formation of Y123 from Y211[169]. At 860 ◦C more wires begin to appear, again in isolated regions that have taken the form of straight nanowires with parallel sides and truncated ends by 880 ◦C. EDX analysis of these wires gives an elemental stoichiometry of Y=1.00, Ba=1.91 and Cu=2.63 when given as a ratio compared to Y, which suggests that the wires are in fact Y123. Unfortunately, due to a low number of these sites, efforts to get these wires onto TEM grids were unsuccessful. By 920 ◦C these wires had all but gone leaving a fairly consistent bulk morphology. Although it is possible that such small structures could suffer oxygen depletion at these elevated temperatures and erode, it is also possible that due to the small coverage of wires across the surface, they were just overlooked despite searching. The larger structure of the Na2CO3 sample maintained a more angular particle shape throughout all temperatures. 5.4.1.2 5 % Na·X At 5 % Na·X as an additive, both NaCl and Na2CO3 result in a majoritivly Y123 mixture at 920 ◦C as seen at 2 % however the NaCl sample sees an increase in Y123 formed, from 45.0 % to 74.5 % but the Na2CO3 analogue remains fairly constant from 65.7 % at 2 % to 65.5 % at 5 %. Both samples increase drastically in their Y123 percentage over the last 20 ◦C. In the NaCl sample at 5 %, the ultimate volume fraction of all the other phases remains fairly constant once at 920 ◦C apart from CuO which is down to 10.2 % from 23.8 %. Indicating that the higher amount of NaCl is facilitating the uptake of CuO by the Y211 to form Y123. CuO is again present at all temperatures, however in contrast to the 2 % sample, the reduction of CuO volume fraction is significantly greater at each temperature. The formation of BaCuO2 109 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.14. Comparison of the solid state reaction of YBCO precursors calcined at various temperatures. All reactions contain 5 % Na by mass as either NaCl ((a) and (c)) or Na2CO3 ((b) and (d)). (a) and (b) display pXRD patterns calcined for 2 h at the labelled temperature. (b) and (d) show the relative amounts of phases in each powder pattern calculated using multi-phase Rietveld refinement. The colours for each phase (Y123, Y211,Y 2Cu2O5, BaCuO2 ,Y2O3 ,BaCO3, CuO) are consistent across all four graphs. 110 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.15. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 800 C for 2 h. peaks at 840 ◦C up to 25.0 % before decreasing consistently towards maximum temperature. Y211 ◦ ◦ fraction increases at the higher temperatures, 860 C to 900 C, as the BaCuO2 begins to decline. All other phases appear to decrease steadily with temperature apart from a small dip in Y2Cu2O5 ◦ volume fraction at 860 C coinciding with the formation of Y211. Similarly, the Na2CO3 mixture reaches a maximum volume fraction of Y211 of 12.7 % at 880 ◦C, with a decrease over the last 40 ◦C to 5.3 % at 920 ◦C. Again, a decrease in volume fraction of Y211 coincides with an increase in Y123 fraction. Regarding phase composition of the 5 % Na2CO3 sample as a function of temperature, the story is far more complicated when compared to the 5 % NaCl sample. Surprisingly, there is a higher fraction of Y123 at 820 ◦C than any other temperature except from 920 ◦C. 840 ◦C to 920 ◦C show an increase in Y123 as temperature increases but 800 ◦C and 820 ◦C contain a higher amount than one would expect when comparing it to the 2 % sample. Both of these lower 111 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.16. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 820 C for 2 h. temperature experiments appear to be inconsistent to trends seen in the other samples regarding Y123, BaCO3 and CuO. It is possible that, as the volume fraction data has been collected via pXRD, that a large amorphous content is skewing the data at these temperatures, however this is difficult to quantify. Unlike the 2 % samples, in both cases Y2Cu2O5 is present at all temperatures. The higher Na content appears to facilitate the formation of this intermediate phase at the lower temperatures which slowly decreases in volume fraction as the temperature increases. At 800 ◦C both 5 % samples are mixtures of various phases however some of these phases have manifest as anisotropic structures. These structures are rare on the surface and appear to grow in clusters. The measured mean width of these structures is 168.3 nm (ln σSD=0.528 nm) and 276.6 nm (ln σSD=0.477 nm) for NaCl and Na2CO3 respectively, however with such a small sample size a rigorous comparison cannot be made. Wires from both samples show signs of 112 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.17. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 840 C for 2 h. The small spheres along nanowires in (a) and (b) are silver particles due to a sputter coater equipment issue. stepped ends indicative of a merging of thinner structures, of which are visible in figures 5.15 (a) and (b) most of which appear to have parallel sides. The wider morphology of the sample shows no remarkable features and looks as would be expected from a solid state reaction. After 2 h at 820 ◦C both samples are almost indistinguishable via SEM. Both samples have a granular appearance with a thin scattering of wires across the surface more abundant than ◦ that seen at 800 C. There is a wider distribution of wire widths, ln σSD=0.776 nm and ln σSD=0.693 nm and a higher mean of 250.4 nm and 345.7 nm for NaCl and Na2CO3 respectively. Interestingly, the central wire in figure 5.16 (d) clearly shows how the larger structures are forming with a thinner wire disconnecting from a larger structure and one approximately the same width visible from the end. Presumably, the thinner wire had grown ≈6 µm before the amorphous base had joined with others and grew as one. This is strong evidence that the wires 113 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.18. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 860 C for 2 h. are a product of the MCG mechanism as growth from an end-suspended droplet cannot create such a structure. If the separation of the smaller wire were post growth, the thin tip of the wire would be unlikely to have the shoulders visible in the micrograph. At 840 ◦C the differences between the two samples becomes more apparent. The sample containing 5 % NaCl has substantial coverage of thin nanowires with a mean of 95.2 nm with ln σ=0.400 nm. In stark contrast to this, figure 5.17 (d) shows that the Na2CO3 sample contains rod- like structures unlike any wires seen up until this point in the Na·X samples and instead appears to resemble the nanoparticle seeded mixtures without Na (fig. 5.5). These rod-like structures are not seen in the NaCl sample and the thin wires seen in figure 5.17 (b) are not found in the Na2CO3 sample. These rod-like structures have a mean width of 539.4 nm and ln σ=0.515 nm. In contrast to the samples heated to 840 ◦C, both of the 860 ◦C samples have no examples of thin, high aspect-ratio wires. The samples containing NaCl have evidence of only short rod- 114 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.19. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 880 C for 2 h. like structures with parallel sides and truncated ends shown in figure 5.18 (d). These structures appear to be growing from a single point in multiple directions, reminiscent of a geological acicular habit. It is difficult to know whether these structures are the beginnings or the remnants of larger structures but the jagged edges imply the latter. There is evidence of where other rods may have been across the surface in the form of small nodules of a similar diameter. The small sample of wires that could be measured have a mean width of 281.2 nm and a ln σSD of 0.396 nm. The ◦ Na2CO3 sample resembles the same sample at 840 C more. The rod-like structures are present again but in greater numbers, have a smaller mean width of 476.9 nm and smaller distribution with ln σSD=0.303 nm. Unlike the NaCl sample at the same temperature, these structures do not appear to be decomposing along the edges or growing from a single point in various directions, but in the same direction across a wider area. In the few occasions in which an end is visible, square and hexagonal morphologies can be seen which are both indicators of an orthorhombic 115 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.20. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 900 C for 2 h. crystal structure which all of the expected intermediate phases adopt. By 880 ◦C neither of the samples show any evidence of the rod-like structures and are covered in thinner, high aspect ratio wires. Both samples have a dense covering of wires with mean widths 114.0 nm and 144.1 nm for NaCl and Na2CO3 respectively. The 5 % sample appears to have wires over a higher percentage of its surface as can be seen in figures 5.19 (a) and (b). Wires of the same appearance are also observed at 900 ◦C on both samples, again with the Na2CO3 having a wider coverage. The mean widths are 163.4 nm and 200.6 nm for NaCl and Na2CO3 respectively which happen to be an increase in width of 41 % §2 % for both from the previous temperature. This could be a merging of multiple ‘crucibles’ or micro-crucible creep; a phenomena where the wire widens as a result of the the liquid-solid interface broadening over time[191]. This could also be a function of the extra time in which the system took to get to temperature or, more likely, the higher temperature itself could have exacerbated this process 116 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.21. SEM micrographs of YBCO precursors containing 5 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 920 C for 2 h. resulting in wider wires. If these wires are the same material, it would follow that the two systems would see the same amount of creep across the same temperatures. At the final temperature of 920 ◦C again, both samples have easily observable wires. The ◦ trend of expansion is continued with the wires on the 5 % Na2CO3 sample at 920 C with a further width increase of 46 % to a mean of 293.4 nm but with a larger distribution (ln σ=0.627 nm). In stark contrast to this, wires in the sample containing 5 % NaCl have a remarkably different size visible as thin hair-like structures in figure 5.21 (b) with a mean width of 71.6 nm. These wires are often curved and some appear to exceed 2 µm in length. There is no clear evidence that anisotropic structures of any other size existed previous to those present. At 5 % Na·X content between 800 ◦C to 920 ◦C anisotropic structures are seen at all temperatures and appear to be in stages. The initial stage at 800 ◦C shows thin wires scattered sparingly across the surface of both samples that increase in number but appear similar in 117 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM Table 5.2: Numerical analysis of observed nanowire diameters of samples containing 5% Na content obtained through image analysis of SEM micrographs from figures 5.15- 5.21. Absent numbers indicate samples that had no observable nanowires. Count indicates the total number of wires measured in the image, mean is calculated using a log-normal distribution fitting of the data and lnσSD is the log of the standard deviation. 5% NaCl Na2CO3 Mean Mean Temperature (◦C) Count ln σ (nm) Count ln σ (nm) diameter (nm) SD diameter (nm) SD 800 22 168 0.52 14 276 0.47 820 15 250 0.77 9 345 0.69 840 82 95 0.40 32 539 0.51 860 36 281 0.39 88 476 0.30 880 173 113 0.31 133 144 0.39 900 57 163 0.28 91 200 0.46 920 130 71 0.23 67 293 0.62 morphology when taken to 820 ◦C these grow more abundant still for the NaCl sample up to 840 ◦C. Stage two involves formation of rod-like structures that are much wider than the initial wires with a lower aspect-ratio and ends typical of orthorhombic single-crystals. These structures ◦ ◦ ◦ are visible in Na2CO3 at the temperatures 840 C and 860 C and NaCl at 860 C but are fewer in number. The third stage shows another wave of thinner, high aspect-ratio wires abundant across the surface of both samples. These wires get more abundant and also thicken by 41 % between ◦ ◦ ◦ 880 C and 900 C. For the Na2CO3 sample, this is the final stage as at 920 C the wires are so abundant, it is not possible to see the surface from which they are growing. At each temperature of this stage, it is worth noting that the standard deviation increases significantly, implying that as well as the creep-thickening of existing wires, new ones are also being created. For NaCl, there is a fourth stage that results in the formation of the thinner, scattered wires that do not resemble any previously observed. An attempt to identify the wires at all stages of the 5 % Na·X content will be made in section 5.4.2. 5.4.1.3 10 % Na·X When containing 10 % Na·X as an additive, both the NaCl and Na2CO3 samples contain a majority of Y123 at every temperature tested with a total volume fraction at 920 ◦C of 89.2 % and 80.1 % respectively. For the NaCl sample, all phases decrease as temperature increases with the exceptions of Y123 which increases drastically, Y2Cu2O5, which remains fairly consistent around 14 % between the temperatures of 840 ◦C to 900 ◦C before decreasing at 920 ◦C and Y211 which has a detectable presence at 840 ◦C, increases steadily to 9.0 % at 880 ◦C before declining at the higher temperatures. The sharpest increase in Y123 fraction is seen between 880 ◦C and 900 ◦C and coincides with a sharp decrease in Y211, BaCuO2 and BaCO3. 118 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.22. Comparison of the solid state reaction of YBCO precursors calcined at various temperatures. All reactions contain 10 % Na by mass as either NaCl ((a) and (c)) or Na2CO3 ((b) and (d)). (a) and (b) display pXRD patterns calcined for 2 h at the labelled temperature. (b) and (d) show the relative amounts of phases in each powder pattern calculated using multi-phase Rietveld refinement. The colours for each phase (Y123, Y211,Y 2Cu2O5, BaCuO2 ,Y2O3 ,BaCO3, CuO) are consistent across all four graphs. 119 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.23. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 800 C for 2 h. The 10 % Na2CO3 sample shows much the same trend as the NaCl, with a few notable differences. The amount of Y123 observed at 840 ◦C is more than the trend would have one expect, interestingly as all the other phases are decreasing with increasing temperature. At 860 ◦C both Y211 and Y2Cu2O5 coincide with a decrease in Y123. Unlike the NaCl sample, the Y211 volume fraction is only first detectable at this temperature however, there is significantly more with 18.6 % compared to 5.8 %. 880 ◦C has the highest fraction of Y211 for both samples but again, ◦ ◦ the Na2CO3 mixture has significantly more. It appears that between 900 C and 920 C there is little to no difference between the volume fractions of the phases, showing that the reaction is complete, barring any further grinding and mixing, by 900 ◦C. At 800 ◦C there is already a visible difference between the samples. The 10 % NaCl sample has only rare occurrences of wires locatable. These wires however appear to grow from a specific site, as has been seen in some of the lower Na content samples at the same temperature. The 120 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.24. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 820 C for 2 h. wires that were found have a mean width of 212.0 nm and can be seen in figure 5.23 (b). The Na2CO3 sample however has a significantly higher number of wires on the surface, these have a similar width (mean width of 200.9 nm) but a higher distribution as to be expected from a larger sample size. It is difficult to say whether the wires are growing in bunches, as per the NaCl sample but on careful inspection, is does appear so. This sample has a larger number of wires present at 800 ◦C than any sample so far. By 820 ◦C, as the previous experiments would lead to believe, there are a larger number of ◦ wires that resemble those grown at At 800 C. Both NaCl and Na2CO3 samples have increased in number but interestingly, the wires no longer appear to be growing in bunches. This may be a function of the fact that the higher temperature has generated more of the amorphous pre-cursor necessary for wire growth, giving the surface more opportunity for growth to occur. Both figures 5.24 (b) and (d) show clear examples of <60 nm wires that have grown multiple µm long, next to 121 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.25. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 840 C for 2 h. thicker wires. When the samples have been held at 840 ◦C for 2 h they take on quite a different appearance. The NaCl sample clearly appears to be covered in anisotropic nanostructures giving the greater morphology a ‘furry’ appearance. These wires have a mean width of 94.0 nm but have a relatively high distribution (ln σSD=0.447 nm) however they do not resemble the wires from the cooler two temperatures and are much thinner. In contrast the Na2CO3 sample is practically featureless and has absolutely no wires, nor obvious evidence of any previous structures, anywhere on the surface. The sample has taken on a granular appearance typical of solid state reactions. With 5 % Na content, at the same temperature, the sample was showing rod-like structures which are nowhere to be seen here. At this temperature the Na2CO3 sample has yet to form any Y211 which is likely a product of the excess oxygen available to reaction during the break down at this temperature, of Na2CO3 forming the more oxygenated Y123. 122 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.26. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 860 C for 2 h. At 860 ◦C the two are still clearly different. The NaCl sample now has significant nanowire coverage over the majority of the sample surface. These wires have a mean width 131.8 nm and again do not appear to be growing in bunches as seen at 800 ◦C which implies a more uniform distribution of the material necessary for wire growth. At this temperature, other than Y123 both samples (though the Na2CO3 more than NaCl) have an increase of Y211 and Y2Cu2O5. The Na2CO3 sample has a similar appearance to the previous temperature, a granular, featureless morphology with the exception of small, rare pockets of very thin wires (mean width of 99.9 nm). These wires were difficult to find and definitely not representative of the entire sample, similar to most wires found at 840 ◦C. All of the wires that were found were within cracks and crevices across the sample. None were found pointing away from the sample completely. At 880 ◦C,wires are again present in both samples. Although the surface of nanowire coverage in the NaCl sample has decreased, they appear shorter and restricted more to cracks and crevices. 123 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.27. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 880 C for 2 h. The Na2CO3 sample has increased in the number of wires visible however they have the thinnest mean of any of the nanowires measured during this experiment of 63.3 nm. Some of these wires are barely visible even under 6000 times magnification as shown in figure 5.27 (d) but can be seen bending and curling back on them selves, most likely a consequence of non-uniform oxygen distribution across the width of the wire. The highest temperature appearance of nanowires in the 10 % Na·X samples is at 900 ◦C. Both samples appear to have similar sized wires but due to the high abundance of thin wires in the NaCl sample, the mean widths are significantly different being 84.5 nm and 111.3 nm for NaCl and Na2CO3 respectively. Both samples appear similar, which is to be expected as they are both primarily Y123 by this point but with the NaCl sample has a higher coverage of nanowires. Again, wires pointing out into ‘open space’ appear to be rare. At 920 ◦C, both of the samples are blocky and featureless resembling standard Y123 made via 124 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.28. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 900 C for 2 h. other solid state means. This is a stark contrast to the same temperature with only 5 % Na·X which have a thick coverage of structures. At 10 % Na·X content between 800 ◦C to 920 ◦C there are noticeably fewer nanowires throughout. From the lower temperature experiments the NaCl sample gradually increases in number and coverage of nanowires peaking in number by 860 ◦C. After which, the numbers decline towards 920 ◦C by which point they are completely absent. The 10 % Na2CO3 sample has a noticeably different reaction across the temperatures. Although a relatively high number of wires are seen at the lower temperatures, at 840 ◦C there are no visible wires at all which is a stark contrast to the NaCl analogue, an interesting observation considering the presence of all materials required for nanowire growth, exemplified 125 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.29. SEM micrographs of YBCO precursors containing 10 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 920 C for 2 h. by the emergence of nanowires at the higher temperatures. Nanowires at the higher temperatures are far more rare respective to other experiments carried out at the same temperatures even when at their most abundant at 900 ◦C. As with the NaCl sample, all features resembling nanowires are absent once reacted at 920 ◦C for 2 h and the morphology is typical of Y123. 126 5.4. THE USE OF SODIUM AS A FLUX Table 5.3: Numerical analysis of observed nanowire diameters of samples containing 10% Na content obtained through image analysis of SEM micrographs from figures 5.23- 5.29. Absent numbers indicate samples that had no observable nanowires. Count indicates the total number of wires measured in the image, mean is calculated using a log-normal distribution fitting of the data and lnσSD is the log of the standard deviation. 10% NaCl Na2CO3 Mean Mean Temperature (◦C) Count ln σ (nm) Count ln σ (nm) diameter (nm) SD diameter (nm) SD 800 12 212 0.33 69 200 0.51 820 7 221 0.54 24 289 0.70 840 124 94 0.44 - - - 860 131 131 0.37 17 99 0.32 880 119 100 0.36 55 63 0.35 900 105 84 0.37 57 111 0.33 920 ------ 5.4.1.4 15 % Na·X Both 15 % Na·X samples, have similar trends to that of the 10 % samples. Specifically the increase ◦ in Y123 being strictly monotonic, however with no anomalies as per 10 % Na2CO3 at 800 C and 820 ◦C. Both contain a majority of Y123 at every temperature tested (with the exception of 800 ◦C ◦ for NaCl) with a total volume fraction at 920 C of 82.6 % and 81.6 % for NaCl and Na2CO3 respectively. The fraction of Y211 in both samples also increases towards the higher temperatures. These values are either lower or within a percent of the 10 % volume fractions, suggesting that any further addition of sodium is unlikely to increase the total amount of Y123 formed after 2 h at 920 ◦C. Other than Y123, the NaCl sample, sees all phases decrease as temperature increases with the exceptions of Y211 which increases to a plateau at 880 ◦C at 7.1 % before remaining steady at ≈5 % for the final two temperatures. The 15 % Na2CO3 sample again shows a general trend for less of all YBCO starting material phases as temperature increases. The most obvious break from this trend is the fraction of the ◦ ◦ ◦ Y2Cu2O5 increasing between 820 C to 860 C which reaches as high as 25.8 % at 860 C before steeply declining between 860 ◦C to 900 ◦C. ◦ With 15 % Na·X content and 2 h at 800 C the morphology of the NaCl and Na2CO3 samples are already different. The NaCl sample has adopted a plate-like morphology more reminiscent [198] of Bi2Sr2CaCu2O8+x (B2212) . From a lower magnification, these side-on plates are easily confused for wires [fig. 5.31 (a)] but upon closer inspection, none are observed. These plates have not been observed in any of the previous experiments. The Na2CO3 sample has a much more mixed morphology. Again, from lower magnifications, it appears that wires can be seen extruding directly from the material surface, however, upon further magnification these structures take on the form of broad ‘tapes’ or ‘ribbons’ which themselves are rare and not representative of the sample as a whole. Again, no nanowires were observed at any location across the entire surface. 127 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.30. Comparison of the solid state reaction of YBCO precursors calcined at various temperatures. All reactions contain 15 % Na by mass as either NaCl ((a) and (c)) or Na2CO3 ((b) and (d)). (a) and (b) display pXRD patterns calcined for 2 h at the labelled temperature. (b) and (d) show the relative amounts of phases in each powder pattern calculated using multi-phase Rietveld refinement. The colours for each phase (Y123, Y211,Y 2Cu2O5, BaCuO2 ,Y2O3 ,BaCO3, CuO) are consistent across all four graphs. 128 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.31. SEM micrographs of YBCO precursors containing 15 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 800 C for 2 h. Taking the NaCl sample to 820 ◦C shows the same plate-like morphology observed at the previous temperature however only in localised regions. These regions appear to have clusters of plates with each plate influencing the growth direction of its neighbours. Although, even at higher magnifications, it is difficult to tell the difference between a wire and a plate from the side. As no obvious wires were easily identifiable, only structures with no observable second dimension were measured, giving a mean width of 219.3 nm. However with such a small sample size, this value is merely illustrative of size. With 15 % as an additive at 840 ◦C anisotropic structures of a more rod-like morphology are visible. These structures are fairly common across the surface of the sample, have parallel sides, flat ends and a mean width of 651.1 nm [figures 5.33 (c) and (d)]. At 840 ◦C there is now a clear difference between the two samples. The 15 % NaCl sample is covered in nanowires with a mean width of 188.4 nm and lengths of up to 20 µm which is the longest measured at any stage of this experiment. These wires appear to grow both in bunches 129 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.32. SEM micrographs of YBCO precursors containing 15 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 820 C for 2 h. and individually and the vast majority have parallel sides and are without bends. The sample containing 15 % Na2CO3 has no signs of wires at all. The surface has a featureless morphology with only a possible hint of the rod-like structures found when reacted 20 ◦C cooler. Both reactions left at 860 ◦C look remarkably similar considering how contrasting they were at 840 ◦C. The wider granular morphology has a scattering of thin and relatively short wire- like structures growing directly out of a grain measured in µm [figures 5.34 (b) and (d)]. It is not possible to state whether these structures were the beginnings of larger structures or the remains. However, based on the previous data, the fact that the wires are all so thin likely means that wire merging and micro-crucible creep have not yet had sufficient time to manifest implying that they are the former. As the system is reacted at 880 ◦C, again the concomitant samples begin to look different from one another. Although with only sparse coverage, the 15 % NaCl sample has clusters of nanowire 130 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.33. SEM micrographs of YBCO precursors containing 15 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 840 C for 2 h. The particles apparent on the surface of the wires in figure 5.33 (b) are silver wetting due to a sputter-coater issue. bunches across the surface [fig. 5.35 (a)]. These wires have a mean width of 106.7 nm but are ◦ noticeably shorter than those observed at 840 C with any exceeding 3 µm rare. The Na2CO3 sample however appears to have much longer wire-like structures growing across the sample surface instead of away from it. This could be a result of structural collapse due to fragile, long wires being mounted on the sample holder and sputter coated. If this is the case, it is interesting that shorter wires appear mostly absent. For the sake of completeness, both samples reacted at 900 ◦C and 920 ◦C are included here in figure 5.36. As can be seen, all four reactions resulted in the standard Y123 morphology with no indications of any nanostructures of any form. With 15 % Na·X as an additive to the reaction, with the exception of NaCl at 840 ◦C, nanowires were either absent or scarce. The first obvious sign of any one-dimensional structures appeared 131 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.34. SEM micrographs of YBCO precursors containing 15 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 860 C for 2 h. ◦ in the Na2CO3 sample at 820 C and were thick and rod-like, with a lower aspect-ratio than is typically seen in YBCO nanowires. Stepping the reaction temperature up 20 ◦C these structures do not reappear. The next anisotropic structures found in the 15 % Na2CO3 sample appear at 860 ◦C as short and thin structures becoming more of a texture to the surface than as independent wires. Increasing the reaction temperature another 20 ◦C does indeed produce wires that are long and lie flat against the surface of the sample which has not been seen in any of the other samples in this section. The 15 % NaCl sample shows no sign of one-dimensional structures until 840 ◦C where there is a high coverage of long nanowires pointing away from the surface, as seen in the other samples. This is a stark contrast to the same mixture heated to 860 ◦C which resembles that described ◦ earlier for the Na2CO3 sample. At 880 C there is another ‘crop’ of wires but markedly shorter than those seen at 840 ◦C. 132 5.4. THE USE OF SODIUM AS A FLUX (a) (b) (c) (d) FIGURE 5.35. SEM micrographs of YBCO precursors containing 15 % NaCl (a)-(b) and ◦ Na2CO3 (c)-(d) held at 880 C for 2 h. 5.4.2 Identification of nanowires Electron diffraction is a common technique for analysing lattice spacing, space groups and symmetry of crystal structures in a transition electron microscope. The use of the electron beam to generate patterns based on Bragg diffraction of electrons is both convenient and informative. In a TEM when using standard imaging mode, the image is made up of contrast generated by scattered electrons from an electron beam, transmitted through the sample. Using post- sample apertures and lenses electrons scattered at any point on the sample cause a contrast differential up at the equivalent place on the image, i.e a point on the sample corresponds to a point on the image. When performing SAED however, elastically scattered electrons that are diffracted by the sample at the same angle end up at the same place on the viewing screen, i.e a specific scattering angle relates to a specific point on the screen. SAED relies on these elastically scattered electrons generated from parallel illumination, leaving the sample at a discrete set of 133 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) FIGURE 5.36. SEM micrographs of YBCO precursors containing 15 % Na·X. (a) NaCl ◦ ◦ ◦ held at 900 C (b) Na2CO3 held at 900 C (c) NaCl held at NaCl held at 920 C and ◦ (d) Na2CO3 held at NaCl held at 920 C, all for 2 h. angles to generate a series of dots we know as a diffraction pattern as Bragg law states that a constructively interfering, elastically scattered electron travels at 2θ from its initial trajectory. In contrast to pXRD and scXRD the parallel incident beam requires only a small area of the sample to be illuminated (≈0.5 µm), allowing for lattice determination of far smaller crystals. This size restriction unfortunately limits the ability to optimally position the sample under the electron beam, which necessarily gives rise to a high probability of diffraction by high index planes that become hard to index. This, coupled with the difficult task of transferring nanowires on the surface of a bulk material to a TEM grid makes identification of the nanowires observed in the previous section difficult. For this reason an an attempt to characterise the 5 % Na·X samples, the samples with the highest population of nanowires, via SAED has been attempted. 134 5.4. THE USE OF SODIUM AS A FLUX Table 5.4: Numerical analysis of observed nanowire diameters of samples containing 15% Na content obtained through image analysis of SEM micrographs from figures 5.31- 5.36. Absent numbers indicate samples that had no observable nanowires. Count indicates the total number of wires measured in the image, mean is calculated using a log-normal distribution fitting of the data and lnσSD is the log of the standard deviation. 15% NaCl Na2CO3 Mean Mean Temperature (◦C) Count ln σ (nm) Count ln σ (nm) diameter (nm) SD diameter (nm) SD 800 - - - 6 679 0.77 820 7 219 0.28 15 651 0.26 840 158 188 0.44 - - - 860 64 65 0.32 22 115 0.50 880 154 106 0.36 17 228 0.62 900 ------920 ------ 5.4.2.1 Selected area of electron diffraction of nanowires As mentioned in section 2.3.5.1, in order to get wires off of the sample surface and onto the 3 mm sample grid, a small amount of the pellet was submerged in ethanol and sonicated until slightly turbid. At this point, a single drop of the turbid ethanol was placed on top of a sample grid and left to evaporate. Unfortunately this sample preparation technique does not guarantee the intended part of a specimen to be observable on the carbon film. The issue is compounded by the With NaCl as an additive, despite multiple attempts to analyse the sample via TEM, no wires were observed on the sample grid for the temperatures 800 ◦C, 820 ◦C, 860 ◦C, 900 ◦C and 920 ◦C. This is likely due to the scarcity of wires observed on the sample surface during SEM analysis. The SAED analysis for 840 ◦C and 880 ◦C for 5 % is shown in figure 5.37. (a) 840 ◦C 5 % NaCl (b) 880 ◦C 5 % NaCl ◦ ◦ FIGURE 5.37. TEM and SAED of nanowires grown with 5 % NaCl at 820 C and 880 C for (a) and (b) respectively. 135 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM For NaCl, the sample formed at 840 ◦C, dark single anisotropic structures typical of single crystals were observed. These structures overwhelmingly had truncated ends similar to that seen in figure 5.37 (a). The sides of the wires were not seamlessly parallel, but did not taper towards the tip. The non-uniformity of the sides however was not consistent across the sample. When in diffraction mode, a regular array of spots are immediately apparent. Calculating the d-spacings, the reflections down the long axis of the wire show d=3.566 Å which matches the (200) reflection of the Y211 phase. This is perpendicular to the (002) reflection across the width of wire with d=2.892 Å as expected from the Y211 orthorhombic unit cell. Once identified as the Y211 phase, the (202) (d=2.2163 Å) and (201) (d=3.168 Å) reflections are easily identified based on inter-planar angles of 32.22° and 51.57° and from [100], respectively. The sample formed at 880 ◦C shows quite a different morphology within the anisotropic structures. As can be seen in figure 5.37 (b), each anisotropic structure appears to be made up of a group of thinner structures. Although with a polycrystalline sample a set of diffraction rings would be expected under SAED, diffraction spots are still present and indexable as Y211 reflections. The slight smearing of these spots indicates some misalignment between sub-structures but generally they appear to be in the same orientation, which indicates that either the thinner wires started independently and aligned through epitaxy or that the larger structure has somehow split down the long axis at some point during the growth and sample preparation process. The ‘stepped’ end of the larger structure and comparable widths of the smaller structures suggest the former. In this sample, the same reflections are present at the same angles and (200) is still seen as propagating along the wire length. With no wires harvested from the 800 ◦C sample, figure 5.38 shows anisotropic structures ◦ ◦ ◦ isolated with Na2CO3 as an additive at 5 % formed at temperatures of 820 C, 840 C and 860 C. At 820 ◦C all structures found, were clearly not single crystals and resembled the ‘current bun’ arrangement described in reference 191. Interestingly however, the wire-like structures remain implying that the current arrangement is the result of a wire breaking down or reacting post- growth. As expected of a polycrystalline sample SAED of these structures showed diffraction rings that have remained unindexed despite repeated attempts to match them with phases containing any combination of Y, Ba, Cu, Na, C and O. At 840 ◦C observed wires have taken on a more homogeneous form [fig. 5.38 (b)] again with a truncated end and parallel sides. SAED revealed these wires to again be Y211 with the (200) reflections along the length of the structure. The (024) (d=1.38 Å) reflection is seen normal to the (200) and sequential slicing of the a-axis is seen with reflections (124), (224) (324) and (424) visible with d-spacings of 1.35 Å, 1.29 Å, 1.19 Å ◦ and 1.09 Å. The SEM micrographs of the Na2CO3 sample heated to 860 C show the more rod- like structures [fig. 5.18 (d)] and were not easy to isolate on a TEM grid. Anisotropic structures were found in small regions but these appeared to be thinner than any observed in the SEM micrographs. SAED of these structures again revealed the Y211 structure. Indexable reflections included the long axis (200) as seen in the other structures, (220), (490) and (410) with respective 136 5.4. THE USE OF SODIUM AS A FLUX ◦ ◦ (a) 820 C 5 % Na2CO3 (b) 840 C 5 % Na2CO3 ◦ (c) 860 C 5 % Na2CO3 ◦ ◦ FIGURE 5.38. TEM and SAED of nanowires grown with 5 % Na2CO3 at 820 C, 840 C and 860 ◦C for (a), (b) and (c) respectively. inter planar angles from (200) of 30.35°, 52.80° and 8.33° and d spacings of 3.08 Å, 1.08 Å and 1.76 Å. ◦ The 5 % Na2CO3 sample heated to 880 C, again had single crystal anisotropic structures. Upon analysis of various SAED patterns recorded throughout this sample revealed two different YBCO identities present as wires, both Y211 and Y123. These wires were again uniform in width but occasionally had domed tips, although the diameter of these domes did not exceed that of the wire width. Figure 5.39 (a) is another Y211 nanowire and (b) is the first appearance of Y123. Although reflections along the length of the wire were not strong enough to measure, visible diffraction spots to measure d-spacings and inter-planar angles for indexing reflections (020), (022) and (017) were available. Based on the indexed reflections the growth along the length of the wire would be the c-axis. Raising the temperature again to 900 ◦C again, wires of different identities are observed however this time, along with Y211, wires of BaCuO2 appear to be present. Although only a trace 137 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM ◦ ◦ (a) 880 C 5 % Na2CO3 (b) 880 C 5 % Na2CO3 ◦ FIGURE 5.39. TEM and SAED of nanowires grown with 5 % Na2CO3 at 880 C. (a) Y211 and (b) Y123. ◦ ◦ (a) 900 C 5 % Na2CO3 (b) 900 C 5 % Na2CO3 ◦ FIGURE 5.40. TEM and SAED of nanowires grown with 5 % Na2CO3 at 900 C. (a) Y211 and (b) BaCuO2. amount of the total volume fraction of the bulk material based on the refinement of the powder pattern by 900 ◦C, it does appear to occupy at least a small percentage of the wire population. Figure 5.40 (a) is another example of Y211 although corroded edges are now representative of the sample as a whole. It should be noted that finding wires that produced a serviceable SAED pattern were hard to find on the 900 ◦C sample. Many such structures were visible but apparently not as highly crystalline as in other samples. The final sample indexed at 920 ◦C again had two species evident; Y211 [fig. 5.41 (a)] and Y123 [fig. 5.41 (b)]. The clear difference between these two wires is the state of the wire edges. At temperatures above 880 ◦C the edges to Y211 wires have begun to lose their uniformity however 138 5.5. BRIEF ATTEMPTS ON ANOTHER SYSTEM ◦ ◦ (a) 920 C 5 % Na2CO3 (b) 920 C 5 % Na2CO3 ◦ FIGURE 5.41. TEM and SAED of nanowires grown with 5 % Na2CO3 at 920 C. (a) Y211 and (b) Y123. the straight non-corroded edges are found on all Y123 indexed. Although Y211 and BaCuO2 are know to be pre-cursors to Y123, it is worth noting that once a wire of Y211 has grown, it is no longer surrounded by anything other than atmosphere, impeding any further reaction that may be favourable at the higher temperatures. This leaves the outer layers of a high-surface area material susceptible to oxygen/ion depletion by leeching into the atmosphere, which may indeed be why the phase formed at lower temperatures is seen to be corroding. Other causes of ‘rough’ edges on nanowires have been reported [202] and the pseudo-periodicity of the edges in figure 5.41 (a) is intriguing, further study would be definitely be required before suggesting that any kind of oscillatory growth was involved. 5.5 Brief attempts on another system Although direct control over width, length and wire stoichiometry is not yet perfected, a brief experiment was attempted in another system. BSCCO is another group of cuperate, high temperature superconductors with BSCCO-2201, BSCCO-2212 and BSCCO-2223 stoichiometries [203,204] having a Tc of 15, 85 and 110 K respectively. Previous work has detailed the growth of BSCCO-2212 from solubalised nitrate precursors using bio-templates[198]. The relative abundance of precursors and intermediate phases involved in the formation of BSCCO-2212 are shown in figure 5 in reference 198 as a function of temperature. Using all nitrate salts as starting materials, after calcining above 350°C the organic template has been removed leaving oxides and carbonates. The formation of the intermediate phases BSCCO-2201 and Bi2Ca1+xSr3−xO7 can clearly be seen as essential to the formation of BSCCO-2212 above 800 ◦C as all starting materials have vanished by this point. As these data were obtained via a quench study, that is the reaction was stopped at the reported temperature with no isotherm, to garner the mechanism of 139 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM (a) (b) (c) (d) (e) FIGURE 5.42. Nanowires visible in the BSCCO system containing 5 % Na2CO3 at various temperatures. (a)-(e) show the system held for 2 h at 760 ◦C, 780 ◦C, 800 ◦C, 820 ◦C and 840 ◦C respectively. phase evolution, it can be used as a step by step insight into how the phases form and contribute to the formation of the ultimate phase. Although previously grown in numerous ways only as ‘whiskers’ on the micron-scale[205], to date, nanowires of BSCCO remain unreported making it a good candidate for a trial. As per the YBCO system, a pellet of BiO and CuO was seeded with Ca and Sr carbonate nanoparticles, synthesised as described in section 2.2.10, in a stoichiometric ratio for BSCCO- ◦ ◦ ◦ 2212 with 5 % Na2CO3 added. Micrographs of the system after 2 h at 760 C, 780 C, 800 C, 820 ◦C and 840 ◦C are shown in figure 5.42. As can be seen, with 5 % Na2CO3 in the system, nanowires can be seen at every temperature 140 5.6. SUMMARY tested. Unfortunately, due to their scarcity, attempts to extract and mount the nanowires on a TEM grid are as yet unsuccessful making full characterisation difficult. However, as the two systems attempted in this chapter only share Cu and O and no wires characterised so far were exclusively CuO, the nanowires in this sample must therefore be some combination of Bi, Sr, Ca, Cu and O, apparently induced by the introduction of carbonate nanoparticles and the addition of Na2CO3 to the mixture. 5.6 Summary The aim of this chapter was to explore methods for a general approach for complex metal-oxide nanowire growth by exploiting mechanisms known to occur when using a biotemplate, without the use of any organic component. Initially a solid state reaction of oxide pre-cursors were seeded with BaCO3 nanoparticles to mimic the initial ‘current bun’ growth phase known to be an intermediate phase for nanowire growth. The resultant material, although it did not resemble the alginate templated wires, did show signs of nano-texturing in the form of anisotropic structures across the surface instead of projecting from it. There was also an increase in the target phase due to a higher surface area of reactants. In order to make the reactants more mobile, two Na salts; Na2CO3 and NaCl were added in various amount with the intention of forming eutectic intermediates that would not only increase mobility throughout the system but also reduce the melting points of some of these intermediate phases. Even at 2 % Na content, production of the target phase was increased substantially by 920 ◦C. This effect increased as more Na was added up until 10 % at which point adding more did not result in more Y123. Regardless of the pursuit of nanowires, the addition of 10 % by mass of a Na salt increased the Y123 yield by ≈100 % under the same reaction conditions. It is noteworthy that in no patterns collected at any temperature, or amount of Na·X tested, detected any crystalline NaCl or Na2CO3 as a raw material. Based on the amount added being far over the detection for a Bruker D8 advance (≈ 1 %) coupled with a melting point of 801 ◦C ◦ and decomposition temperature of 851 C for NaCl and Na2CO3 respectively, it is implicit that even at the lower temperatures, the Na containing species are already either reacted, lost to the atmosphere or amorphous. As no crystalline Na containing species at all are detected at any level, it is assumed that it is the latter as Na still appears to be affecting the system, even at elevated temperatures. Upon inspection of the surface of the material anisotropic structures were observed throughout at various stages of growth and surface population. 2 % by mass of Na·X caused the growth of nanowires in the system. Though not ubiquitous across the surface they were found in small isolated clusters and were thinner (<100 nm) when grown at higher temperatures 900 ◦C to 920 ◦C. Wires were far more abundant with higher amounts of Na·X between 5 % to 10 % but thicker overall with a mean thickness reaching 539.358 nm at 840 ◦C with the carbonate. Wire 141 CHAPTER 5. GROWTH OF COMPLEX-OXIDE NANOWIRES VIA SYNTHETIC APPLICATION OFTHEMICRO-CRUCIBLEMECHANISM thickness becomes harder to ascribe to temperature apart from a general tendency for wire thickness to increase towards the mid range temperatures (840 ◦C to 860 ◦C) before becoming thinner at the higher temperatures. Adding more than 10 % Na·X begins to appear excessive. At these fractions the total amount of Y123 formed has not increased from that seen at 10 % and the formation of nanowires appears to be somewhat hindered as far lower numbers were seen across the whole temperature range and none at all at the two highest temperatures. This could suggest that the elevated Na content is facilitating the growth and subsequent decline of nanowires within the 2 h hold time at any given temperature, so by the time they are observed they have been and gone. Of course with only this present data it is impossible to tell if this had indeed occurred however, the persistence of nanowires in other systems shown the same conditions is at least indicative that this is not occurring. Identifying the wires present in each sample is non trivial. As these structures appear to be a surface phenomena, the overwhelming majority of sample mass is bulk reactant with no nano texturing making dispersion across a substrate difficult. However after numerous methods were unsuccessfully employed in attempts to isolate these structures for further characterisation, gentle sonication in ethanol was found to be the best choice. It is clear from the various micrographs that anisotropic structures throughout the system grow in a range of morphologies, lengths and thickness’s indicative of different materials. Issues arising from the preparation technique however became apparent; structures observed on the TEM grids did not necessarily resemble those seen under SEM implying that this technique was perhaps selective regarding which structures were suspended in the ethanol. To compound the issues with wire identification further, the discovery of different materials being present concurrently is suggests that each ‘crop’ of wires is perhaps a mixture of materials growing in and among each other. Indeed it is entirely possible that even wires in the same bunches may be of different identities, starting with a chemically different microcrucible growing at slightly different temperatures and at slightly different speeds. This would also explain why attempts to characterise the wires with EDX often resulted in confusing and sometimes contradicting data, with wires thinner than the penetration depth of the incident electron beam returning an averaged data set of different materials. Y211 is by far the most common nanowire observed via SAED and is present at almost every temperature when 5 % Na·X is used. No system successfully analysed was void of Y211 nanowires, but these appeared to become more damaged at the higher temperatures. Y123 breaks [169] down to Y211, BaCuO2 and CuO under a oxygen restricted atmosphere over time but if Y123 is only found in the higher temperature samples, it is likely that the Y211 wires were grown at the lower temperatures before being subjected to prolonged times at higher ones, having plenty of time for such a high surface-area structure to become oxygen deficient or react with volatile metals being released into the atmosphere. Ultimately, it has been shown that nano-seeding of metal-oxide superconductors is possible 142 5.6. SUMMARY with the addition of a sodium salt flux and the initial results for a second, more complex system are encouraging. The parameter space in which to develop this method is vast. Here it has been shown how a change in Na content can affect wire growth but with little control over width and overall composition. The next steps in the development of this method include In situ observation of how these structures form on the surface of a pellet. This would reveal much about the growth of these structures and would be the obvious next step as the ability to observe structure growth at these elevated temperatures would greatly expedite the design of conditions conducive for wire growth in any oxide system. Observing their growth and evolution may eliminate the problem of concomitant materials through a change in ramp rate, hold time or atmospheric composition, perhaps even creep time for fine width adjustment. Comprehensive knowledge of the intermediate phases will be key in expanding this method to other systems so quench analysis of each system will also be a priority. In lieu of easy access to a heating stage, many quench studies will be necessary. 143 h rto hs,bsdo ctee eot fhg ult rsa rwho rtisudrthe under proteins of growth crystal quality high of reports scattered on based these, of first The in growth crystal of outcome the manipulate to which in means novel explored has project This 99ppr lsiyn lnraoai oeuepcigmtf nt ordsic rusbased groups distinct four to in motifs packing molecule aromatic planar classifying paper, 1989 a Conclusions 6.1 ntenaetnihorpcigageadui elsotai.Cmae oteubiquitous the to Compared short-axis. cell structure, unit coronene and angle packing nearest-neighbour the on a of has discovery which the molecule, researched in the and designated manifested characterised by been thoroughly This affected this growth. be of polymorph crystal to new during shown completely fields was magnetic molecule, PAH of researched the application well growth, crystal occurring initial naturally During a diamagnetic field. high magnetic coronene, a static with a crystals, in organic solution smaller from growing susceptibility, involved field, magnetic a of influence coronene in polymorphism induced Magnetically 6.1.1 growth. crystal of method then defined novel and well a a achieve phenomena taking to second, growth it the exploiting crystal and and involved mechanism reported mechanisms quantify under and an identify to finding different attempting significantly of two that using attempted firstly been has approaches, This materials. ionic and molecular both Awy owaee’ next.” whatever’s do “Always β crnn ae nanmn ovninpooe yDsrj n aeot in Gavezotti and Desiraju by proposed convention naming a on based -coronene β crnn ssont aeamc mle akn nl of angle packing smaller much a have to shown is -coronene ereCarlin George – 145 C ONCLUSIONS 49 C HAPTER . 71° 6 relative γ - CHAPTER 6. CONCLUSIONS to 95.30° observed in the γ polymorph. This new polymorph was structurally characterised with 3 scXRD, showing a short-axis of 3.7405 Å and a volume of 664.017 Å , the new polymorph has a slightly higher density when compared to the known form. Further characterisation was carried out in order to better understand this new crystal and to report the new structure. Initially, periodic boundary conditions (PBC) calculations were run in an attempt to understand the structural stability of the new polymorph showing that both the γ and β form sit in energetic minima. Depending on the calculation set used, results showed either a small difference in lattice energy < 2 kJmol−1 or a large difference suggesting that the β form is the lowest energy form at 0 K. Physical characterisation was carried out using differential scanning calorometry (DSC) and AFM shows the β form having a slightly depressed melting point and enthalpy of fusion but being substantially softer. All of these results are consistent with an enantiotropic system with the β form being metastable at room temperature. The optical behaviour was also shown to differ significantly between the polymorphs with a broad, featureless absorption band replacing sharp absorption peaks seen in the γ polymorph. Cryogenic pXRD was then carried out which highlighted a partial phase transition below 150 K which was indexable to the β form, again providing evidence that coronene is an enantiotropic system with the γ polymorph metastable at lower temperatures. The reason for the partial transformation was shown to be size specific with a much higher percentage of the volume seeing a transformation below 150 K when measured as micron sized crystallites. As larger crystals apparently do not transform polymorph, even when thermally stimulated, in order to understand what role the magnetic field plays in the formation of the β polymorph a deeper look into the inception of crystal growth was required, before clusters or crystallites become too large to fall on either side of this structural ‘knife-edge’. After the role of some magnetic impurities were investigated and shown to have no effect, even at high concentrations, the packing of small numbers of molecules was simulated in order to understand if the differences in inter-molecular interactions are something that are changed to or changed from during the clustering of molecules and resultant crystal growth. Based on work designed specifically to determine the geometries of PAHs molecules with respect to one another, clusters of up to 10 coronene molecules were optimised. As per previous work, the molecules overwhelmingly adopted a face···face stacking overlap of 1.61 Å which is effectively the overlap seen in the β-coronene structure. This, coupled with the large variation of within nearest-neighbour angle suggested that not only was the stacking of molecules the primary driving force during the early stages of nucleation, but that in small numbers, the β form is preferred. Although this data was generated using the optimisation of free molecules, not those already crystallised, it is still in agreement with the size dependant structural analysis. With, evidence that the effect must be occurring at the inception of crystal nucleation and these small clusters initially involve molecules interacting differently that seen in the γ polymorph, attempts to locate and then characterise these clusters were made. UV-vis 146 6.1. CONCLUSIONS spectroscopy of the saturated coronene solution was undertaken in order to locate absorbance peaks predicted by calculations on the optimised geometries calculated earlier. This analysis did reveal discrete absorptions that were in good agreement with these calculations, implying that the small β like clusters were both present at high concentrations and of a range of sizes. As it now looked likely that as coronene crystals grow they take on the β form below a critical size and that upon exceeding that size the γ polymorph becomes energetically favourable and that the magnetic field was somehow suppressing the change to γ, not actively causing β to form. Identification of coronene cluster sizes were attempted via various techniques, namely DLS and DOSY, both of which gave ambiguous results making quantification of the particle sizes difficult. DLS however did appear to change in the system at 42 ◦C, this could be the formation on crystallites as a function of temperature however this remains speculation. With regards to higher magnetic-fields, crystals were grown in fields of up to 20 T whilst being observed spectroscopically in situ. The results of these experiments showed a suppression of crystal growth with a reflexive relationship as a function of field, specifically that crystal growth is drastically suppressed, increasing supersaturation, up to 0.9 T above which, this effect lessens. This relationship is indicative of (at least) two competing effects, the summation of which see a minimum at 0.9 T in the coronene-toluene system. This minimum is allowing the supersaturation of the system to reach levels higher than those accessible via other means, which appears to be allowing the new, β polymorph of coronene to form from solution. 6.1.2 Templated growth of metal-oxide nanowires The second direction presented here involved exploitation of the well-characterised method of biotemplated growth of YBCO nanowires in order to create a general method for metal-oxide nanowire growth. The high-temperature superconductor YBCO was synthesised in various ways in the solid state as to make Y123 the target phase, as opposed to from aqueous salts used in the biotemplating process. Initially, barium carbonate nanoparticles were synthesised and added to the solid-state reaction mixture, replacing standard powdered barium carbonate. The resultant material had a higher volume fraction of the target phase and had anisotropic surface structures across the whole sample. While not morphologically resembling the wires observed from the alginate templated synthesis, this was indeed evidence that the nano-structure of the material, is controllable by changing the size of the inorganic starting materials. In order to allow more ionic transport throughout the reaction material, sodium salts were added intended as a flux, 2 %, 5 %, 10 % and 15 % by weight with both the chloride and carbonate salts. Powder diffraction immediately revealed a higher fraction of target phase, implying a more mobile environment for reactants even at 2 % by weight. This increase in target phase volume appeared to reach a maximum by 10 % sodium salt as increasing it more appeared to have no further effect. Analysis of the surface of the reactants showed varying morphologies and numbers of 147 CHAPTER 6. CONCLUSIONS nanowires dependant on the temperature and amount of sodium salt added. It can be stated unambiguously that both addition of the sodium salt and nano-sized barium carbonate starting material are instrumental in the growth of nanowires in this system. Nanowire formation was tracked at temperature isotherms of 800 ◦C, 820 ◦C, 840 ◦C, 860 ◦C, 880 ◦C, 900 ◦C and 920 ◦C with a dwell time of 2 h in every case. These parameters were chosen in order to identify the ideal variants for nanowire growth in this particular system. Encouragingly, every reaction mixture used in this set of experiments saw anisotropic structures at some temperature tested. In order to identify some set of optimal parameters for growth, numbers of observed wires were used to quantify the success of an experiment. With no conclusive general pattern regarding, how higher amounts of Na·X and higher temperatures affect the number of nanowires, each mixture appeared to have an optimum temperature. Nanowire thickness, thickness distribution and length varied greatly within the parameter space and again appeared to not follow any obvious pattern. During analysis of specific nanowires using SAED, it is apparent that the vast majority of nanowires in all systems, at least with 5 % Na·X is Y211. This could be due to Y211 being more amenable to separation from the bulk for analysis or it could be that Y211 is just overwhelmingly formed based on the reaction conditions and available oxygen. Interestingly, nanowires of Y123 and BaCuO2 were also observed, which is good evidence for the generality of the technique. A brief foray into another system, that of BSCCO produced some promising results, again of nanowires of varying morphologies based on the temperature used. Sharing only copper and oxygen, these nanowires must be of a different elemental composition to any found in the YBCO system. Although only qualitative data, these micrographs confirm the general applicability of the method making it a good starting point for growth of nanowires in any metal oxide system. 6.2 Further work 6.2.1 Magnetically induced polymorphism in coronene Moving forward, there are some key threads to follow towards finding both a mechanism for the suppression of crystal growth which seems intimately linked to β crystal growth. Reliable particle size analysis as a function of both saturation and magnetic field is paramount. How the size, shape, internal structure and distribution of pre-nucleation clusters behave as a function of applied magnetic field is absolutely necessary information for resolving this effect. Unfortunately, the complicated nature of both preparing a usable coronene sample for use in a technique that would allow for collection of this data has proven difficult. To compliment this data, a more thorough set of simulations should be generated, specifically at higher temperatures. Although adding thermal motion to a simulation is non-trivial, the fast pace at which non-specialist software has been developing should make such calculations both accessible and increasingly accurate in the near future. Calculations of this nature may provide 148 6.2. FURTHER WORK some critical insight into when a cluster of coronene molecules are have a thermal incentive to adopt the γ polymorph as a function of size. Secondary to this, based on the computational data run for the stability of the two coronene polymorphs, it should be possible to map a likely physical pathway between γ and β when transforming. Calculating the energies related to the unit cells and interactions between the molecules, gives the contribution of each interaction to the stability of the bulk crystal giving a ‘path of lowest energy’ for a structural flip. Most importantly, it is desirable to find a general method of polymorph selection through the application of magnetic-fields. Of course, as the understanding of how magnets affect the mechanisms of crystal growth increases, so does the ability to pick other suitable candidates. 6.2.2 Tempted growth of metal-oxide nanowires The obvious place to move forward with this technique would be to attempt it on a variety of systems that have never been reported to produce nanowires. However, due to the variety of wire species found using SAED, gaining tighter control over the identity of the wires would be a preferable next step. Ideally, this would be done by direct observation of wire growth using a heating stage within a SEM. With this technique, the temperature at which wires grow could be unambiguously stated, assuming that wires of different species grow at different temperatures based on the reaction temperature. This allows for quenching of the reaction, at the optimal time in which to harvest wires of a specific stoichiometry. This would also allow observation of wire merging, and possibly creep, in order to use time and temperature to control width and allow for quick determination of whether other variables can have an effect (heating rate, etc.). Once adequately quantified, this technique can seemingly be applied to other systems, creating nanowires of a wider range of systems for analysis and potential applications. 149 BIBLIOGRAPHY [1] A. K. Geim and K. Novoselov, “The rise of graphene,” Nature Materials, vol. 6, no. 3, pp. 183–191, 2007. [2] B. Y. Shekunov, M. E. Aulton, R. W. Adama-acquah, and D. J. W. Grant, “Effect of temperature on crystal growth and crystal properties of paracetamol,” Journal of the Chemical Society, vol. 92, no. 3, pp. 439–444, 1996. [3] R. K. Harris, P. Y. Ghi, H. Puschmann, D. C. Apperley, U. J. Griesser, R. B. Hammond, C. Ma, K. J. Roberts, G. J. Pearce, J. R. Yates, and C. J. Pickard, “Structural studies of the polymorphs of carbamazepine, its dihydrate, and two solvates,” Organic Process Research & Development, vol. 9, no. 6, pp. 902–910, 2005. [4] L. Bonnefond, P. Schellenberger, J. Basquin, G. Demangeat, C. Ritzenthaler, R. Chên- evert, C. Balg, M. Frugier, J. Rudinger-Thirion, R. Giegé, B. Lorber, and C. Sauter, “Exploiting protein engineering and crystal polymorphism for successful X-ray structure determination,” Crystal Growth & Design, vol. 11, no. 10, pp. 4334–4343, 2011. [5] G. Schweicher, Y. Olivier, V. Lemaur, and Y. H. Geerts, “What currently limits charge carrier mobility in crystals of molecular semiconductors?,” Israel Journal of Chemistry, vol. 54, no. 5-6, pp. 595–620, 2014. [6] D. Erdemir, A. Y. Lee, and A. S. Myerson, “Nucleation of crystals from solution: classical and two-step models,” Accounts of Chemical Research, vol. 42, no. 5, pp. 621–629, 2009. [7] S. W. Young, “Mechanical Stimulus to Crystallization in Super-Cooled Liquids,” Journal of the American Chemical Society, vol. 33, no. 2, pp. 148–162, 1911. [8] E. of Berkeley, “Solubility and supersolubility from the osmotic standpoint,” The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, vol. 24, no. 140, pp. 254–268, 1912. [9] E. Khamskii, “Crystallization from solutions,” Consultants Bureau, 1969. [10] F. Ito, Y. Suzuki, J. I. Fujimori, T. Sagawa, M. Hara, T. Seki, R. Yasukuni, and M. L. de la Chapelle, “Direct visualization of the two-step nucleation model by fluorescence color 151 BIBLIOGRAPHY changes during evaporative crystallization from solution,” Scientific Reports, vol. 6, 2016. [11] R. Demichelis, P. Raiteri, J. D. Gale, D. Quigley, and D. Gebauer, “Stable prenucleation mineral clusters are liquid-like ionic polymers,” Nature Communications, vol. 2, no. 1, p. 590, 2011. [12] D. Zahn, “Thermodynamics and Kinetics of Prenucleation Clusters, Classical and Non- Classical Nucleation,” ChemPhysChem, vol. 16, no. 10, pp. 2069–2075, 2015. [13] S. Karthika, T. K. Radhakrishnan, and P. Kalaichelvi, “A Review of Classical and Nonclassical Nucleation Theories,” Crystal Growth & Design, vol. 16, no. 11, pp. 6663– 6681, 2016. [14] S. T. Yau and P. G. Vekilov, “Direct observation of nucleus structure and nucleation pathways in apoferritin crystallization,” Journal of the American Chemical Society, vol. 123, no. 6, pp. 1080–1089, 2001. [15] P. W. G. Poodt, M. C. R. Heijna, K. Tsukamoto, W. J. De Grip, P. C. M. Christianen, J. C. Maan, W. J. P. Van Enckevort, and E. Vlieg, “Suppression of convection using gradient magnetic fields during crystal growth of NiSO4 · 6H2O,” Applied Physics Letters, vol. 87, no. 21, pp. 1–3, 2005. [16] K. Onuma and A. Ito, “Cluster Growth Model for Hydroxyapatite,” Chemistry of Materials, vol. 10, no. 11, pp. 3346–3351, 1998. [17] P. R. ten Wolde and D. Frenkel, “Enhancement of Protein Crystal Nucleation by Critical Density Fluctuations,” Science, vol. 277, no. 5334, pp. 1975–1978, 1997. [18] A. Gavezzotti, “Molecular Aggregation of Acetic Acid in a Carbon Tetrachloride Solution: A Molecular Dynamics Study with a View to Crystal Nucleation,” Chemistry - A European Journal, vol. 5, no. 2, pp. 567–576, 1999. [19] K. G. Soga, J. R. Melrose, and R. C. Ball, “Metastable states and the kinetics of colloid phase separation,” Journal of Chemical Physics, vol. 110, no. 4, pp. 2280–2288, 1999. [20] P. G. Vekilov, “Nucleation,” Crystal Growth & Design, vol. 10, no. 12, pp. 5007–5019, 2010. [21] Wöhler and Liebig, “Untersuchungen über das Radikal der Benzoesäure,” Annalen der Pharmacie, vol. 3, no. 3, pp. 249–282, 1832. [22] T. Matsukawa, M. Yoshimura, M. Uchiyama, M. Yamagishi, A. Nakao, Y. Takahashi, J. Takeya, Y. Kitaoka, Y. Mori, and T. Sasaki, “Polymorphs of rubrene crystal grown from solution,” Japanese Journal of Applied Physics, vol. 49, no. 8 PART 1, p. 085502, 2010. 152 BIBLIOGRAPHY [23] R. G. Della Valle, E. Venuti, A. Brillante, and A. Girlande, “Inherent structures of crystalline tetracene,” Journal of Physical Chemistry A, vol. 110, no. 37, pp. 10858–10862, 2006. [24] A. D. McNaught and A. Wilkinson, IUPAC Compendium of Chemical Terminology. Blackwell Scientific Publications, 2nd ed., 2012. [25] M. I. Kay, Y. Okaya, and D. E. Cox, “A refinement of the structure of the room-temperature phase of phenanthrene, C14H10, from X-ray and neutron diffraction data,” Acta Crystallographica Section B, vol. 27, no. 1, pp. 26–33, 1971. [26] S. A. Surwase, J. P. Boetker, D. Saville, B. J. Boyd, K. C. Gordon, L. Peltonen, and C. J. Strachan, “Indomethacin: New polymorphs of an old drug,” Molecular Pharmaceutics, vol. 10, no. 12, pp. 4472–4480, 2013. [27] A. Corma, M. Moliner, Á. Cantín, M. J. Díaz-Cabañas, J. L. Jordá, D. Zhang, J. Sun, K. Jansson, S. Hovmöller, and X. Zou, “Synthesis and structure of polymorph B of zeolite beta,” Chemistry of Materials, vol. 20, no. 9, pp. 3218–3223, 2008. [28] G. R. Desiraju and A. Gavezzotti, “Crystal structures of polynuclear aromatic hydrocar- bons. Classification, rationalization and prediction from molecular structure,” Acta Crystallographica Section B, vol. 45, no. 5, pp. 473–482, 1989. [29] A. Srivastava, H. Rotella, S. Saha, B. Pal, G. Kalon, S. Mathew, M. Motapothula, M. Dykas, P. Yang, E. Okunishi, D. D. Sarma, and T. Venkatesan, “Selective growth of single phase VO2(A, B, and M) polymorph thin films,” APL Materials, vol. 3, no. 2, p. 026101, 2015. [30] A. L. Grzesiak, M. Lang, K. Kim, and A. J. Matzger, “Comparison of the Four Anhydrous Polymorphs of Carbamazepine and the Crystal Structure of Form I,” Journal of Pharmaceutical Sciences, vol. 92, no. 11, pp. 2260–2271, 2003. [31] S. T. Salammal, J. Y. Balandier, J. B. Arlin, Y. Olivier, V. Lemaur, L. Wang, D. Beljonne, J. Cornil, A. R. Kennedy, Y. H. Geerts, and B. Chattopadhyay, “Polymorphism in bulk and thin films: The curious case of dithiophene-DPP(Boc)-dithiophene,” Journal of Physical Chemistry C, vol. 118, no. 1, pp. 657–669, 2014. [32] W. Ostwald, “Studien über die Bildung und Umwandlung fester Körper,” Zeitschrift für Physikalische Chemie, vol. 22, pp. 289–330, 1897. [33] W. Mullin, J, Crystallisation. Oxford: Butterworth-Heinemann, 4th ed., 2002. [34] B. V. Spitsyn, “Crystal Growth under Thermodynamically Metastable Conditions,” in Growth of Crystals, pp. 58–66, Boston, MA: Springer, 1986. 153 BIBLIOGRAPHY [35] N. Tsapatsaris, B. A. Kolesov, J. Fischer, E. V. Boldyreva, L. Daemen, J. Eckert, and H. N. Bordallo, “Polymorphism of paracetamol: A new understanding of molecular flexibility through local methyl dynamics,” Molecular Pharmaceutics, vol. 11, no. 3, pp. 1032–1041, 2014. [36] C. Sudha and K. Srinivasan, “Supersaturation dependent nucleation control and separation of mono, ortho and unstable polymorphs of paracetamol by swift cooling crystallization technique,” CrystEngComm, vol. 15, no. 10, p. 1914, 2013. [37] L. Yu, “Polymorphism in molecular solids: An extraordinary system of red, orange, and yellow crystals,” Accounts of Chemical Research, vol. 43, no. 9, pp. 1257–1266, 2010. [38] W. H. Bragg and W. L. Bragg, “The Reflection of X-rays by Crystals,” Proceedings of the Royal Society A, vol. 88, no. 605, pp. 428–438, 1913. [39] J. A. Bearden, “X-ray wavelengths,” Reviews of Modern Physics, vol. 39, no. 1, pp. 78–124, 1967. [40] A. N. Christensen, M. S. Lehmann, and M. Nielsen, “Solving crystal structures from powder diffraction data,” Australian journal of physics, vol. 38, no. 3, pp. 497–506, 1985. [41] A. Meden, “Crystal Structure Solution from Powder Diffraction Data – State of the Art and Perspectives ,” Croatica Chemica Acta, vol. 71, no. 3, pp. 615–633, 1998. [42] A. Hewat, W. I. F. David, and L. van Eijck, “Hugo Rietveld (1932–2016),” Journal of Applied Crystallography, vol. 49, no. 4, pp. 1394–1395, 2016. [43] H. M. Rietveld, “A profile refinement method for nuclear and magnetic structures,” Journal of Applied Crystallography, vol. 2, no. 2, pp. 65–71, 1969. [44] B. H. Toby, “R factors in Rietveld analysis: How good is good enough?,” Powder Diffraction, vol. 21, no. 01, pp. 67–70, 2006. [45] J. Potticary, L. R. Terry, C. Bell, A. M. Collins, C. Fontanesi, G. Kociok-Kohn, S. Crampin, E. Da Como, and S. R. Hall, “Control of polymorphism in coronene by the application of magnetic fields,” arXiv Preprint Physics, pp. 1–22, 2015. [46] J. Potticary, L. R. Terry, C. Bell, A. N. Papanikolopoulos, P. C. M. Christianen, H. Engelkamp, A. M. Collins, C. Fontanesi, G. Kociok-Köhn, S. Crampin, E. D. Como, S. R. Hall, E. Da Como, and S. R. Hall, “An unforeseen polymorph of coronene by the application of magnetic fields during crystal growth,” Nature Communications, vol. 7, no. May, p. 11555, 2016. 154 BIBLIOGRAPHY [47] J. Potticary, R. Boston, L. Vella-Zarb, A. Few, C. Bell, and S. R. Hall, “Low temperature magneto-morphological characterisation of coronene and the resolution of previously observed unexplained phenomena,” Scientific Reports, vol. 6, p. 38696, 2016. [48] J. Potticary, M. P. Avery, D. Mills, and S. R. Hall, “DONALD: A 2.5 T wide sample space permanent magnet,” HardwareX, 2018. [49] L. Huang, Q. Liao, Q. Shi, H. Fu, J. Ma, and J. Yao, “Rubrene micro-crystals from solution routes: their crystallography, morphology and optical properties,” Journal of Materials Chemistry, vol. 20, no. 1, p. 159, 2010. [50] H. G. Davis and S. Gottlieb, “Density and refractive index of multi-ring aromatic compounds in the liquid state,” Fuel, vol. 42, no. 37, 1963. [51] S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M. I. J. Probert, K. Refson, and M. C. Payne, “First principles methods using CASTEP,” Zeitschrift fur Kristallographie, vol. 220, no. 5-6, pp. 567–570, 2005. [52] G. Kresse and J. Furthmüller, “Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,” Physical Review B, vol. 54, no. 16, pp. 11169– 11186, 1996. [53] J. D. Pack and H. J. Monkhorst, “Special points for Brillouin-zone integrations,” Physical Review B, vol. 16, no. 4, pp. 1748–1749, 1977. [54] E. R. McNellis, J. Meyer, and K. Reuter, “Azobenzene at coinage metal surfaces: Role of dispersive van der Waals interactions,” Physical Review B, vol. 80, no. 20, p. 205414, 2009. [55] S. Grimme, “Semiempirical GGA-type density functional constructed with a long-range dispersion correction,” Journal of Computational Chemistry, vol. 27, no. 15, pp. 1787– 1799, 2006. [56] I. Fedorov, Y. Zhuravlev, and V. Berveno, “Properties of crystalline coronene: Dispersion forces leading to a larger van der Waals radius for carbon,” Physica Status Solidi (B) Basic Research, vol. 249, no. 7, pp. 1438–1444, 2012. [57] P. E. Blöchl, “Projector augmented-wave method,” Physical Review B, vol. 50, no. 24, pp. 17953–17979, 1994. [58] A. Tkatchenko and M. Scheffler, “Accurate molecular van der Waals interactions from ground-state electron density and free-atom reference data,” Physical Review Letters, vol. 102, no. 7, p. 073005, 2009. 155 BIBLIOGRAPHY [59] T. Bucko,ˇ S. Lebègue, J. Hafner, and J. G. Ángyán, “Tkatchenko-Scheffler van der Waals correction method with and without self-consistent screening applied to solids,” Physical Review B, vol. 87, no. 6, p. 064110, 2013. [60] R. Dovesi, R. Orlando, A. Erba, C. M. Zicovich-Wilson, B. Civalleri, S. Casassa, L. Maschio, M. Ferrabone, M. De La Pierre, P. D’Arco, Y. Noël, M. Causà, M. Rérat, and B. Kirt- man, “CRYSTAL14: A program for the ab initio investigation of crystalline solids,” International Journal of Quantum Chemistry, vol. 114, no. 19, pp. 1287–1317, 2014. [61] R. Dennington, T. A. Keith, and J. M. Milliam, “GaussView,” Software version 5.0.9, 2008. [62] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery-Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, and D. J. Fox, “Gaussian,” Software version 9.5, 2010. [63] O. V. Dolomanov, L. J. Bourhis, R. J. Gildea, J. A. K. Howard, and H. Puschmann, “OLEX2: a complete structure solution, refinement and analysis program,” Journal of Applied Crystallography, vol. 42, no. 2, pp. 339–341, 2009. [64] N. Doebelin and R. Kleeberg, “Profex: A graphical user interface for the Rietveld refinement program BGMN,” Journal of Applied Crystallography, vol. 48, no. 5, pp. 1573–1580, 2015. [65] J. Zhu, C. Dahlstrand, J. R. Smith, S. Villaume, and H. Ottosson, “On the importance of clar structures of polybenzenoid hydrocarbons as revealed by the π-contribution to the electron localization function,” Symmetry, vol. 2, no. 3, pp. 1653–1682, 2010. [66] A. Y. Ganin, Y. Takabayashi, P. Jeglic,ˇ D. Arcon,ˇ A. Potocnik,ˇ P. J. Baker, Y. Ohishi, M. T. McDonald, M. D. Tzirakis, A. McLennan, G. R. Darling, M. Takata, M. J. Rosseinsky, and K. Prassides, “Polymorphism control of superconductivity and magnetism in Cs3C60 close to the Mott transition,” Nature, vol. 466, no. 7303, pp. 221–225, 2010. 156 BIBLIOGRAPHY [67] J. Bernstein, Polymorphism in Molecular Crystals. Oxford: Oxford University Press, 2nd ed., 2010. [68] G. Giri, E. Verploegen, S. C. Mannsfeld, S. Atahan-Evrenk, H. Kim do, S. Y. Lee, H. A. Becerril, A. Aspuru-Guzik, M. F. Toney, and Z. Bao, “Tuning charge transport in solution- sheared organic semiconductors using lattice strain,” Nature, vol. 480, no. 7378, p. 504, 2011. [69] E. Beaugnon, D. Bourgault, D. Braithwaite, P. de Rango, R. P. de la Bathie, A. Sulpice, and R. Tournier, “Material processing in high static magnetic field. A review of an experimental study on levitation, phase separation, convection and texturation,” Journal de Physique I, vol. 3, no. 2, pp. 399–421, 1993. [70] A. Nakamura, J. Ohtsuka, K. Miyazono, A. Yamamura, K. Kubota, R. Hirose, N. Hirota, M. Ataka, Y. Sawano, and M. Tanokura, “Improvement in Quality of Protein Crystals Grown in a High Magnetic Field Gradient,” Crystal Growth & Design, vol. 12, no. 1, pp. 1141–1150, 2012. [71] N. Micali, H. Engelkamp, P. G. van Rhee, P. C. M. Christianen, L. M. Scolaro, and J. C. Maan, “Selection of supramolecular chirality by application of rotational and magnetic forces,” Nature Chemistry, vol. 4, no. 3, pp. 201–207, 2012. [72] M. I. Boamfa, S. V. Lazarenko, E. C. M. Vermalen, A. Kirilyuk, and T. Rasing, “Magnetic field alignment of liquid crystals for fast display applications,” Advanced Materials, vol. 17, no. 5, pp. 610–614, 2005. [73] M. Gopinadhan, P. Deshmukh, Y. Choo, P. W. Majewski, O. Bakajin, M. Elimelech, R. M. Kasi, and C. O. Osuji, “Thermally switchable aligned nanopores by magnetic-field directed self-assembly of block copolymers,” Advanced Materials, vol. 26, no. 30, pp. 5148–5154, 2014. [74] M. B. J. Atkinson, D. K. Bwambok, J. Chen, P. D. Chopade, M. M. Thuo, C. R. Mace, K. A. Mirica, A. A. Kumar, A. S. Myerson, and G. M. Whitesides, “Using magnetic levitation to separate mixtures of crystal polymorphs,” Angewandte Chemie - International Edition, vol. 52, no. 39, pp. 10208–10211, 2013. [75] S. Honjo, M. Yokota, N. Doki, and K. Shimizu, “Magnetic Field Influence on the Crystal Structure of 2,2':6',2''-Terpyridine,” Kagaku Kogaku Ronbunshu, vol. 34, no. 3, pp. 383– 387, 2008. [76] A. Seidell, Solubilities Of Organic Compounds Vol - II. New York: D. Van Norstrand Company, 3rd ed., 1941. 157 BIBLIOGRAPHY [77] R. Rieger and K. Müllen, “Forever young: Polycyclic aromatic hydrocarbons as model cases for structural and optical studies,” Journal of Physical Organic Chemistry, vol. 23, no. 4, pp. 315–325, 2010. [78] S. Xiao, M. Myers, Q. Miao, S. Sanaur, K. Pang, M. L. Steigerwald, and C. Nuckolls, “Molec- ular wires from contorted aromatic compounds,” Angewandte Chemie - International Edition, vol. 44, no. 45, pp. 7390–7394, 2005. [79] K. Bhattacharyya, T. K. Mukhopadhyay, and A. Datta, “Controlling electronic effects and intermolecular packing in contorted polyaromatic hydrocarbons (c-PAHs): towards high mobility field effect transistors,” Physical Chemistry Chemical Physics, vol. 18, no. 22, pp. 14886–14893, 2016. [80] J. Trotter, “The crystal structure of 1:14-benzbisanthrene,” Acta Crystallographica, vol. 11, no. 6, pp. 423–428, 1958. [81] V. A. Nefedov, “Pyrene dimerization into 1,1'-dipyrenyl,” Russian Journal of Organic Chemistry, vol. 43, no. 8, pp. 1163–1166, 2007. [82] J. Wang, H. Wu, L. Jin, J. Zhang, Y. Yuan, and P. Wang, “A Perylene-Based Polycyclic Aromatic Hydrocarbon Electron Donor for a Highly Efficient Solar Cell Dye,” Chem- SusChem, vol. 10, no. 14, pp. 2962–2967, 2017. [83] I. G. Lezama and A. F. Morpurgo, “Progress in organic single-crystal field-effect transistors,” MRS Bulletin, vol. 38, no. 01, pp. 51–56, 2013. [84] T. Hasegawa and J. Takeya, “Organic field-effect transistors using single crystals,” Science and Technology of Advanced Materials, vol. 10, no. 2, p. 024314, 2009. [85] L. Loots and L. J. Barbour, “A simple and robust method for the identification of π–π packing motifs of aromatic compounds,” CrystEngComm, vol. 14, no. 1, pp. 300–304, 2012. [86] B. Schatschneider, J. Phelps, and S. Jezowski, “A new parameter for classification of polycyclic aromatic hydrocarbon crystalline motifs: a Hirshfeld surface investigation,” CrystEngComm, vol. 13, no. 24, p. 7216, 2011. [87] E. R. T. Tiekink and J. Zukerman-Schpector, The Importance of Pi-Interactions in Crystal Engineering: Frontiers in Crystal Engineering. John Wiley & Sons, Ltd., 2012. [88] L. Farina, A. Brillante, R. G. Della Valle, E. Venuti, M. Amboage, and K. Syassen, “Pressure- induced phase transition in pentacene,” Chemical Physics Letters, vol. 375, no. 5-6, pp. 490–494, 2003. 158 BIBLIOGRAPHY [89] T. Siegrist, C. Besnard, S. Haas, M. Schiltz, P. Pattison, D. Chernyshov, B. Batlogg, and C. Kloc, “A polymorph lost and found: The high-temperature crystal structure of pentacene,” Advanced Materials, vol. 19, no. 16, pp. 2079–2082, 2007. [90] C. C. Mattheus, A. B. Dros, J. Baas, G. T. Oostergetel, A. Meetsma, J. L. De Boer, and T. T. M. Palstra, “Identification of polymorphs of pentacene,” Synthetic Metals, vol. 138, no. 3, pp. 475–481, 2003. [91] A. M. Hiszpanski, R. M. Baur, B. Kim, N. J. Tremblay, C. Nuckolls, A. R. Woll, and Y. L. Loo, “Tuning polymorphism and orientation in organic semiconductor thin films via post-deposition processing,” Journal of the American Chemical Society, vol. 136, no. 44, pp. 15749–15756, 2014. [92] M. Botoshansky, F. H. Herbstein, and M. Kapon, “Towards a complete description of a polymorphic crystal: The example of perylene redetermination of the structures of the (Z = 2 and 4) polymorphs,” Helvetica Chimica Acta, vol. 86, no. 4, pp. 1113–1128, 2003. [93] W. S. Hu, S. Z. Weng, Y. T. Tao, H. J. Liu, and H. Y. Lee, “Oriented growth of rubrene thin films on aligned pentacene buffer layer and its anisotropic thin-film transistor characteristics,” Organic Electronics: physics, materials, applications, vol. 9, no. 3, pp. 385–395, 2008. [94] A. Hamaguchi, T. Negishi, Y. Kimura, Y. Ikeda, K. Takimiya, S. Z. Bisri, Y. Iwasa, and T. Shiro, “Single-crystal-like organic thin-film transistors fabricated from dinaphtho[2,3- b:2',3'-f ]thieno[3,2-b]thiophene (dntt) precursor–polystyrene blends.,” Advanced Mate- rials, vol. 27, no. 42, pp. 6606–6611, 2015. [95] Y. Zhao and D. G. Truhlar, “A Prototype for graphene material simulation: Structures and interaction potentials of coronene dimers,” Journal of Physical Chemistry C, vol. 112, no. 11, pp. 4061–4067, 2008. [96] G. L. Piotrovsky, “Karpatite – a new organic mineral from Transcarpathia,” Mineralogich- eskii Sbornik, vol. 9, pp. 120–127, 1955. [97] T. Echigo, M. Kimata, and T. Maruoka, “Crystal-chemical and carbon-isotopic characteris- tics of karpatite (C24H12) from the Picacho Peak Area, San Benito County, California: Evidences for the hydrothermal formation,” American Mineralogist, vol. 92, no. 8-9, pp. 1262–1269, 2007. [98] J. I. Aihara, “Hückel-like rule of superaromaticity for charged paracyclophanes,” Chemical Physics Letters, vol. 381, no. 1-2, pp. 147–153, 2003. 159 BIBLIOGRAPHY [99] W. Schmidt, “Handbook of Combustion: Polycyclic Aromatic Hydrocarbons and Astro- physics,” Handbook of Combustion, Vol. 2: Combustion Diagnostics and Pollutants, vol. 2, pp. 149–164, 1986. [100] R. Scholl and K. Meyer, “Synthese des anti-diperi-dibenz-coronens und dessen abbau zum coronen (hexabenzo-benzol).(mitbearbeitet von horst v. hoeßle und solon brissimdji),” Berichte der deutschen chemischen Gesellschaft (A and B Series), vol. 65, no. 5, pp. 902– 915, 1932. [101] M. S. Newman, “A New Synthesis of Coronene,” Journal of the American Chemical Society, vol. 62, no. 7, pp. 1683–1687, 1940. [102] W. Baker, J. F. W. McOmie, and W. K. Warburton, “Eight- and Higher-membered Ring Compounds. Part IX. Improved Synthesis of Coronene.,” Journal of the Chemical Society, no. V, p. 2991, 1952. [103] P. Jessup and J. Reiss, “Cyclophanes. VII. Naphthalene and phenanthrene cyclophanes for the synthesis of coronene,” Australian Journal of Chemistry, vol. 30, no. 4, p. 843, 1977. [104] J. T. M. van Dijk, A. Hartwijk, A. C. Bleeker, J. Lugtenburg, and J. Cornelisse, “Gram Scale Synthesis of Benzo[ghi]perylene and Coronene,” The Journal of Organic Chemistry, vol. 61, no. 13, p. 1136, 1996. [105] H. C. Shen, J. M. Tang, H. K. Chang, C. W. Yang, and R. S. Liu, “Short and efficient synthesis of coronene derivatives via ruthenium-catalyzed benzannulation protocol,” Journal of Organic Chemistry, vol. 70, no. 24, pp. 10113–10116, 2005. [106] T. M. Krygowski, M. Cyranski,´ A. Ciesielski, B. Swirska,´ and P. Leszczynski,´ “Separation of the energetic and geometric contributions to aromaticity. 2. analysis of the aromatic character of benzene rings in their various topological environments in the benzenoid hydrocarbons. crystal and molecular structure of coronene,” Journal of Chemical Information and Computer Science, vol. 2338, no. 1, pp. 1135–1141, 1996. [107] W. E. Moffitt and C. A. Coulson, “The Electronic Structure and Bond Lengths of Coronene and Pyrene,” Proceedings of the Physical Society, vol. 60, no. 340, pp. 309–315, 1948. [108] G. E. Bacon, N. A. Curry, and S. A. Wilson, “A Crystallographic Study of Solid Benzene by Neutron Diffraction,” Proceedings of the Royal Society A, vol. 279, no. 1376, pp. 98–110, 1964. [109] K. Harigaya and T. Enoki, “Mechanism of magnetism in stacked nanographite with open shell electrons,” Chemical Physics Letters, vol. 351, no. 1-2, pp. 128–134, 2002. 160 BIBLIOGRAPHY [110] M. Darvish Ganji, S. M. Hosseini-khah, and Z. Amini-tabar, “Theoretical insight into hydrogen adsorption onto graphene: a first-principles B3LYP-D3 study,” Physical Chemistry Chemical Physics, vol. 17, no. 4, pp. 2504–2511, 2015. [111] M. Rubio, O. Enrique, and J. Sánchez-Marín, “A Study of Coronene–Coronene Association Using Atom–Atom Pair Potentials,” International Journal of Quantum Chemistry, vol. 57, pp. 567–573, 1996. [112] E. Steiner and P. Fowler, “Current-density maps as probes of aromaticity: Global and Clar π ring currents in totally resonant polycyclic aromatic hydrocarbons,” Faraday Discussions, vol. 135, p. 309, 2007. [113] J. Li, Y. Liu, Y. Qian, L. Li, L. Xie, J. Shang, T. Yu, M. Yi, and W. Huang, “Describing curved- planar π-π interactions: modeled by corannulene, pyrene and coronene.,” Physical Chemistry Chemical Physics, vol. 15, no. 30, pp. 12694–701, 2013. [114] R. Rieger, M. Kastler, V. Enkelmann, and K. Müllen, “Entry to coronene chemistry - Making large electron donors and acceptors,” Chemistry - A European Journal, vol. 14, no. 21, pp. 6322–6325, 2008. [115] V. Z. Kresin, V. A. Litovchenko, and A. G. Panasenko, “Effects related to pair correlation of π-electrons,” The Journal of Chemical Physics, vol. 63, no. 8, p. 3613, 1975. [116] S. Yoshimoto, R. Narita, M. Wakisaka, and K. Itaya, “The structure of a coronene adlayer formed in a benzene solution: Studies by in situ STM and ex situ LEED,” Journal of Electroanalytical Chemistry, vol. 532, no. 1-2, pp. 331–335, 2002. [117] E. Steiner, P. W. Fowler, and L. W. Jenneskens, “Counter-rotating ring currents in coronene and corannulene,” Angewandte Chemie - International Edition, vol. 40, no. 2, pp. 362– 366, 2001. [118] M. Randic, M. Novic, M. Vracko, D. Vukicevic, and D. Plavsic, “π-electron currents in polycyclic conjugated hydrocarbons: Coronene and its isomers having five and seven member rings,” International Journal of Quantum Chemistry, vol. 112, no. 4, pp. 972– 985, 2012. [119] A. H. Matsui and K. Mizuno, “Crystallization and excitonic luminescence of coronene crystals,” Journal of Physics D, vol. 26, pp. B242–B244, 1999. [120] S. Nakatani, T. Nakamura, K. Mizuno, and A. H. Matsui, “Interband and intraband exciton scattering in coronene crystal,” Journal of Luminescence, vol. 58, no. 1-6, pp. 343–346, 1994. 161 BIBLIOGRAPHY [121] A. H. Matsui, “Excitonic processes in aromatic molecular crystals of strong exciton-phonon coupling,” Pure and Applied Chemistry, vol. 67, no. 3, pp. 429–436, 1995. [122] A. Ghosh, K. V. Rao, S. J. George, and C. N. R. Rao, “Noncovalent functionalization, exfoliation, and solubilization of graphene in water by employing a fluorescent coronene carboxylate,” Chemistry - A European Journal, vol. 16, no. 9, pp. 2700–2704, 2010. [123] F. Würthner, C. Thalacker, S. Diele, and C. Tschierske, “Fluorescent J-type aggregates and thermotropic columnar mesophases of perylene bisimide dyes,” Chemistry - A European Journal, vol. 7, no. 10, pp. 2245–2253, 2001. [124] S. Kurata, H. Hirano, and M. Nagai, “Development of fluorescent markers using polycyclic aromatic hydrocarbons with vaseline.,” Journal of Forensic Science, vol. 47, pp. 244–253, 2002. [125] Y. Kubozono, H. Mitamura, X. Lee, X. He, Y. Yamanari, Y. Takahashi, Y. Suzuki, Y. Kaji, R. Eguchi, K. Akaike, T. Kambe, H. Okamoto, A. Fujiwara, T. Kato, T. Kosugi, and H. Aoki, “Metal-intercalated aromatic hydrocarbons: a new class of carbon-based superconductors,” Physical Chemistry Chemical Physics, vol. 13, no. 37, p. 16476, 2011. [126] T. Kambe, X. He, Y. Takahashi, Y. Yamanari, K. Teranishi, H. Mitamura, S. Shibasaki, K. Tomita, R. Eguchi, H. Goto, Y. Takabayashi, T. Kato, A. Fujiwara, T. Kariyado, H. Aoki, and Y. Kubozono, “Synthesis and physical properties of metal-doped picene solids,” Physical Review B, vol. 86, no. 21, 2012. [127] G. A. Artioli and L. Malavasi, “Superconductivity in metal-intercalated aromatic hydrocar- bons,” Journal of Materials Chemistry C, vol. 2, pp. 1577–1584, 2014. [128] E. H. Lee, “A practical guide to pharmaceutical polymorph screening & selection,” Asian Journal of Pharmaceutical Sciences, vol. 9, no. 4, pp. 163–175, 2014. [129] J. Potticary, “An Investigation into Emergent Exotic Magnetism in Coronene,” Master’s thesis, The University of Bristol, 2015. [130] R. M. Ibberson, M. T. Telling, and S. Parsons, “Structure determination and phase transition behaviour of dimethyl sulfate,” Electrochemical and Solid-State Letters, vol. 9, no. 5, pp. 280–286, 2006. [131] F. Moosavi and M. Gholizadeh, “Magnetic effects on the solvent properties investigated by molecular dynamics simulation,” Journal of Magnetism and Magnetic Materials, vol. 354, pp. 239–247, 2014. 162 BIBLIOGRAPHY [132] T. Steiner and G. R. Desiraju, “Distinction between the weak hydrogen bond and the van der Waals interaction,” Chemical Communications, vol. 0, no. 8, pp. 891–892, 1998. [133] F. H. Allen, “A systematic pairwise comparison of geometric parameters obtained by X-ray and neutron diffraction,” Acta Crystallographica Section B, vol. 42, no. 5, pp. 515–522, 1986. [134] T. Yamamoto, S. Nakatani, T. Nakamura, K. Mizuno, A. H. Matsui, Y. Akahama, and H. Kawamura, “Exciton-phonon coupling and pressure-induced structural phase changes in coronene crystals,” Chemical Physics, vol. 184, no. 1-3, pp. 247–254, 1994. [135] K. Hummer, P. Puschnig, and C. Ambrosch-Draxl, “Lowest optical excitations in molecular crystals: Bound excitons versus free electron-hole pairs in anthracene,” Physical Review Letters, vol. 92, no. 14, p. 147402, 2004. [136] P. Cudazzo, F. Sottile, A. Rubio, and M. Gatti, “Exciton dispersion in molecular solids.,” Journal of Physics, vol. 27, no. 11, p. 113204, 2015. [137] N. Jonathan, S. Gordon, and B. P. Dailey, “Chemical Shifts and Ring Currents in Condensed Ring Hydrocarbons,” Journal of Chemical Physics, vol. 36, no. 9, pp. 2443–2448, 1962. [138] E. Matito, J. Poater, M. Duran, and M. Solà, “An analysis of the changes in aromaticity and planarity along the reaction path of the simplest Diels-Alder reaction. Exploring the validity of different indicators of aromaticity,” Journal of Molecular Structure, vol. 727, no. 1-3 SPEC. ISS., pp. 165–171, 2005. [139] S. R. Byrn, G. Gray, R. R. Pfeiffer, and J. Frye, “Analysis of solid-state Carbon-13 NMR spectra of polymorphs (benoxaprofen and nabilone) and pseudopolymorphs (cefazolin).,” Journal of Pharmaceutical Sciences, vol. 74, no. 5, pp. 565–8, 1985. [140] D. Li, G. Wang, H.-C. Cheng, C.-Y. Chen, H. Wu, Y. Liu, Y. Huang, and X. Duan, “Size- dependent phase transition in methylammonium lead iodide perovskite microplate crystals,” Nature Communications, vol. 7, p. 11330, 2016. [141] P. D. Todorov, L. W. Jenneskens, and J. H. van Lenthe, “Assignment of phantom bands in the solid-state infrared and Raman spectra of coronene: The importance of a minute out-of-plane distortion,” Journal of Chemical Physics, vol. 132, no. 3, p. 034504, 2010. [142] M. T. Rogers, “The Magnetic Anisotropy of Coronene, Naphthazarin and Other Crystals,” Journal of the American Chemical Society, vol. 69, no. 1933, pp. 1506–1508, 1945. [143] H. Akamatu and M. Kinoshita, “On the Magnetic Susceptibility of Hexabenzocoronene and Triphenylene,” Bulletin of the Chemical Society of Japan, vol. 32, no. 7, pp. 773–774, 1959. 163 BIBLIOGRAPHY [144] T. L. Makarova, B. Sundqvist, R. Höhne, P. Esquinazi, Y. Kopelevich, P. Scharff, V. A. Davydov, L. S. Kashevarova, and A. V. Rakhmanina, “Magnetic carbon,” Nature, vol. 413, no. 6857, pp. 716–718, 2001. [145] T. L. Makarova, B. Sundqvist, R. Höhne, P. Esquinazi, Y. Kopelevich, P. Scharff, V. Davydov, L. S. Kashevarova, and A. V. Rakhmanina, “Retraction: Magnetic carbon,” Nature, vol. 440, no. 7084, pp. 707–707, 2006. [146] R. Mitsuhashi, Y. Suzuki, Y. Yamanari, H. Mitamura, T. Kambe, N. Ikeda, H. Okamoto, A. Fujiwara, M. Yamaji, N. Kawasaki, Y. Maniwa, and Y. Kubozono, “Superconductivity in alkali-metal-doped picene,” Nature, vol. 464, no. 7285, pp. 76–79, 2010. [147] S. Heguri, M. Kobayashi, and K. Tanigaki, “Questioning the existence of superconducting potassium doped phases for aromatic hydrocarbons,” Physical Review B, vol. 92, no. 1, p. 014502, 2015. [148] P. Sanderson, S. S. Su, I. T. Chang, J. M. Delgado Saborit, D. M. Kepaptsoglou, R. J. Weber, and R. M. Harrison, “Characterisation of iron-rich atmospheric submicrometre particles in the roadside environment,” Atmospheric Environment, vol. 140, pp. 167–175, 2016. [149] A. S. Myerson and B. L. Trout, “Nucleation from Solution,” Science, vol. 341, no. 6148, pp. 855–856, 2013. [150] C. Feng, C. S. Lin, W. Fan, R. Q. Zhang, and M. A. Van Hove, “Stacking of polycyclic aromatic hydrocarbons as prototype for graphene multilayers, studied using density functional theory augmented with a dispersion term,” Journal of Chemical Physics, vol. 131, no. 19, p. 194702, 2009. [151] D. Chen, T. S. Totton, J. W. J. Akroyd, S. Mosbach, and M. Kraft, “Size-dependent melting of polycyclic aromatic hydrocarbon nano-clusters: A molecular dynamics study,” Carbon, vol. 67, pp. 79–91, 2014. [152] A. Petr, D. Wei, C. Kvarnström, A. Ivaska, and L. Dunsch, “π-Dimer of an Aniline Dimer: An ESR–UV–Vis Spectroelectrochemical Study,” The Journal of Physical Chemistry B, vol. 111, no. 43, pp. 12395–12398, 2007. [153] J. C. Fetzer, “The Chemistry and Analysis of Large PAHs,” Polycyclic Aromatic Compounds, vol. 27, no. 2, pp. 143–162, 2007. [154] J. W. Patterson, “The Ultraviolet Absorption Spectra of Coronene,” Journal of the American Chemical Society, vol. 64, no. 6, pp. 1485–1486, 1942. [155] A. He, X. Kang, Y. Xu, I. Noda, Y. Ozaki, and J. Wu, “Investigation on intermolecular interaction between berberine and β-cyclodextrin by 2D UV–Vis asynchronous spectra,” Spectrochimica Acta - Part A, vol. 185, pp. 343–348, 2017. 164 BIBLIOGRAPHY [156] J. Zaccaro, J. Matic, A. S. Myerson, and B. A. Garetz, “Nonphotochemical, Laser-Induced Nucleation of Supersaturated Aqueous Glycine Produces Unexpected γ-Polymorph,” Crystal Growth & Design, vol. 1, no. 1, pp. 5–8, 2001. [157] B. A. Garetz, J. E. Aber, N. L. Goddard, R. G. Young, and A. S. Myerson, “Nonphotochemical, polarization-dependent, laser-induced nucleation in supersaturated aqueous urea solutions,” Physical Review Letters, vol. 77, no. 16, pp. 3475–3476, 1996. [158] D. Voss, “The 110% Solution,” Science, vol. 274, no. 5291, p. 1325, 1996. [159] P. E. Hansen, “13C NMR of polycyclic aromatic compounds. A review,” Organic Magnetic Resonance, vol. 12, no. 3, pp. 109–142, 1979. [160] G. R. Fulmer, A. J. M. Miller, N. H. Sherden, H. E. Gottlieb, A. Nudelman, B. M. Stoltz, J. E. Bercaw, and K. I. Goldberg, “NMR chemical shifts of trace impurities: Common laboratory solvents, organics, and gases in deuterated solvents relevant to the organometallic chemist,” Organometallics, vol. 29, no. 9, pp. 2176–2179, 2010. [161] F. J. Santos, C. A. Nieto De Castro, J. H. Dymond, N. K. Dalaouti, M. J. Assael, and A. Nagashima, “Standard reference data for the viscosity of toluene,” Journal of Physical and Chemical Reference Data, vol. 35, no. 1, pp. 1–8, 2006. [162] Y. Jiang, X. Zhong, W. Shi, Q. Peng, H. Geng, Y. Zhao, and Z. Shuai, “Nuclear quantum tunnelling and carrier delocalization effects to bridge the gap between hopping and bandlike behaviors in organic semiconductors,” Nanoscale Horizons, vol. 1, no. 1, pp. 53– 59, 2016. [163] J. A. A. J. Perenboom, S. A. J. Wiegers, P. C. M. Christianen, U. Zeitler, and J. C. Maan, “Research in High Magnetic Fields: The Installation at the University of Nijmegen,” Journal of Low Temperature Physics, vol. 133, no. 1-2, pp. 181–201, 2003. [164] C. Pareja-Rivera, M. Cuéllar-Cruz, N. Esturau-Escofet, N. Demitri, M. Polentarutti, V. Stojanoff, and A. Moreno, “Recent Advances in the Understanding of the Influence of Electric and Magnetic Fields on Protein Crystal Growth,” Crystal Growth & Design, vol. 17, no. 1, pp. 135–145, 2017. [165] A. Nakamura, J. Ohtsuka, T. Kashiwagi, N. Numoto, N. Hirota, T. Ode, H. Okada, K. Nagata, M. Kiyohara, E. I. Suzuki, A. Kita, H. Wada, and M. Tanokura, “In-situ and real-time growth observation of high-quality protein crystals under quasi-microgravity on earth,” Scientific Reports, vol. 6, no. October 2015, pp. 1–5, 2016. [166] G. Sazaki, E. Yoshida, H. Komatsu, T. Nakada, S. Miyashita, and K. Watanabe, “Effects of a magnetic field on the nucleation and growth of protein crystals,” Journal of Crystal Growth, vol. 173, no. 1-2, pp. 231–234, 1997. 165 BIBLIOGRAPHY [167] A. T. Karunanithi, C. Acquah, L. E. K. Achenie, S. Sithambaram, S. L. Suib, and R. Gani, “An experimental verification of morphology of ibuprofen crystals from CAMD designed solvent,” Chemical Engineering Science, vol. 62, no. 12, pp. 3276–3281, 2007. [168] J. Chen, J. Wang, J. Ulrich, Q. Yin, and L. Xue, “Effect of solvent on the crystal structure and habit of hydrocortisone,” Crystal Growth & Design, vol. 8, no. 5, pp. 1490–1494, 2008. [169] T. Łada, A. Paszewin, R. Molinski,´ A. Morawski, and W. Pachla, “Investigation of phase stability within the YBCO system under pressure of nitrogen and argon,” Applied Superconductivity, vol. 1, no. 3-6, pp. 591–597, 1993. [170] Y. Kamihara, T. Watanabe, M. Hirano, and H. Hosono, “Iron-based layered superconductor La[O1−xFx]FeAs (x= 0.05-0.12) with Tc = 26 K,” Journal of the American Chemical Society, vol. 130, no. 11, pp. 3296–3297, 2008. [171] M. K. Wu, J. R. Ashburn, C. J. Torng, P. H. Hor, R. L. Meng, L. Gao, Z. J. Huang, Y. Q. Wang, and C. W. Chu, “Superconductivity at 93 K in a new mixed-phase Yb–Ba–Cu–O compound system at ambient pressure,” Physical Review Letters, vol. 58, no. 9, pp. 908– 910, 1987. [172] K. S. Martirosyan and D. Luss, “Carbon combustion synthesis of complex oxides: Process demonstration and features,” American Institute of Chemical Engineers, vol. 51, no. 10, pp. 2801–2810, 2005. [173] J. Prado-Gonjal, R. Schmidt, and E. Morán, “Microwave-Assisted Routes for the Synthesis of Complex Functional Oxides,” Inorganics, vol. 3, no. 2, pp. 101–117, 2015. [174] C. Thelander, P. Agarwal, S. Brongersma, J. Eymery, L. F. Feiner, A. Forchel, M. Scheffler, W. Riess, B. J. Ohlsson, U. Gösele, and L. Samuelson, “Nanowire-based one-dimensional electronics,” Materials Today, vol. 9, no. 10, pp. 28–35, 2006. [175] D. C. Green, M. R. Lees, and S. R. Hall, “Designed 3D architectures of high-temperature superconductors,” Chemical Communications, vol. 49, no. 29, p. 2974, 2013. [176] Z. Zhang, H. Suo, L. Ma, A. Kursumovic, Y. Xu, M. Liu, J. L. Macmanus-Driscoll, and S. C. Wimbush, “Biopolymer-mediated inorganic crystal growth at reduced temperature with enhanced kinetics and morphological direction,” Crystal Growth & Design, vol. 14, no. 5, pp. 2296–2301, 2014. [177] S. R. Hall, Biotemplating: Complex Structures from Natural Materials. World Scientific Publishing Co., 1st ed., 2009. ISBN:1908978392. 166 BIBLIOGRAPHY [178] S. R. Hall, “Biomimetic synthesis of high-Tc, type-II superconductor nanowires,” Advanced Materials, vol. 18, no. 4, pp. 487–490, 2006. [179] D. Walsh, S. C. Wimbush, and S. R. Hall, “Use of the Polysaccharide Dextran as a Mor- phological Directing Agent in the Synthesis of High-Tc Superconducting YBa2Cu3O7−δ Sponges with Improved Critical Current Densities,” Chemistry of Materials, vol. 19, no. 4, pp. 647–649, 2007. [180] E. Culverwell, S. C. Wimbush, and S. R. Hall, “Biotemplated synthesis of an ordered macroporous superconductor with high critical current density using a cuttlebone template,” Chemical Communications, no. 9, pp. 1055–1057, 2008. [181] S. R. Hall, S. C. Wimbush, Y. Shida, and W. Ogasawara, “Biotemplated synthesis of superconducting plate-like YBa2Cu3O7−δ using oligosaccharides,” Chemical Physics Letters, vol. 507, no. 1-3, pp. 144–150, 2011. [182] D. P. DiVincenzo and IBM, “The Physical Implementation of Quantum Computation,” Fortschritte der Physik, vol. 48, no. 9-11, pp. 771–783, 2000. [183] S. Stanisic, N. Linden, A. Montanaro, and P. S. Turner, “Generating entanglement with linear optics,” Physical Review A, vol. 96, no. 4, p. 043861, 2017. [184] X. Duan, Y. Huang, Y. Cui, J. Wang, and C. M. Lieber, “Indium phosphide nanowires as building blocks for nanoscale electronic and optoelectronic devices,” Nature, vol. 409, no. 6816, pp. 66–69, 2001. [185] A. M. Morales and C. M. Lieber, “A Laser Ablation Method for the Synthesis of Crystalline Semiconductor Nanowires,” Science, vol. 279, no. 5348, pp. 208–211, 1998. [186] M. H. Huang, Y. Wu, H. Feick, N. Tran, E. Weber, and P. Yang, “Catalytic growth of zinc oxide nanowires by vapor transport,” Advanced Materials, vol. 13, no. 2, pp. 113–116, 2001. [187] G. A. Ozin, A. C. Arsenault, L. Cademartiri, Nanochemistry: a chemical approach to nanomaterials. Cambridge: The Royal Society of Chemistry, 2nd ed., 2009. ISBN:978-1-84755-895-4. [188] R. S. Wagner and W. C. Ellis, “Vapor-liquid-solid mechanism of single crystal growth,” Applied Physics Letters, vol. 4, no. 5, pp. 89–90, 1964. [189] K. Nagashima, T. Yanagida, K. Oka, H. Tanaka, and T. Kawai, “Mechanism and control of sidewall growth and catalyst diffusion on oxide nanowire vapor-liquid-solid growth,” Applied Physics Letters, vol. 93, no. 15, 2008. 167 BIBLIOGRAPHY [190] Z. A. C. Schnepp, S. C. Wimbush, S. Mann, and S. R. Hall, “Structural evolution of superconductor nanowires in biopolymer gels,” Advanced Materials, vol. 20, no. 9, pp. 1782–1786, 2008. [191] R. Boston, Z. Schnepp, Y. Nemoto, Y. Sakka, and S. R. Hall, “In Situ TEM Observation of a Microcrucible Mechanism of Nanowire Growth,” Science, vol. 344, no. 6184, pp. 623–626, 2014. [192] I. Matsubara, R. Funahashia, T. Ogura, Y. Hiroshi, K. Tsuru, and T. Kawai, “Growth mechanism of Bi2Sr2CaCu2Ox superconducting whiskers,” Journal of Crystal Growth, vol. 141, pp. 131–140, 1994. [193] H. K. Onnes, “Disappearance of the electrical resistance of mercury at helium tempera- tures,” Proceedings Koninklijke Akademie van Wetenschappen te Amsterdam, vol. 14, pp. 113–115, 1911. [194] J. G. Bednorz and K. A. Müller, “Possible High Tc Superconductivity in the Ba–La–Cu–O System,” in Zeitschrift for Physik B, vol. 64, pp. 267–271, Springer-Verlag, 1986. [195] M. Atikur Rahman, “A Review on Cuprate Based Superconducting Materials Including Characteristics and Applications,” American Journal of Physics and Applications, vol. 3, no. 2, p. 39, 2015. [196] A. Schilling, M. Cantoni, J. D. Guo, and H. R. Ott, “Superconductivity above 130 K in the Hg–Ba–Ca–Cu–O system,” Nature, vol. 363, no. 6424, pp. 56–58, 1993. [197] Z. Schnepp, J. Mitchells, S. Mann, and S. R. Hall, “Biopolymer-mediated synthesis of anisotropic piezoelectric nanorods,” Chemical Communications, vol. 46, no. 27, p. 4887, 2010. [198] D. C. Green, R. Boston, S. Glatzel, M. R. Lees, S. C. Wimbush, J. Potticary, W. Ogasawara, and S. R. Hall, “On the Mechanism of Cuprate Crystal Growth: The Role of Mixed Metal Carbonates,” Advanced Functional Materials, vol. 25, no. 29, pp. 4700–4707, 2015. [199] L. A. Klinkova, V. I. Nikolaichik, N. V. Barkovskii, and K. V. Van, “On the thermal stability ◦ of the barium-copper oxide BaCuO2 at 900–1100 C in air,” Bulletin of the Russian Academy of Sciences: Physics, vol. 73, no. 8, pp. 1107–1109, 2009. [200] M. Muralidhar, N. Sakai, M. Jirsa, and S. Tanaka, “Solid Phase (RE2BaO4)-Liquid Phase (BaCuO2) Reaction: The Way to Highly Oriented ErBa2Cu3Oy Superconducting Thick Films on Commercial Silver Substrates,” Crystal Growth & Design, vol. 9, no. 5, pp. 2404–2408, 2009. 168 BIBLIOGRAPHY [201] R. Boston, K. Awaya, T. Nakayama, W. Ogasawara, and S. R. Hall, “Formation of superconducting yttrium barium copper oxide using sulphur-containing templates,” RSC Advances, vol. 4, no. 51, pp. 26824–26828, 2014. [202] F. M. Ross, J. Tersoff, and M. C. Reuter, “Sawtooth faceting in silicon nanowires,” Physical Review Letters, vol. 95, no. 14, p. 146104, 2005. [203] J. L. Tallon, R. G. Buckley, P. W. Gilberd, M. R. Presland, I. W. M. Brown, M. E. Bowden, L. A. Christian, and R. Goguel, “High-Tc superconducting phases in the series Bi2.1(Ca, Sr)n+1CunO2n+4+δ,” Nature, vol. 333, no. 6169, pp. 153–156, 1988. [204] M. Shalaby, H. Hashem, T. Hammad, L. Wahab, K. Marzouk, and S. Soltan, “Higher critical current density achieved in Bi-2223 High-Tc superconductors,” Journal of Radiation Research and Applied Sciences, vol. 9, no. 3, pp. 345–351, 2016. [205] P. Badica, K. Togano, S. Awaji, K. Watanabe, and H. Kumakura, “Review on Bi-Sr-Ca-Cu-O whiskers,” Superconductor Science and Technology, vol. 19, no. 10, pp. R81–R99, 2006. [206] C. P. Brock and J. D. Dunitz, “Temperature dependence of thermal motion in crystalline naphthalene,” Acta Crystallographica Section B, vol. 38, no. 8, pp. 2218–2228, 1982. [207] R. Mason, “The crystallography of anthracene at 95 K and 290 K,” Acta Crystallographica, vol. 17, no. 5, pp. 547–555, 1964. [208] G. B. Robertson, “Crystal and Molecular Structure of Diphenyl,” Nature, vol. 191, no. 4788, pp. 593–594, 1961. [209] G. Ferraris, D. W. Jones, and J. Yerkess, “A neutron-diffraction study of the crystal and molecular structure of triphenylene,” Zeitschrift für Kristallographie, vol. 138, no. 1-4, pp. 113–128, 1973. [210] P. H. Friedlander and D. Sayre, “Crystal Structure of 1:2-Benzanthracene,” Nature, vol. 178, no. 4540, pp. 999–1000, 1956. [211] D. M. Burns and J. Iball, “Refinement of the Structure of Chrysene,” Proceedings of the Royal Society A, vol. 257, no. 1291, pp. 491–514, 1960. [212] F. L. Hirshfeld, S. Sandler, and G. M. J. Schmidt, “The structure of overcrowded aromatic compounds. Part VI. The crystal structure of benzo[c]phenanthrene and of 1,12- dimethylbenzo[c]phenanthrene,” Journal of the Chemical Society, p. 2108, 1963. [213] A. De, R. Ghosh, S. Roychowdhury, and P. Roychowdhury, “Structural analysis of picene, C22H14,” Acta Crystallographica Section C, vol. 41, no. 6, pp. 907–909, 1985. 169 BIBLIOGRAPHY [214] J. Iball, C. H. Morgan, and D. E. Zacharias, “Refinement of the crystal structure of orthorhombic dibenz[a,h]-anthracene,” Journal of the Chemical Society, no. 12, p. 1271, 1975. [215] Y. Delugeard, J. Desuche, and J. L. Baudour, “Structural transition in polyphenyls. II. The crystal structure of the high-temperature phase of quaterphenyl,” Acta Crystallographica Section B, vol. 32, no. 3, pp. 702–705, 1976. [216] A. C. Hazell, F. K. Larsen, and M. S. Lehmann, “A neutron diffraction study of the crystal structure of pyrene, C16H10,” Acta Crystallographica Section B, vol. 28, no. 10, pp. 2977– 2984, 1972. [217] A. Camerman and J. Trotter, “The Crystal and Molecular Structure of Perylene,” Proceed- ings of the Royal Society A, vol. 279, no. 1376, pp. 129–146, 1964. [218] J. G. White, “The crystal structure of 1:12-benzperylene: a quantitative X-ray investigation,” Journal of the Chemical Society, p. 1398, 1948. [219] B. Hummelink-Peters, T. E. M. van den Hark, J. H. Noordik, and P. T. Beurskens, “Dinaphtho[1,2-a;1',2',-h]anthrecene, C30H18,” Crystal Structure Communications, vol. 4, pp. 281–284, 1975. [220] K. A. Kerr, J. P. Ashmore, and J. C. Speakman, “The Crystal and Molecular Structure of Quaterrylene: A Redetermination,” Proceedings of the Royal Society A, vol. 344, no. 1637, pp. 199–215, 1975. [221] J. Iball, S. N. Scrimgeour, and D. W. Young, “3,4-Benzopyrene (a new refinement),” Acta Crystallographica Section B, vol. 32, no. 1, pp. 328–330, 1976. [222] J. Bregman, F. L. Hirshfeld, D. Rabinovich, and G. M. J. Schmidt, “The crystal structure of [18]annulene, I. X-ray study,” Acta Crystallographica, vol. 19, no. 2, pp. 227–234, 1965. [223] W. N. Lipscomb, J. M. Robertson, and M. G. Rossmann, “Crystal-structure studies of polynuclear hydrocarbons. Part I. 2:3-8:9-dibenzoperylene,” Journal of the Chemical Society, p. 2601, 1959. [224] J. M. Robertson and J. G. White, “The crystal structure of coronene: a quantitative X-ray investigation,” Journal of the Chemical Society, p. 607, 1945. [225] J. M. Robertson and J. Trotter, “Crystal-structure studies of polynuclear hydrocarbons. Part V. 1,2:7,8-dibenzocoronene,” Journal of the Chemical Society, p. 1115, 1961. [226] R. G. Hazell and G. S. Pawley, “The crystal and molecular structure of ovalene: some three-dimensional refinements,” Zeitschrift für Kristallographie, vol. 137, no. 1-6, 1973. 170 BIBLIOGRAPHY [227] J. M. Robertson and J. Trotter, “248. Crystal-structure studies of polynuclear hydrocarbons. Part VI. 1,12:2,3:4,5:6,7:8,9:10,11-hexabenzocoronene,” Journal of the Chemical Society, vol. 20, no. 175, p. 1280, 1961. [228] H. A. Staab, F. Diederich, C. Krieger, and D. Schweitzer, “Cycloarenes, a New Class of Aromatic Compounds, II. Molecular Structure and Spectroscopic Properties of Kekulene,” Chemische Berichte, vol. 116, no. 10, pp. 3504–3512, 1983. [229] P. J. Roberts and G. Ferguson, “β-Tribenzopyrene (naphtho[1,2,3,4-rst]pentaphene),” Acta Crystallographica Section B, vol. 33, no. 4, pp. 1244–1247, 1977. [230] I. Oonishi, S. Fujisawa, J. Aoki, Y. Ohashi, and Y. Sasada, “Conformation of Condensed Poly- cyclic Aromatic Hydrocarbons. II. The Molecular and Crystal Structure of Violanthrene B (Dibenzo[a,rst]naphtho[8,1,2-cde]pentaphene),” Bulletin of the Chemical Society of Japan, vol. 59, no. 7, pp. 2233–2236, 1986. [231] Y. Kohno, M. Konno, Y. Saito, and H. Inokuchi, “The crystal structure of tetra- benzo[a,cd,j,lm]perylene,” Acta Crystallographica Section B, vol. 31, no. 8, pp. 2076– 2080, 1975. [232] I. Oonishi, S. Fujisawa, J. Aoki, and T. Danno, “A Redetermination of Diphenanthro[5,4,3- abcd : 5',4',3'- jklm]perylene,” Bulletin of the Chemical Society of Japan, vol. 51, no. 8, pp. 2256–2260, 1978. [233] S. Fujisawa, I. Oonishi, J. Aoki, Y. Ohashi, and Y. Sasada, “Conformation of Condensed Poly- cyclic Aromatic Hydrocarbons. I. The Molecular and Crystal Structure of Anthra[2,1,9,8- hijkl]benzo[de]naphtho[2,1,8,7-stuv]pentacene,” Bulletin of the Chemical Society of Japan, vol. 55, no. 11, pp. 3424–3428, 1982. [234] A. Williams, G. H. Kwei, R. B. Von Dreele, I. D. Raistrick, and D. L. Bish, “Joint X-ray and neutron refinement of the structure of superconducting YBa2Cu3O7−x: Precision structure, anisotropic thermal parameters, strain, and cation disorder,” Physical Review B, vol. 37, no. 13, pp. 7960–7962, 1988. [235] C. Michel and B. Raveau, “Les oxydes A2BaCuO5(A = Y, Sm, Eu, Gd, Dy, Ho, Er, Yb),” Journal of Solid State Chemistry, vol. 43, no. 1, pp. 73–80, 1982. [236] J. L. García-Muñoz and J. Rodríguez-Carvajal, “Structural characterization of R2Cu2O5(R = Yb, Tm, Er, Y, and Ho) oxides by neutron diffraction,” Journal of Solid State Chemistry, vol. 115, no. 2, pp. 324–331, 1995. [237] C. Santos, K. Strecker, P. A. Suzuki, S. Kycia, O. M. Silva, and C. R. Silva, “Stabilization of α-SiAlONs using a rare-earth mixed oxide (RE2O3) as sintering additive,” Materials Research Bulletin, vol. 40, no. 7, pp. 1094–1103, 2005. 171 BIBLIOGRAPHY [238] J. P. R. De Villiers, “Crystal structures of aragonite , strontianite , and witherite,” The American Mineralogist, vol. 56, pp. 758–767, 1971. [239] D. P. Volanti, M. O. Orlandi, J. Andrés, and E. Longo, “Efficient microwave-assisted hydrothermal synthesis of CuO sea urchin-like architectures via a mesoscale self- assembly,” CrystEngComm, vol. 12, no. 6, p. 1696, 2010. [240] K. Halbach, “Design of Permanent Multiple Magnets with Orientated Rare Earth Cobalt Material,” Nuclear Instruments and Methods, vol. 169, pp. 1–10, 1980. [241] K. Halbach, “Strong Rare Earth Cobalt Quadrupoles,” Transactions on Nuclear Science, no. 3, 1979. [242] H. H. A. Leupold, A. S. A. Tilak, E. Potenziani, and A. S. A. Tilak, “Adjustable multi- tesla permanent magnet field sources,” Transactions on Magnetics, vol. 29, no. 6 pt 1, pp. 2902–2904, 1993. [243] H. Raich and P. Blümler, “Design and construction of a dipolar Halbach array with a homogeneous field from identical bar magnets: NMR mandhalas,” Concepts in Magnetic Resonance Part B: Magnetic Resonance Engineering, vol. 23, no. 1, pp. 16–25, 2004. [244] Webpage:, “High energy magnet,” https://www.supermagnete.de/eng/Magnet- applications/High-energy-magnet?category=blocks_big, accessed: 2017-10-19. [245] D. Meeker, “Finite Element Methods Magnetics,” Software version 4.2, 2012. 172 Altema n l h cheese.” the all and meat the “All oi .Cre,2016 Carter, O. Robin – 173 A PPENDIX A PPENDIX A APPENDIXA.APPENDIX 3 1 2 E -1.8363762527596E+03 DE-4.8E-08( 21) 0.970803966845 E+01 0.604383707763 E-15 −0.130202068334 E+00 0.229042060855 E-17 0.456171094224 E+01 −0.366929206654 E-17 −0.445495023322 E+01 0.918669773212 E-15 0.141779640479 E+02 4 0.100000000000 E+01 0.000000000000 E+00 0.555111512313 E-16 0.000000000000 E+00 0.100000000000 E+01 0.000000000000 E+00 0.000000000000 E+00 0.000000000000 E+00 0.100000000000 E+01 0.000000000000 E+00 0.000000000000 E+00 0.000000000000 E+00 -0.100000000000 E+01 0.000000000000 E+00 -0.555111512313 E-16 -0.122464679915 E-15 0.100000000000 E+01 -0.163687025695 E-15 0.000000000000 E+00 0.000000000000 E+00 -0.100000000000 E+01 0.262654471762 E+01 0.228085547111 E+01 0.702388098982 E+01 -0.100000000000 E+01 0.000000000000 E+00 -0.555111512313 E-16 0.000000000000 E+00 -0.100000000000 E+01 0.000000000000 E+00 0.000000000000 E+00 0.000000000000 E+00 -0.100000000000 E+01 0.000000000000 E+00 0.000000000000 E+00 0.000000000000 E+00 0.100000000000 E+01 0.000000000000 E+00 0.555111512313 E-16 0.122464679915 E-15 -0.100000000000 E+01 0.163687025695 E-15 0.000000000000 E+00 0.000000000000 E+00 0.100000000000 E+01 0.262654471762 E+01 0.228085547111 E+01 0.702388098982 E+01 18 6 -1.858570539950 0.795229605331 -2.005540239958 6 -3.210661603567 1.202086985750 -1.901203526709 6 -3.764583703110 2.203930685080 -2.732160007970 6 4.643863981760 -1.963331957357 -2.716449104588 6 3.794030126874 1.987958712140 -1.766598569892 6 4.658821069599 0.046020116673 -5.490201466554 6 1.614006348444 1.701877886241 -0.758255489619 6 0.498999261637 -1.541118911720 6.876708781859 6 -1.332251724019 -0.143140139516 -1.158073670235 6 -4.028443354701 0.605132035860 -0.916693816396 6 2.763000900102 -1.296671402436 -6.227258135334 6 -0.861346411320 -1.907457877225 6.977512782458 1 -0.554272970493 -1.942117355116 0.725005914081 1 -2.023159061846 1.475532134891 2.335316318505 1 -4.238910637040 1.179391405969 3.348184910117 1 3.140203203907 1.878093186277 3.482848601031 1 1.240293201086 -1.272513339988 2.766766791856 1 0.283363905685 0.431356111921 1.241383862665 FIGURE A.1. A β-coronene fort.34 geometry input file for CRYSTAL14. 174 Coronere EXTERNAL OPTGEOM FULLOPTG ENDOPT END 6 4 0 0 6 2.0 1.00 3047.5249000 0.0018347 457.3695100 0.0140373 103.9486900 0.0688426 29.2101550 0.2321844 9.2866630 0.4679413 3.1639270 0.3623120 0 1 3 4.0 1.00 7.8682724 −0.1193324 0.0689991 1.8812885 −0.1608542 0.3164240 0.5442493 1.1434564 0.7443083 0 1 1 0.0 1.00 0.1687144 1.0000000 1.0000000 0 3 1 0.0 1.00 0.8000000 1.0000000 1 2 0 0 3 1.0 1.00 18.7311370 0.03349460 2.8253937 0.23472695 0.6401217 0.81375733 0 0 1 0.0 1.00 0.1612778 1.0000000 99 0 END DFT HSEsol XLGRID END SCFDIR GRIMME 1.05 20. 25. 2 6 1.75 1.75 1 0.14 1.09 TOLINTEG 7 7 7 7 14 SHRINK 4 4 EXCHSIZE 5000000 BIPOSIZE 3500000 MAXCYCLE 2000 PPAN LEVSHIFT 6 0 FMIXING 30 TOLDEE 7 END FIGURE A.2. A typical input input file for CRYSTAL14. The command ’EXTERNAL’ forces the program to search for the fort.34 (A.1) file in the same folder for starting geometry input. 175 APPENDIXA.APPENDIX ◦ ◦ 3 Form Temp (K) a (Å) b (Å) c (Å) β ( ) πd ∠mol ( ) Volume (Å ) γ 150 10.02 4.67 15.060 106.7 3.43 85.79 699.00 γ 200 10.04 4.681 15.604 106.32 3.43 85.76 703.77 γ 250 10.072 4.690 15.650 106.18 3.44 85.59 710.1 β 80 10.386 3.821 17.211 96.24 3.47 49.7 679 β 150 10.392 3.839 17.229 96.24 3.48 50.0 683 TABLE A.1. Unit cell parameters of the two forms of coronene collected at various temperatures. πd is the stacking distance between coronene molecules and ∠mol is the nearest neighbour angle between molecules. 176 3 −1 −1 Basis set Functional a (%) b (%) c (%) β (%) Volume (Å ) Energy (A.U.) Energy (kJmol )Eγ-Eβ (kJmol ) -3.22 -2.56 -8.22 3.57 -15.33 -1837.951519 -4833812.495 621G -0.48522974 -2.72 -8.06 -3.14 1.07 -13.54 -1837.951335 -4833812.01 B1WC -2.82 -2.33 -4.31 1.60 -10.01 -1839.074783 -4836766.679 631G 0 DNC DNC DNC DNC DNC 0 0 -1.23 4.46 -6.17 1.63 -4.09 -1841.655461 -4843553.862 621G 1.93400206 -1.62 -0.01 -0.01 -0.01 -2.61 -1841.656196 -4843555.796 B3LYP -1.71 2.29 -1.57 0.01 -1.36 -1842.780686 -4846513.205 631G 15.76151636 -1.23 -0.01 0.01 -1.48 -0.01 -1842.786679 -4846528.967 -2.44 0.00 -7.09 3.05 -10.90 -1840.687862 -4841009.078 621G -3.62660168 -2.25 -4.90 -1.74 0.00 -8.71 -1840.686483 -4841005.451 HSE06 -2.15 -1.02 -2.42 0.01 -5.91 -1841.826628 -4844004.032 631G 16.28207489 -1.94 -2.96 -0.00 -1.02 -5.08 -1841.832819 -4844020.314 -3.46 -2.65 -8.58 3.62 -15.98 -1835.265321 -4826747.794 621G -0.53096018 -3.06 -8.17 -3.32 1.03 -14.09 -1835.265119 -4826747.263 HSEsol -3.13 -2.36 -4.76 1.71 -10.80 -1836.376253 -4829669.545 631G 20.15147817 -2.86 -5.13 -1.78 -0.00 -9.41 -1836.383915 -4829689.696 -1.69 2.43 -7.10 3.03 -8.14 -1840.503477 -4840524.145 621G -0.33106703 -1.74 -3.55 -0.01 0.00 -5.97 -1840.503351 -4840523.814 PBE -0.07 -0.01 -0.20 -0.18 -0.58 -1841.669745 -4843591.43 631G -0.15471238 177 -1.15 -2.13 0.00 -0.01 -3.06 -1841.669686 -4843591.274 -2.41 -0.00 -6.82 2.98 -10.89 -1840.708072 -4841062.229 621G -4.85196865 -2.14 -5.00 -1.73 0.00 -8.68 -1840.706227 -4841057.377 PBE0 -2.32 -0.01 -2.51 1.06 -5.88 -1841.853585 -4844074.929 631G 16.71603804 -1.90 -2.95 -0.00 -1.02 -5.04 -1841.859941 -4844091.645 -2.65 -0.01 -6.89 3.01 -11.76 -1840.790561 -4841279.175 621G 0 DNC DNC DNC DNC DNC 0 0 PBE0-13 -2.47 -1.24 -2.55 1.01 -6.67 -1841.932083 -4844281.378 631G 11.61389064 -2.12 -3.27 -0.01 -0.01 -5.79 -1841.936499 -4844292.992 -3.44 -2.81 -8.43 3.57 -15.93 -1835.317978 -4826886.282 621G -1.73435613 -3.03 -8.24 -3.25 0.01 -14.05 -1835.317319 -4826884.548 PBESOL0 -3.25 -2.16 -4.66 1.85 -10.71 -1836.434753 -4829823.399 631G 24.38533107 -2.96 -5.27 -1.70 -0.00 -9.57 -1836.444025 -4829847.785 -2.78 0.00 -6.26 2.41 -9.71 -1837.381324 -4832312.882 621G 4.372827359 -2.77 -3.93 -1.38 -0.00 -7.83 -1837.382987 -4832317.254 WC1LYP DNC DNC DNC DNC DNC 0 0 631G 0 DNC DNC DNC DNC DNC 0 0 TABLE A.2. PBC parameters of γ (grey rows) and β (white rows) calculated using a range of basis sets and functionals within the Crystal14 software package. Cell parameters are given as a percentage of the experimental values collected in table −1 3.2 and lattice energies are given in both A.U. (Hartree) and kJmol .Eγ-Eβ shows the difference in energy between the two forms and is given in kJmol−1. APPENDIXA.APPENDIX (a) (b) (c) (d) FIGURE A.3. Unit cell parameters (a) lengths of the a, b and c axes, (b) monoclinic angle β and (c) cell volume of coronene as a function of temperature calculated using Rietveld analysis. (d) shows the Rwp for each pattern collected. Green circles indicate starting points for the experiments. On all graphs, the red ‘down’ triangles indicate cooling and the black ‘up’ arrows indicate warming. 178 Unit cell Reference Compound Formula Space group Z ∠ (°) short axis (Å) mol Herringbone 108 Benzene C6H6 Pbca 4 6.92 88 206 Naphthalene C10H8 P21/a 2 5.97 52 207 Anthracene C14H10 P21/a 2 6.00 49 25 Phenanthrene C14H10 P21 2 6.16 58 208 Biphenyl C12H10 P21/a 2 5.58 64 209 Triphenylene C18H12 P212121 4 5.26 80 210 Benzanthracene C18H12 P21 2 6.50 45 211 Chrysene C18H12 I2/c 4 5.78 58 212 Benzophenanthrene C18H12 P212121 4 5.78 85 213 Picene C22H14 P21 2 6.15 58 214 Dibenzanthracene C22H14 P21 2 6.59 45 215 Quaterphenyl C24H18 P21/a 2 5.61 66 Sandwich herringbone 216 Pyrene C16H10 P21/a 4 8.47 83 217 Perylene C20H12 P21/a 4 10.26 69 218 Benzperylene C22H12 P21/a 4 9.89 64 219 Dinaphthoanthracene C30H18 P21/c 4 8.16 62 220 Quaterrylene C40H20 P21/a 4 10.63 69 γ herringbone 221 Benzopyrene C20H12 P21/c 4 4.53 73 222 Annulene C18H18 P21/a 2 4.80 79 223 Dibenzoperylene C28H16 A2/a 4 5.23 81 224 Coronene C24H18 P21/a 2 4.70 85 80 Benzobisanthrene C30H14 Pa 4 4.68 - 225 Dibenzocoronene C32H16 C2/c 4 5.22 83 226 Ovalene C32H14 P21/a 2 4.70 86 227 Hexabenzocoronene C42H18 P21/a 2 5.11 86 228 Kekulene C48H24 C2/c 4 4.58 86 β herringbone 229 Tribenzopyrene C28H16 Pn21m 2 4.02 30 230 Violanthrene C34H18 P21/c 4 3.80 - 231 Tetrabenzoperylene C34H18 Pcab 8 7.65 3 232 Diphenanthroperylene C38H18 C2 4 3.83 0 233 Anthrabenzonaphthopentacene C38H18 Pa 4 3.78 29 223 Diperinaphthyleneanthracene C34H18 P21 2 7.83 9 Table A.3: The 32 hydrocarbons used by Desiraju and Gavezzotti’s 1989 paper (reference 28) to identify packing motifs in condensed ring systems. 179 APPENDIXA.APPENDIX Y123 Y211 Y2Cu2O5 Y2O3 BaCO3 CuO Crystal Orthorhombic Orthorhombic Orthorhombic Cubic Orthorhombic Monoclinic system a (Å) 3.82 7.132 10.796 10.608 5.313 4.684 b (Å) 3.88 12.181 3.494 10.608 8.896 3.423 c (Å) 11.68 5.658 12.455 10.608 6.428 5.129 β (°) - - - - - 99.54 Volume (Å3) 173.4 491.5 469.8 1193.7 303.8 81.1 Space group Pmmm Pbnm Pna21 Ia-3 Pmcn C2/c Z 1 4 2 16 4 4 source (ref) 234 235 236 237 238 239 TABLE A.4. Unit cell parameters phases used for identification and analysis of YBCO synthesis. 180 NaCl ◦ Temperature ( C) Y123 er Y2Cu2O5 er Y211 er Y2O3 er BaCO3 er BaCuO2 er CuO er 800 0.0628 0.0065 3E-4 0.0039 2E-4 0.0027 0.2169 0.0053 0.1144 0.0031 0.2876 0.0064 0.3179 0.0055 820 0.157 0.013 3E-4 0.0074 7E-4 0.0087 0.2368 0.0076 0.0494 0.0046 0.2508 0.0093 0.3062 0.0083 840 0.206 0.011 2E-4 0.0038 0.0475 0.0016 0.1902 0.0047 0.0338 0.0027 0.2331 0.0053 0.2902 0.0092 2% 860 0.251 0.035 2E-4 0.0039 0.1286 0.01 0.1224 0.0056 0.0348 0.0034 0.193 0.01 0.27 0.013 880 0.2955 0.0049 1E-4 0.0023 0.168 0.0034 0.0988 0.0031 0.0262 0.0024 0.1447 0.0038 0.2677 0.0036 900 0.3582 0.008 2E-4 0.0033 0.2098 0.0034 0.0202 0.0028 0.0285 0.002 0.1206 0.0033 0.2626 0.0065 920 0.45 0.011 3E-4 0.0053 0.1912 0.0044 0.02 0.0043 0.0246 0.0036 0.0759 0.0043 0.2383 0.0087 800 0.183 0.0077 0.1445 0.0074 2E-4 0.0037 0.0936 0.0035 0.2451 0.0049 0.1138 0.0055 0.2198 0.0048 820 0.2144 0.016 0.1236 0.0071 0.01023 0.0027 0.0193 0.0046 0.1903 0.004 0.23 0.006 0.21217 0.011 840 0.2471 0.0049 0.1061 0.0046 0.0407 0.0033 0.0085 0.003 0.16 0.0038 0.2501 0.0039 0.1875 0.0039 5% 860 0.31159 0.0098 0.0574 0.0051 0.122 0.011 0.0099 0.0029 0.13 0.0029 0.1947 0.0044 0.17441 0.0037 880 0.3866 0.0073 0.0614 0.0066 0.1336 0.0054 0.0014 0.0014 0.1127 0.0064 0.13952 0.0046 0.16478 0.0049 900 0.465 0.026 0.082 0.015 0.1205 0.0075 0.002 0.0082 0.09 0.0092 0.0935 0.0052 0.147 0.013 181 920 0.7451 0.015 0.01928 0.015 0.0206 0.0046 0 0.0021 0.0306 0.0035 0.08202 0.0019 0.1024 0.0035 800 0.23 0.027 0.201 0.012 2E-4 0.0039 0.0027 0.0026 0.2176 0.0083 0.161 0.0073 0.187 0.016 820 0.362 0.023 0.1565 0.0086 0.007 0.0024 0.0075 0.0013 0.1491 0.0066 0.1348 0.0063 0.1831 0.023 840 0.4343 0.0095 0.1275 0.0071 0.0425 0.0093 0.0024 0.0035 0.1392 0.0031 0.0996 0.0033 0.1545 0.0058 10% 860 0.501 0.014 0.141 0.014 0.058 0.012 0 0.0056 0.1034 0.0073 0.1172 0.0074 0.0795 0.0062 880 0.52 0.01 0.15 0.0058 0.0905 0.0054 0.0151 0.0014 0.0809 0.019 0.0735 0.0033 0.07 0.0044 900 0.754 0.01 0.132 0.011 0.0091 0.0018 0 0.002 0.0214 0.0018 0.0405 0.0014 0.043 0.0032 920 0.892 0.015 2E-4 0.0041 0.0132 0.0065 0.0024 0.0018 0.0071 0.0015 0.0188 0.0026 0.022 0.013 800 0.171 0.022 0.246 0.014 3E-4 0.0042 0.0217 0.0069 0.22 0.0073 0.1379 0.0066 0.203 0.015 820 0.2981 0.0068 0.1993 0.0047 0.0056 0.002 0.0166 0.0021 0.1105 0.0033 0.1929 0.0039 0.1769 0.0039 840 0.351 0.021 0.144 0.016 0.0314 0.0039 0.0042 0.002 0.161 0.018 0.145 0.016 0.163 0.019 15% 860 0.4603 0.0066 0.1294 0.0048 0.0329 0.0038 0.0051 0.0014 0.0641 0.002 0.166 0.0029 0.1422 0.0058 880 0.4806 0.0058 0.1251 0.013 0.071 0.011 0.0096 0.0014 0.0818 0.0094 0.133 0.016 0.12 0.014 900 0.6747 0.017 0.0473 0.019 0.0538 0.0041 0 0.0014 0.0327 0.0027 0.0871 0.0036 0.0999 0.0031 920 0.8259 0.0051 0.0086 0.0034 0.0506 0.0024 0 0.0019 0.0173 0.0019 0.0126 0.0015 0.085 0.0022 TABLE A.5. Relative amounts of phases in YBCO nanowire mixtures containing NaCl at various amounts determined using multiphase Reitveld analysis. APPENDIXA.APPENDIX 15% 10% 5% 2% T Na ABLE eprtr 13e Y er Y123 Temperature sn utpaeRivl analysis. Reitveld multiphase using 2 2 .1 .3 .3 .1 .56003 .3 .28002 .E40040.0068 0.074 5.1E-4 0.0058 0.004 0.1469 0.0033 0.0979 0.0045 0 0.0043 0.0018 0.165 0.1577 0.0022 0.0069 0.0061 0.0023 0.0069 0.0248 0.012 0.0028 0.0019 0.0036 0.0065 0.044 0.0222 0.039 0.018 0.0061 0.0089 0.004 0.094 0.0129 0.0491 0.0645 0.0039 0.0035 0.1971 0.0025 0.1899 0.0079 0.106 0.0062 0.0028 0.1157 0.003 0.0046 0.0033 0.0048 0.008 0 0.12 1E-4 0.0183 0.0024 0.1358 0.0037 0.0464 0.0731 0.045 0.0025 0.0049 0.0911 0.0044 0.0364 0.0088 0 0.0035 0.0526 0.0029 0.007 0.0041 0.0062 0.0441 0.0024 0.0056 0.0551 0.008 0.0037 0.148 0.1819 0 0.0317 0.2016 0.0935 0.011 0.1301 0.0058 0.0467 0.006 0.0323 0.083 0.0046 0.005 0.0048 0.0032 0.0055 0.0026 0.0025 0.0022 0.013 0.0056 0.013 2E-4 0.0407 0.033 0.0053 0.0715 0.0169 0.033 0.0834 0.1154 0.0033 3E-4 1E-4 0.0041 0.0026 0.0065 0.0049 0.011 0.0046 0 0.0019 0.034 0.0047 0.172 0 0.0262 0.1996 2E-4 0.01 0.1789 2E-4 0.0042 0.026 0.816 0.0052 0.0021 0.014 1E-4 0.016 0.258 0.25 0.0042 0.0068 0.013 0.794 0.0075 0.217 0.0091 0.0071 0.678 0.003 0.186 1E-4 0.1763 1E-4 0.0056 0.4052 920 0.247 0.0039 0.031 0.0068 0.018 0.0029 0.38 2E-4 4E-4 900 0.0062 0.3694 0.0077 0.016 880 2E-4 0.034 2E-4 0.2757 0.02 860 0.205 0.0096 0.801 0.0067 0.021 840 0.807 0.139 0.014 820 0.622 0.1767 800 0.1578 0.009 0.405 920 0.0066 0.5266 0.01 900 0.3977 880 0.404 860 840 820 800 2 .5 .1 .09003 .57002 .01002 .17002 .05002 .610.0054 0.009 0.1601 0.2002 0.0044 0.0024 0.0034 0.2174 0.0083 0.0046 0.0015 0.0037 0.2862 0.0219 0.2529 0.0051 0.0027 0.004 0.006 0.0087 0.0048 0.0327 0.008 0.1693 0.2089 0.1177 0.0032 0.2159 0.0034 0.0038 0.0018 0.269 0.1417 0.0039 0.0023 0.1839 0.1482 0.0036 0.0036 0.2826 0.2889 0.0016 0.0046 0.0048 0.0031 0.0033 0.0092 0.1895 0.1045 0.0041 0.0493 0.0033 0.0079 2E-4 0.1947 0.0024 0.2781 0.0033 0.3116 0.0026 0.0052 0.0043 0.0024 0.0041 0.017 0.0052 0.0527 0.0921 0.0032 0.0046 4E-4 0.0072 0.3437 0.1646 0.0385 0.1176 0.1786 0.0036 0.0024 0.0104 0.0236 0.0098 0.0042 0.003 0.0724 0.0054 0.0028 0.361 0.0022 0.0032 0.0035 0.0047 0.0019 0.0811 0.0041 0.1273 0.0138 0.0031 0.0313 0.0067 0.0099 0.0274 0.0387 0.0986 0.0035 0.0065 3E-4 0.0019 0.0896 0.0053 1E-4 0.0573 3E-4 0.0026 0.0079 0.0033 0.011 0.1011 0.1436 0.0051 0.0022 0.0448 0.0043 0.0049 0.0095 0.0046 0.033 0.0984 0.0038 0.0043 0.1932 0.0367 0.0032 0.655 0.0065 0.1491 0.0335 0.0032 0.1668 0.0039 0.1017 0.0047 0.294 0.1794 0.0086 0.1694 0.0058 0.0046 0.1396 0.1941 0.0028 0.2348 0.2055 0.0056 0.0045 0.1322 0.2027 0.0078 0.1115 920 0.004 0.0037 0.2634 0.2082 0.0047 0.1043 0 900 0.004 3E-4 0.0477 0.0068 0.0065 0.3137 880 2E-4 0.0077 0.0036 0.2556 0.0293 860 2E-4 0.0058 0.6574 840 0.0045 2E-4 0.0047 0.6135 820 0 0.0054 0.5457 2E-4 800 0.0051 0.4392 920 3E-4 0.005 0.3414 900 0.0084 0.1111 880 0.0782 860 840 820 800 CO A.6. 3 eaieaonso hssi BOnnwr itrscnann Na containing mixtures nanowire YBCO in phases of amounts Relative 2 Cu 2 O 5 rY1 rYO rBC3e au2e u er CuO er BaCuO2 er BaCO3 er Y2O2 er Y211 er 2 Cl 3 tvrosaonsdetermined amounts various at 182 ◦ ◦ (a) 800 C 2 % Na2CO3 (b) 820 C 2 % Na2CO3 ◦ ◦ (c) 840 C 2 % Na2CO3 (d) 860 C 2 % Na2CO3 ◦ ◦ (e) 880 C 2 % Na2CO3 (f) 900 C 2 % Na2CO3 ◦ (g) 920 C 2 % Na2CO3 FIGURE A.4. Size distribution histograms for nanowires with 2 % Na2CO3. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 183 APPENDIXA.APPENDIX ◦ ◦ (a) 800 C 5 % Na2CO3 (b) 820 C 5 % Na2CO3 ◦ ◦ (c) 840 C 5 % Na2CO3 (d) 860 C 5 % Na2CO3 ◦ ◦ (e) 880 C 5 % Na2CO3 (f) 900 C 5 % Na2CO3 ◦ (g) 920 C 5 % Na2CO3 FIGURE A.5. Size distribution histograms for nanowires with 5 % Na2CO3. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 184 ◦ ◦ (a) 800 C 10 % Na2CO3 (b) 820 C 10 % Na2CO3 ◦ ◦ (c) 860 C 10 % Na2CO3 (d) 880 C 10 % Na2CO3 ◦ (e) 900 C 10 % Na2CO3 FIGURE A.6. Size distribution histograms for nanowires with 10 % Na2CO3. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 185 APPENDIXA.APPENDIX ◦ ◦ (a) 800 C 15 % Na2CO3 (b) 820 C 15 % Na2CO3 ◦ ◦ (c) 860 C 15 % Na2CO3 (d) 880 C 15 % Na2CO3 FIGURE A.7. Size distribution histograms for nanowires with 15 % Na2CO3. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. ◦ (a) 900 C 2 % NaCl (b) 920 ◦C 2 % NaCl FIGURE A.8. Size distribution histograms for nanowires with 2 % NaCl. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 186 (a) 800 ◦C 5 % NaCl (b) 820 ◦C 5 % NaCl ◦ (c) 840 C 5 % NaCl (d) 860 ◦C 5 % NaCl ◦ (e) 880 ◦C 5 % NaCl (f) 900 C 5 % NaCl (g) 920 ◦C 5 % NaCl FIGURE A.9. Size distribution histograms for nanowires with 5 % NaCl. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 187 APPENDIXA.APPENDIX ◦ (a) 800 C 10 % NaCl (b) 820 ◦C 10 % NaCl ◦ (c) 840 C 10 % NaCl (d) 860 ◦C 10 % NaCl ◦ (e) 880 C 10 % NaCl (f) 900 ◦C 10 % NaCl FIGURE A.10. Size distribution histograms for nanowires with 10 % NaCl. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 188 ◦ (a) 820 ◦C 15 % NaCl (b) 840 C 15 % NaCl ◦ (c) 860 ◦C 15 % NaCl (d) 880 C 15 % NaCl FIGURE A.11. Size distribution histograms for nanowires with 15 % NaCl. Blue lines show the log normal fitting curves used to calculate the mean and standard deviation. 189 (> imtro -hpdwt oecvt of unit. cavity the pole order a of in with applicability C-shaped necessary generic or above is diameter the sale system increase for the magnets and of permanent collection possible Existing data as for angle strategies maximised. an maximise is of space to wide sample as the from to during observation access system the which Direct of in observation arrangement direct an for of requiring allow period also prolonged crystallisation, would a environment over sample fields magnetic ideal under An held time. be to specific, gasses a of or atmosphere volumes large non-ambient mixes, solvent/anti-solvent complex require can systems crystallisation high as innovation curb laboratory. can This electromagnets. expensive the sacrifice to normally resort must or one space fields, sample higher the fields, At of essential. magnetic size is of field application magnetic the the involving of experiments magnet homogeneity in permanent reproducibility T ensure to 2.5 order a In of construction and Design B.1 herein. contained text and figures all of preparation 48. reference as T 1 hsbcmsapolmseiclyi antcfil-eitdcytlgot,weethe where growth, crystal field-mediated magnetic in specifically problem a becomes This the involving myself by submitted and written was chapter this create to used paper The a of build and design the describes appendix This antcfilsaeiacsil oms nadiybss vni eleupe research well-equipped a in even basis, daily a on most to inaccessible are fields magnetic ) Sinei at n at r facts” are facts and facts is “Science ∼ oadTup 2016 Trump, Donald – mm 1 T 2 191 r otyclnrclwt small a with cylindrical mostly are hsui a rdcdi re opoiea provide to order in produced was unit This . 2 . T 5 ttcmge n spbihdi part in published is and magnet static A PPENDIX A PPENDIX B ∼ cm 1 B in APPENDIXB.APPENDIXB sample environment with a relatively high, sustained magnetic field, without having to resort to expensive solutions. Currently available permanent magnets above 2 T are Halbach cylinders, arrays designed to maximise magnetic flux within a circular cavity whilst minimising that outside by selective arrangement of the poles of wedge-shaped magnets[240]. This is necessarily restrictive as the sample space is within a tube accessible only from either end and to maximise magnetic field strength using permanent magnets, this space is restricted to small cavities ∼1 cm in diameter [241]. Sample access can be expanded by the introduction of equatorial radial tunnels or by a thin slit in the walls of the cylinder if full equatorial access is required. This of course reduces the field at the centre of the gap and also the radial homogeneity of the field[242]. In theory such designs can produce fields of any strength, however in practice the logarithmic relationship between the outer radius of the array and the resulting field strength mean that fields above 3 T to 4 T become impractical[242,243]. For research purposes, a c-shaped arrangement is preferred, as this allows, 360° access to the sample space. Current examples of such c-shaped magnets have been able to achieve high field strengths of 3.2 T between pole faces of 1 cm2 but with only a small 1.5 mm separation[244]. The design is intended to maximise field strength while increasing the pole gap to 5 mm to provide a larger sample volume. The production of this equipment aims to provide a homogeneous, high field (∼2.5 T) magnetic sample environment with a volume large enough to accommodate solution crystallisation experiments in sample chambers such as NMR tubes and cuvettes whilst simultaneously allowing direct observation of the sample from a wide angle. Whereas the resulting rig is not lightweight at 26.5 kg it is eminently more portable than an equivalent electromagnet system (of the order of 625 kg), provides a max field strength of 2.468 T with relatively low stray field. B.1.1 Component design Practically, the only way in which permanent NIB magnet array would reach magnetic fields >1 T would be by ‘focussing’ the field lines by using angled pole-shoes, as would be found on a standard commercial electromagnet. The greater the magnetic flux at the tip, the larger the gap can be whilst still retaining a relatively high field. Stacks of three NIB block magnets were used as preliminary measurements indicated that adding a forth only has a negligible effect on the strength of the stack end in respect to three, which has a field strength 134 % that of two. Figure B.1 is a 3D representation of a single pole design, intended to be symmetrical across the dotted line. Green sections indicate the ferromagnetic mild steel components designed to carry the magnetic field and to concentrate the magnetic flux between the pole shoes. The red blocks are NIB permanent magnets, arranged to be the source of the magnetic field and the grey pieces are both structural, keeping stacks of three magnets next to each other and some basic shielding in an attempt to stop the field lines ‘short-circuiting’ ultimately reducing the strength of the field. Optimisation of component shape and position was carried out in order to achieve maximum 192 B.1. DESIGN AND CONSTRUCTION OF A 2.5 T PERMANENT MAGNET FIGURE B.1. Graphical representation of the structure of a single pole of the 2.5 T permanent magnet. The actual magnet is symmetrical from 2.5 mm above the pole shoe tip as indicated by the dashed black line. field strength. Figure B.2 displays the magnetic field line shape and magnitude calculated using the software FEMM[245]. Each block has been given the properties associated with the particular material shown in figures B.1 and B.4- B.5 and calculated in an air environment. Grid squares represent 1 cm2 and the white spots labelled A, B, C, D and E represent positions of calculated magnetic field strength of 1.9484 T, 1.5 T, 1.0 T, 0.1 T and 0.01 T respectively. Figures B.4- B.5 give a detailed description of the magnet size and component position. All magnet poles are aligned. The yoke and pole shoes were produced from mild steel instead of pure iron, trading a relatively small increase in field strength for a significant increase in corrosion resistance while reducing material and machining costs of elemental iron. To avoid inhomogeneity caused by air gaps and material interfaces within, where possible, steel components were crafted from a single piece. 6082 aluminium alloy was used for all non-magnetic structural pieces a low magnetic permeability of 1.26 × 10−6 Hm while remaining relatively inexpensive. Both the aluminium shield and frame require high mechanical strength whilst limiting the effects of stray field on its homogeneity within the yoke and pole shoes. An increased homogeneity within the magnet ultimately leads to increased field strength between the pole shoes. 193 APPENDIXB.APPENDIXB FIGURE B.2. Profile view 2D Magnetic field calculation produced using FEMM soft- ware[245]. Points marked A, B, C, D, E correspond to calculated field strengths of 1.9484 T, 1.5 T, 1 T, 0.1 T and 0.01 T respectively. Grid square correspond to 1 cm. The 2D nature of the software does not take into account the flux focussing from the 3rd dimension of the pole-shoe, lowering the calculated field. B.1.2 Component construction The yoke was cut using a water jet from a single piece of mild steel of dimensions 330 mm × 151 mm × 77 mm before being milled to shape, threads were then machined into the sides in preparation for the aluminium magnet frames. Pole shoes were milled from single pieces of mild steel of size 80 mm × 80 mm × 50 mm. For stability, a thread was bored into the base and the whole piece clamped using a 10 mm steel screw. The screw was then cut and machined to form a smooth surface. All aluminium parts were milled from 6082 aluminium alloy sheets of thickness’s 16 mm, 7 mm and 5 mm. Aluminium magnet frames were assembled using brass screws after machining threaded holes into the sides of the thicker, 16 mm plate. The finished fabricated components are shown in figure B.7 (a). 194 B.1. DESIGN AND CONSTRUCTION OF A 2.5 T PERMANENT MAGNET FIGURE B.3. Profile schematic of a 2.5 T permanent magnet with edge lengths tabulated. The dashed red line indicates the plane at which the stray field was measured. FIGURE B.4. Top down schematic of the 2.5 T permanent magnet. The light grey zone represents the yoke and should be considered behind the angled pole-shoes. 195 APPENDIXB.APPENDIXB FIGURE B.5. Schematic of the end of the 2.5 T permanent magnet. B.1.3 Build Instructions B.1.3.1 Safety Before working with any magnets, any electronic devices (phones/laptops/watches) should be removed from the vicinity particularly loose items which could jump to contact and eye-protection should be worn as colliding magnets may shatter. The construction of the magnet rig requires the handling and manipulation of NIB block magnets. These individual, unmodified magnets already possess a strong magnetic field, 588 N of force and able to support (∼0.66 T at the surface). The supplier suggesting a convenient hanger for a sledge hammer (up to 10 kg) as a possible use. The manipulation of multiple block magnets must therefore be undertaken with great care due to the high risk of pinch hazards and flying metal shards as a result of shattering. The strong attraction between block magnets can cause magnets to jump distances of up to 35 cm either shattering on impact or trapping items in between. The block magnets should therefore be stored in the shipping packaging with accompanying shielding until required. During the build, individual magnets or magnets stacks must be stored at least 50 cm apart and with no ferromagnetic or magnetic material nearby. Magnets should be brought into contact in a controlled manner one at a time, approaching from a direction perpendicular to the magnet poles and making use of non-magnetic material (e.g. thick wooden blocks) as guides. More detail of basic magnet handling methods is given below. The steel c-shaped yoke is heavy and should be handled with care to avoid pinch hazards. B.1.3.2 Bringing two magnets into contact To bring two magnets into contact, the first magnet was secured within a wooden template of thickness 20 mm, on a flat surface as shown in figure B.6. A square cavity (∼40 mm × 40 mm) 196 B.1. DESIGN AND CONSTRUCTION OF A 2.5 T PERMANENT MAGNET cut into the template restricts transverse movement of the magnet in place whilst the surface of the template provided a flat surface level with the top of the magnet. The thick edge of a wedge was placed over the magnet and a second magnet was lowered into position until in contact the wedge directly over the first, being careful to make sure the poles were aligned (fig. B.6). Whilst keeping the second magnet above the first the wedge was slowly withdrawn, bringing the magnets together. To produce three-magnet stacks, this process was repeated, bringing the stack of two down onto the wedge. FIGURE B.6. Schematic showing the attachment and removal of a block magnet from a magnet stack. (a) the wooden base template with approx. 40 mm × 40 mm slot to hold a block magnet securely. (b) Attachment of two block magnets using a wooden wedge. (c) Separation of two block magnets with lower template held flush to a wall. When required to separate magnets, the wooden template was held such that a single magnet protrudes above the wooden surface. The template was then be secured in place by being placed flush to a wall. A second wooden panel was then placed flat on the template and used to push against the protruding magnet sliding it off the stack. The second panel was also used to prevent the magnet in the template from jumping out of the cavity. B.1.3.3 Magnet rig assembly Before beginning any assembly of the magnet, a gaussmeter was used to identify and mark the poles of each magnet stack to avoid inserting a stack in an incorrect orientation as this would will greatly decrease the final field strength. Assembly of the magnet rig is broadly outlined pictorially in figure B.7 with a detailed step by step guide provided below. The method for producing the magnet stacks used in this process is describe above. B.1.3.4 Bringing two magnet stacks into contact When strong magnets are brought together side by side, they repel each other, attempting to flip and turn in order to come together end to end with poles aligned. For this reason, when bringing these stacked magnets together, they were restricted in all dimensions other than that 197 APPENDIXB.APPENDIXB FIGURE B.7. Visual documentation of the build process. (a) Shows the fabricated magnet rig components. From left to right: Steel pole shoes × 2, Steel Yoke with Aluminium shielding fitted to upright section, Aluminium magnet frame × 2. (b) Shows the insertion of two magnet stacks into a magnet frame with lowest magnet in the stack resting on the notch cut into the frame. The C-clamp provides extra support to the frame, resisting repulsion between the magnet stacks. Additional shielding from the magnetic field, provided by card or wooden blocks placed between the clamp and the frame, enabled easier removal of the clamps when required. (c) shows two magnet stacks held together by a shielded C-clamp and placed on the foot plate of the yoke piece. (d) the result of sliding the frame in (b) into place around the prepared yoke shown in (c), secured with brass screws. At this point clamps are no longer necessary for support the frame. (e) and (f) show the assembled magnet after stages (a-d) have been repeated for the second magnet frame and the pole shoes mounted. in which they are required to move. To bring two stacks into contact, wooden guides were used as shown in figure B.8. Two stacks were placed on a flat surface 40 cm apart and held between two wooden planks with a width of 6 cm to match the stack height, to prevent lateral motion. Wooden blocks were then placed on the outside edges of the magnets to prevent direct contact between the magnets and the c-clamp which was then used to drive the stacks together. Finally a wooden 198 B.1. DESIGN AND CONSTRUCTION OF A 2.5 T PERMANENT MAGNET lid was placed on top of the wooden planks to prevent any upward motion of the stacks. Pressure was kept on all sides of the housing as the clamp was wound together. Once the stacks are in contact, the wooden housing was removed as the stacks were be held securely by the clamp. FIGURE B.8. Formation of dual magnet stacks using wooden guides and lid. B.1.3.5 Mounting the dual magnet stacks in the aluminium magnet holder The aluminium frame was held securely in a bench mounted vice. The dual magnet stack (held securely between wooden blocks by a c-clamp) was brought to the open end of the frame. The lower edge of the magnets, level with the notch cut into the aluminium frame, to support the base of the stacks. A third wooden block was placed on the top surface of the stack (shown as a dotted line in figure B.9) to prevent vertical movement. A second c-clamp was then fitted perpendicular to the first, making contact (shielded by wooden blocks) with the back plate of the aluminium frame and the wide face of the dual stack. This clamp was then used to slowly slide the magnet stack from between the restraining c-clamp into the aluminium holder. FIGURE B.9. Top view of mounting a dual magnet stack in aluminium magnet holder. Dotted line represents location of wooden top plate. 199 APPENDIXB.APPENDIXB Once the dual stack was within the side-arms of the of the frame, the restraining c-clamp was then be used to apply force to the arms of the frame as show in figure B.7 (c). Through continued application of even pressure the dual stack was moved into position flush with the back plate of the magnet holder. B.1.3.6 Mounting the dual magnet stacks onto the yoke base The remaining two stacks to form the 2 × 2 array at a pole were attached directly to the yoke arm once the aluminium shields have been attached, shown in figure B.7 (c). As with joining individual block magnets together, a wooden wedge was used to lower each stack onto the surface of the yoke arm. Once attached, each stack was moved into position by easily pushing it laterally and when adjacent, the stacks were forced together using another c-clamp. To complete the 2 × 2 array of magnets on each yoke arm, the loaded magnet frame was placed on the same surface as the yoke. Using a clamp, the loaded frame was pushed towards the two magnet stacks attached to the yoke. Once all four stacks were in contact, the aluminium frame was fixed in place using brass screws before the clamps were removed (fig. B.7 (d)). This was repeated on both arms before the pole shoes were attached. B.1.3.7 Mounting pole shoe Pole shoes were placed on a platform raised to the same height as the surface of the 2 × 2 magnet array. Using wood to apply pressure to the pole tip in order to keep the shoe in position, the shoe was then moved laterally onto the array surface. A rubber mallet was used to adjust the fine position of the shoe once attached. B.1.4 Operation Instructions The finished magnet was sited on a stable surface, able to comfortably support a mass of 26.5 kg, 15 cm from any ferromagnetic materials or electronic devices. The sample space can be interrogated, from the remaining angles, for example using a video camera, perpendicular to the field direction. Any sample holder that fits within a 5 mm cavity and is not ferromagnetic can be used with the equipment. Currently 5 mm cuvettes and NMR tubes are used to hold liquid samples between poles. Although the stray field generated drops to background levels within 15 cm of the pole shoes (fig. B.10) electronic devices are kept at a safe distance. The high magnetic field between pole shoes and in immediate vicinity present potential dangers for wearable items (e.g. rings, chains, watches etc.) Due to the shape of the sample space, any interrogable equipment requiring direct observation of the sample, can be used. 200 B.1. DESIGN AND CONSTRUCTION OF A 2.5 T PERMANENT MAGNET B.1.5 Field Characterisation To characterise the successful operation of the magnet, measurements of the field strength between the poles and mapping of the stray field is required. Detection of the stray field is necessary when used in a laboratory setting as applicability of a magnet reduces significantly if it cannot be used safely near other equipment. Magnetic interaction with any surrounding ferromagnetic metals or electronic devices is dangerous thus the stray field must be measured in order to assess a minimum safe distance for such equipment. To measure the stray field, a gaussmeter with transverse Hall probe was used to map the magnetic field strength parallel to the pole direction. Points were recorded at 5 mm increments across the plane to record the field contours which can be seen in figure B.10. FIGURE B.10. Stray field of the 2.5 T magnet measured in a plane that does not cross the yoke represented by a red dashed line in figure B.3. Grey shapes represent the machined pole-shoes seen from above. The red areas at each side indicate a flipping of the field pole. 201 APPENDIXB.APPENDIXB The maximum field strength between the poles (2.468 T) is higher than expected from the field calculations in figure B.2 which predicted a maximum field strength of 1.9484 T. This increased experimental value is expected since the calculations were performed using a 2D slice of the magnet rig, rather than a full 3D model. The stray field measurements correspond well with the calculations of the stray field, dropping to 0.1 T within 5.5 cm from the pole centre and 0.01 T within 11.5 cm. For the calculated field these values are approximately 6 cm and 10.75 cm respectively. Beyond 12 cm the stray field is therefore negligible so maintaining a safe zone of 15 cm around the magnet rig which is free of ferromagnetic materials or electronic devices is recommended. 202