"Applied Science, Faculty of"@en . "Materials Engineering, Department of"@en . "DSpace"@en . "UBCV"@en . "Lodha, Rahul"@en . "2014-04-30T00:00:00Z"@en . "2013"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "Magnesia-chrome refractory ceramics are used in non-ferrous industry because of their corrosion resistance against fayalite-type slags, rich in FeO. Unfortunately, Cr\u00C2\u00B3\u00E2\u0081\u00BA may oxidize to Cr\u00E2\u0081\u00B6\u00E2\u0081\u00BA during the smelting and converting processes, thus making their spent pot-lining hazardous. Two strategies are explored in this present work to develop chrome-free refractories sinterable at lower temperatures. The first is to replace Cr\u00C2\u00B3\u00E2\u0081\u00BA with tetravalent cations (Me\u00E2\u0081\u00B4\u00E2\u0081\u00BA) that form complex spinel with MgO and Al\u00E2\u0082\u0082O\u00E2\u0082\u0083 at 1350-1550 \u00C2\u00BAC. These ions could promote simultaneous synthesis and sintering through formation of non-stoichiometric spinel at high temperatures and re-precipitation of the complex spinel during cooling. The second objective is to study the contribution of nano-size powders of the spinel formers to achieve both synthesis and sintering of Cr-free spinel solid solutions.\nThe current research proved that synthesis and sintering of such spinels, or their solid solutions, could be performed at much lower temperatures (1350-1550 \u00C2\u00BAC) than the Cr\u00C2\u00B3\u00E2\u0081\u00BA-bearing version, which requires firing at 1750-1850 \u00C2\u00BAC to achieve equivalent properties. It is hypothesised that the defect super-structure due to Me\u00E2\u0081\u00B4\u00E2\u0081\u00BA ions inducing the spinel phase formation at relatively low temperatures (1350-1550 \u00C2\u00BAC) results in the simultaneous synthesis and sintering of these complex spinels. The defect super-structures formed at lower temperatures have a higher lattice parameter than the final spinel solid solution phase. The volume expansion due to spinel formation that retards densification is also overcome due to the shrinkage in volume of the complex spinel phase. The results indicate that addition of Fe\u00E2\u0082\u0082O\u00E2\u0082\u0083 in MgO-Al\u00E2\u0082\u0082O\u00E2\u0082\u0083-Me\u00E2\u0081\u00B4\u00E2\u0081\u00BAO\u00E2\u0082\u0082 systems results in improved reaction sintering with an open porosity of ~4% at 1450 \u00C2\u00BAC. However, the narrow particle size distribution for preparation of aggregates makes it necessary to fire the pressed specimens at higher temperature or in two stages in order to achieve similar open porosity to the co-clinkered aggregates for basic refractory ceramics. This research work contributes to the preparation of chrome-free aggregates and binding systems and show that synthesis and sintering of magnesia-spinel phase led to development of chromium-free binding phase for basic castables with comparable flexural and compressive strength to magnesia-chrome bricks with firing temperatures below 1500 \u00C2\u00BAC."@en . "https://circle.library.ubc.ca/rest/handle/2429/45246?expand=metadata"@en . "Synthesis and sintering of chromium-free complex spinels in the MgO-Al2O3-FeOx-Me4+O2 systems by Rahul Lodha A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate and Postdoctoral Studies (Materials Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) October 2013 ? Rahul Lodha, 2013 ii Abstract Magnesia-chrome refractory ceramics are used in non-ferrous industry because of their corrosion resistance against fayalite-type slags, rich in FeO. Unfortunately, Cr3+ may oxidize to Cr6+ during the smelting and converting processes, thus making their spent pot-lining hazardous. Two strategies are explored in this present work to develop chrome-free refractories sinterable at lower temperatures. The first is to replace Cr3+ with tetravalent cations (Me4+) that form complex spinel with MgO and Al2O3 at 1350-1550 ?C. These ions could promote simultaneous synthesis and sintering through formation of non-stoichiometric spinel at high temperatures and re-precipitation of the complex spinel during cooling. The second objective is to study the contribution of nano-size powders of the spinel formers to achieve both synthesis and sintering of Cr-free spinel solid solutions. The current research proved that synthesis and sintering of such spinels, or their solid solutions, could be performed at much lower temperatures (1350-1550 ?C) than the Cr3+-bearing version, which requires firing at 1750-1850 ?C to achieve equivalent properties. It is hypothesised that the defect super-structure due to Me4+ ions inducing the spinel phase formation at relatively low temperatures (1350-1550 ?C) results in the simultaneous synthesis and sintering of these complex spinels. The defect super-structures formed at lower temperatures have a higher lattice parameter than the final spinel solid solution phase. The volume expansion due to spinel formation that retards iii densification is also overcome due to the shrinkage in volume of the complex spinel phase. The results indicate that addition of Fe2O3 in MgO-Al2O3-Me4+O2 systems results in improved reaction sintering with an open porosity of ~4% at 1450 ?C. However, the narrow particle size distribution for preparation of aggregates makes it necessary to fire the pressed specimens at higher temperature or in two stages in order to achieve similar open porosity to the co-clinkered aggregates for basic refractory ceramics. This research work contributes to the preparation of chrome-free aggregates and binding systems and show that synthesis and sintering of magnesia-spinel phase led to development of chromium-free binding phase for basic castables with comparable flexural and compressive strength to magnesia-chrome bricks with firing temperatures below 1500 ?C. iv Preface This research program was a collaborative R & D between Teck Metals Ltd., Trail, B.C., Clayburn Industries, Abborsford, BC and the University of British Columbia, Vancouver BC. Teck has used the bricks developed in ceramic group (UBCeram) of the Materials Engineering Department of the University of British Columbia and provided data on chemical analysis. Clayburn allowed the usage of their facilities for vibration casting, pressing and firing of samples. UBCeram group is focused on determining the degradation/failure mechanisms and investigating the potential of chromium-free basic refractories, by analyzing the structural changes and mechanisms of failure. Given the long-term nature of this research project and the volume of work carried out over the past few years, many people have contributed to the project at various stages including Rahul Lodha, Waiman Lo, Hamidreza Zargar, Carmen Oprea, Dr. George Oprea, Prof. Tom Troczynski, Monte Brothers, Dominic Verhelst, Terry Mills, Ken Hansen. I was responsible for preparation, laboratory testing of prepared samples for physical and mechanical properties, and conducting SEM/EDX and XRD analyses. Regarding the industrial trials, I was responsible for design, preparation and implementation of laboratory scale experiments that were of industrial relevance. Hamidreza Zargar was responsible for the industrial-scale materials evaluation after use by the industrial partners, including contributing to the design of experiments, samples preparation, v performing physical and mechanical tests, and conducting and interpreting SEM/EDX and XRD analyses on the industrial scale samples. Versions of Chapter 5 have been published in the following publications: ? R. Lodha, G. Oprea and T. Troczynski, Spinel Inversion and Lattice Parameters in Chromium-Free Spinel Solid Solutions, Proceedings of the 13th Biennial Conference on Refractories, UNITECR 2013. ? R. Lodha, C. Oprea, T. Troczynski and G. Oprea, Structural studies on magnesia-rich chromium-free complex-spinel solid solutions, International Conference on Ceramics, CONSILOX XI, Bucharest, Romania, 2012. ? R. Lodha, C. Oprea, T. Troczynski and G. Oprea, Sintering studies on magnesia-rich chromium-free spinel-bonded basic refractories, Industrial Ceramics, 31 (3), 2011, pp. 223-228. (Also presented in CIMTEC 2010 and published in Advances in Science and Technology, 70, 2010, pp. 108-113). ? R. Lodha, H. Zargar, C. Oprea, T. Troczynski and G. Oprea, Effect of iron oxide in activated synthesis & sintering of complex spinel, Proceedings of the 12th Biennial Conference on Refractories, UNITECR 2011, Japan, pp. 890-893. ? R. Lodha, G. Oprea and T. Troczynski, Role of Ti4+ and Sn4+ ions in spinel formation and reactive sintering of magnesia-rich ceramics, Ceramics International, 37 (2), 2011, pp. 465?470. vi ? R. Lodha, C. Oprea, T. Troczynski and G. Oprea, Magnesia-rich chromium-free spinel-bonded basic refractories, Advances in Refractories V ? Proceedings of the 49th Conference of Metallurgists, COM 2010, Vancouver BC Canada, pp. 43-50. ? R. Lodha, G. Oprea and T. Troczynski, Reactive-sintering of magnesia-rich basic refractories, Proceedings of the 11th Biennial Conference on Refractories, UNITECR 2009, Brazil, pp. 466-469. ? R. Lodha, T. Troczynski and G. Oprea, Role of oxide additives in the synthesis and sintering of magnesium aluminate spinel, Interceram, 57 (5), 2008, pp. 324-329. (Also presented and published in the Proceedings of 16th International Conference on Refractories, organized by Czech Silicate Society, 2008, pp. 199-205.) ? R. Lodha, T. Troczynski and G. Oprea, Activated synthesis and sintering of magnesium aluminate spinel, Proceedings of IREFCON 08 organized by Indian Refractory Makers Association, 2008, Kolkata, India, pp. 149-153. Versions of Chapter 6 have been published as follows: ? R. Lodha, H. Zargar, T. Troczynski and G. Oprea, Chromium-free complex spinel bonded basic castables, Proceedings of the 13th Biennial Conference on Refractories, UNITECR 2013. ? R. Lodha, H. Zargar, C. Oprea, G. Oprea and T. Troczynski, Chromium-free binding system for basic refractories, Proceedings of the 51st Conference of Metallurgists, COM 2012, Niagara Falls ON Canada, pp. 73-79. vii ? R. Lodha, H. Zargar, T. Troczynski and G. Oprea, Magnesia rich complex spinel bonded basic castables, Proceedings of the 12th Biennial Conference on Refractories, UNITECR 2011, Japan, 153-156. ? R. Lodha, C. Oprea, H. Zargar, G. Oprea and T. Troczynski, Activated synthesis and sintering of complex spinel bonded basic refractories, AIST Steel Properties & Applications Conference Proceedings, Columbus Ohio USA, 2011, pp. 15-22. The manuscripts for all the above were originally drafted by me with subsequent editorial assistance by Dr. George Oprea and Prof. Tom Troczynski. A version of Chapter 7 has been published as follows: ? G. Oprea, H. Zargar, C. Oprea, R. Lodha, T. Troczynski and D. Verhelst, Chromium-Free Spinel Bonded Castables versus Rebonded Fused Grain Basic Bricks, Proceedings of the 13th Biennial Conference on Refractories, UNITECR 2013, Victoria BC Canada. The manuscript was originally drafted by Dr. G. Oprea with input from all co-authors, and extensive modifications performed by me. viii Table of contents Abstract ............................................................................................................................ ii Preface ............................................................................................................................ iv Table of contents ........................................................................................................... viii List of tables ................................................................................................................... xii List of figures ................................................................................................................. xiv List of symbols ............................................................................................................ xviii List of acronyms ............................................................................................................ xix Acknowledgements ........................................................................................................ xx 1. Introduction ............................................................................................................. 1 2. Literature review ..................................................................................................... 5 2.1. Sintering of pure magnesium aluminate spinel ............................................... 6 2.2. Structure of spinel .......................................................................................... 9 2.3. Crystal structure determination through refinement of X-ray diffraction pattern .......................................................................................................... 17 2.4. Effect of additives on the sintering of magnesium aluminate spinel ............. 18 2.5. Sintering and bonding of magnesia in the presence of MgAl2O4 .................. 25 2.6. Magnesium aluminate spinel formation through physico-chemical methods and its effect on sintering ............................................................... 32 2.7. Hydration of magnesia.................................................................................. 34 2.8. Refractory degradation in simulated industrial conditions............................. 36 ix 2.9. Summary ...................................................................................................... 38 3. Scope and objectives ............................................................................................ 41 4. Methodology ......................................................................................................... 44 4.1. Calculation of theoretical inversion in complex spinel .................................. 44 4.2. Complex spinel preparation and characterization ......................................... 48 4.2.1. Category I compositions ................................................................... 49 4.2.2. Category II compositions .................................................................. 50 4.2.3. Category III compositions ................................................................. 52 4.2.4. Category IV compositions ................................................................ 53 4.3. Hydration of the aggregates ......................................................................... 56 4.4. Preparation and characterization of bricks using aggregate and binder systems ........................................................................................................ 58 4.5. Preparation and characterization of vibratory castables using aggregate and binder systems ...................................................................................... 60 4.6. Characterization of the trial refractory ceramics containing chromium-free binder system and comparison with magnesia-chrome bricks ..................... 64 5. Development of dense ceramic structure as binder systems for refractories ........ 67 5.1. Role of tetravalent oxides in forming complex spinel with MgO and Al2O3 ... 67 5.2. Calculations of inversion in spinel solid solutions ......................................... 81 5.3. Effect of Fe2O3 in forming complex spinel in MgO-Al2O3-SnO2-TiO2 mixes .. 84 5.4. Role of nano-powders in complex spinel systems ........................................ 87 5.5. Comparison of chrome-free mixes to chrome-based mixes ......................... 91 x 5.6. Summary ...................................................................................................... 93 6. Development of refractory ceramics using in-house processed binder systems ... 95 6.1. Microstructure of the binder systems ............................................................ 95 6.2. Fabrication of refractory ceramics by pressing with complex spinel binder .. 99 6.3. Hydration of aggregates ............................................................................. 101 6.4. Development of vibratory castables with complex spinel binder systems... 104 6.5. Summary .................................................................................................... 114 7. Characterization of trial ceramics containing the new developed chromium-free binder system ..................................................................................................... 116 7.1. Properties of cast ceramics for the plant trial .............................................. 116 7.2. Visual observation and volume loss for the trial ceramics .......................... 117 7.3. Comparative variation of physical properties of the ceramics ..................... 120 7.4. Comparison of the microstructures of the ceramics ................................... 123 7.5. Mineralogical composition of the ceramics by X-ray diffractometry ............ 131 7.6. Comparative variation of mechanical properties of the ceramics ............... 133 7.7. Summary .................................................................................................... 136 8. Conclusions ........................................................................................................ 138 8.1. Techniques for fabrication of basic chrome-free complex spinel refractory ceramics ..................................................................................................... 140 8.2. In service performance of developed refractory ceramics in the BBOC through comparative analyses after usage ................................................. 141 9. Future work ......................................................................................................... 143 xi References .................................................................................................................. 145 Appendix A Calculation for predicted inversion in spinel solid solution .................... 156 Appendix B Calculations from molar to weight percent ........................................... 158 Appendix C Materials and manufacturers ................................................................ 160 Appendix D Instrumental and measurement precision ............................................ 162 Appendix E Supplementary experimental data ........................................................ 163 xii List of Tables Table 4.1. Weight percent compositions of stoichiometric spinel mixes with TiO2......... 49 Table 4.2. Weight percent compositions of stoichiometric spinel mixes with SnO2 ....... 50 Table 4.3. Weight percent compositions of magnesia-rich spinel mixes with TiO2 ........ 51 Table 4.4. Weight percent compositions of magnesia-rich spinel mixes with SnO2....... 52 Table 4.5. Weight percent compositions of magnesia-rich spinel mixes ....................... 53 Table 4.6. Weight percent substitution by nano-oxide powders in magnesia-rich spinel mixes ...................................................................................................................... 54 Table 4.7. Composition and calculated inversion of magnesia-chrome aggregates ...... 57 Table 5.1. Spinel peak shift with addition of TiO2 and SnO2 in stoichiometric spinel compositions .......................................................................................................... 68 Table 5.2. Spinel peak shift with addition of tetravalent cation oxide additions in magnesia-rich compositions ................................................................................... 68 Table 5.3. Effect of additives on bulk density of stoichiometric compositions ................ 71 Table 5.4. Effect of additives on % true density of stoichiometric compositions ............ 71 Table 5.5. Effect of additives on bulk density of magnesia-rich compositions ............... 72 Table 5.6. Effect of additives on % true density of magnesia-rich compositions ........... 72 Table 5.7. Composition and calculated inversion of magnesia-chrome bricks .............. 81 Table 5.8. Calculated inversion and lattice parameter of the spinel solid solution in the chrome-free mixes ................................................................................................. 83 Table 5.9. Unit cell dimensions of complex spinel for mixes fired at 1450 ?C ................ 85 xiii Table 5.10. Unit cell dimensions of complex spinel for mixes fired at 1450 ?C .............. 90 Table 5.11. Physical properties of mixes containing nano-oxides ................................. 92 Table 6.1. Physical properties of the fired samples ..................................................... 100 Table 6.2. Hydration results of mag-chrome aggregates ............................................. 102 Table 6.3. Hydration results chrome-free aggregates ................................................. 103 Table 6.4. Flow and setting time of castables with different types of magnesia .......... 105 Table 6.5. Flow and setting time of castables with different additives ......................... 106 Table 6.6. Flow and setting time of castables with different dispersants ..................... 107 Table 6.7. Flow and setting time of castables with different dispersants ..................... 108 Table 6.8. Flow and setting time of castables with different dispersants ..................... 109 Table 6.9. Physical properties of castables after drying and firing .............................. 110 Table 6.10. Mechanical properties of castables after drying and firing ........................ 111 Table 7.1. Properties of cast half wedge bricks, fired at 1450 ?C, 5 hours .................. 116 xiv List of Figures Figure 2.1: Phase diagram of MgO-Al2O3........................................................................ 7 Figure 2.2: Structure of spinel ....................................................................................... 10 Figure 2.3: Variation of bulk density for different additives in spinel .............................. 19 Figure 2.4: SEM of magnesia-5 wt. % in-situ MA spinel ................................................ 26 Figure 2.5: SEM of magnesia-20 wt. % in-situ MA spinel .............................................. 27 Figure 2.6: SEM of magnesia-20wt% preformed MA spinel .......................................... 28 Figure 2.7: Flexural strength of magnesia-spinel composites as a function of MA spinel content ................................................................................................................... 30 Figure 4.1. Pressed sample fired at 1450 ?C, 5 h .......................................................... 59 Figure 4.2: Flow table used for castables ...................................................................... 61 Figure 4.3: Cured and dried vibratory castables after firing at 1450 ?C, 5 h .................. 63 Figure 5.1: XRD patterns of stoichiometric spinel samples containing Mg2SnO4 fired at 1450 ?C, 3h ............................................................................................................ 69 Figure 5.2: XRD patterns of magnesia-rich spinel samples fired at 1450 ?C ................. 70 Figure 5.3: The effect of TiO2 additive on the apparent specific gravity of stoichiometric (S) and magnesia rich (M) spinel compositions fired at 1400 and 1450 ?C. .......... 73 Figure 5.4: The effect of SnO2 additive on the apparent specific gravity of stoichiometric (S) and magnesia rich (M) spinel compositions fired at 1400 and 1450 ?C. .......... 74 Figure 5.5: The effect of additives SnO2 and TiO2 on the apparent porosityof magnesia rich spinel compositions fired at 1400 and 1450 ?C. .............................................. 75 xv Figure 5.6: Variation of unit cell dimensions with mole % Mg2SnO4 in spinel solid solution fired for 3 h ................................................................................................ 77 Figure 5.7: Variation of unit cell dimensions with mole % Mg2TiO4 in spinel solid solution ............................................................................................................................... 78 Figure 5.8: Physical properties of pellets vs. mole % Mg2SnO4 in spinel solid solution fired at 1800?C, 3 h ................................................................................................ 79 Figure 5.9: Physical properties of pellets vs. mole % Mg2TiO4 in spinel solid solution fired at 1800?C, 3 h ................................................................................................ 80 Figure 5.10: XRD pattern of magnesia-rich spinel sample containing 4 weight % Fe2O3 fired at 1350 ?C, 3 h ............................................................................................... 85 Figure 5.11: Effect of iron oxide on the densification of magnesia-alumina-titania-tin oxide mixes ............................................................................................................ 86 Figure 5.12: Densification pattern of magnesia-alumina-tin oxide-ferric oxide-nano titania mixes sintered for 3 hrs at different temperatures ....................................... 88 Figure 5.13: Densification pattern of magnesia-alumina-titania-tin oxide-ferric oxide-nano titania mixes sintered for 3 hrs at different temperatures............................... 88 Figure 5.14: Densification of magnesia-alumina-titania-tin oxide-ferric oxide-nano alumina-nano titania mixes sintered for 3 hrs at different temperatures ................. 89 Figure 6.1: Scanning electron micrograph of MgO-Al2O3-TiO2 fired at 1800 ?C, 3 h ..... 96 Figure 6.2: Scanning electron micrograph of MgO-Al2O3-SnO2 fired at 1800 ?C, 3 h .... 97 xvi Figure 6.3: Scanning electron micrograph of MgO-Al2O3-Fe2O3-SnO2-TiO2 fired at 1350 ?C, 3 h .................................................................................................................... 98 Figure 6.4: Particle size distribution of the castable .................................................... 104 Figure 6.5: Scanning electron micrograph (magnification ? 50) of NB-2 fired at 1450 ?C, 5 h with elemental analysis of aggregate surface .................................. 112 Figure 6.6: Scanning electron micrograph (magnification ? 1000) of NB-2 fired at 1450 ?C, 5 h ......................................................................................................... 113 Figure 7.1: Cross-section of UT1 brick, showing slag penetration and distribution of pores and cracks .................................................................................................. 117 Figure 7.2: Cross-section of F15 brick, showing slag penetration and distribution of pores and cracks .................................................................................................. 118 Figure 7.3: Distribution of water absorption and apparent porosity along the bricks perpendicular to the hot face ................................................................................ 121 Figure 7.4: Distribution of bulk density along the bricks perpendicular to hot face ..... 122 Figure 7.5: Composite image of the analysed UT1 and F15 bricks in the slag zone ... 124 Figure 7.6: Distribution of Ag, Sb, Pb and Bi from hot face to cold face ...................... 125 Figure 7.7: Micrographs showing cracks from hot face of both the used bricks .......... 126 Figure 7.8: Microstructure of used UT1 brick from the hot face to 90 mm towards the cold face ............................................................................................................... 128 Figure 7.9: Microstructure of used F15 brick from the hot face to 90 mm towards the cold face ............................................................................................................... 130 xvii Figure 7.10: X-ray diffraction patterns of used UT1 brick at hot and cold faces .......... 132 Figure 7.11: X-ray diffraction patterns of used F15 brick at the hot and cold faces ..... 133 Figure 7.12: Variation of modulus of rupture from hot to cold face .............................. 134 Figure 7.13: Variation of cold crushing strength across the brick ................................ 135 xviii List of Symbols A: Bivalent cation B: Trivalent cation D: Tetravalent cation L: Span between two supports for modulus of rupture tests P: Applied load T: Temperature X: Bivalent anion b: Breadth d: Depth h: Height s.s.: solid solution t: time ?: Diameter 2?: Diffraction angle xix List of Acronyms AP: Apparent porosity ASG: Apparent specific gravity ASTM: American Society for Testing and Materials BBOC: Bottom Blown Oxygen Convertor BD: Bulk density CCS: Cold crushing strength CSD: Cambridge structural database CSSS: Complex spinel MA: Magnesium aluminate Max: Maximum Min: Minimum MOR: Modulus of rupture PDF: Powder diffraction file PVA: Polyvinyl alcohol SEM: Scanning electron microscopy SPMA: Sodium poly-methyl acrylate TD: True density WA: Water absorption WPF: Whole pattern fit XRD: X-ray diffraction xx Acknowledgements I take this opportunity to acknowledge my supervisors, Prof. Tom Troczynski and Dr. George Oprea for their motivation and advice, patient guidance, immeasurable support, and encouragements through the different facets during this work. Also, this research is based on the industrial support of Teck Metals Ltd. and Clayburn Industries Ltd., whom I thank for their collaborative support. Also, the experience and insights of Mr. Dominic Verhelst, Mr. Terry Mills, Mr. Monte Brothers, and their constant support to whom I am grateful. I acknowledge my committee members, Dr. Peter Barr, and Dr. Akram Alfantazi, who graciously agreed to serve on my committee and provide valuable advice. I take this opportunity to thank profusely Mrs. Carmen Oprea, Dr. Guotian Ye, Mr. Wai Man Lo, Mr. Hamidreza Zargar, and all members of UBCeram Group for their motivation and support during this project. In particular I acknowledge Mr.Hamidreza Zargar for providing me with the selected results of the industrial experiments. In this regard, I am very grateful to Mr. Dominic Verhelst of Teck Metals Ltd., who went ahead with the installation of cast specimens for verification and feedback regarding the plant trials and service life. I acknowledge Mrs. Mary Fletcher, Mr. Gary Lockhart, Mr. Glenn Smith, and all members of my department for their encouragement and assistance. I am xxi also indebted to Prof. Mati Raudsepp and Dr. Steven S. Hepperle for discussions and demonstrations on different phase and microstructural analyses tools. I am also thankful to Abhijit Pandhari, Aniruddha Chatterjee, Amir Hossein Nobari, Tanay Bose, Tirdad Nickchi, Edwin Gershom, Stefan Honisch, Sudip Talukdar, Harish Gautam, and all my friends who have supported and motivated me throughout. I am also grateful to the University of British Columbia Four Year Fellowship (4YF) and the UBCeram consortium on refractories for their support. I am thankful also to Refratechnik Steel GmbH, BASF ? The Chemical Company, Alcoa Inc., Almatis, Washington Mills, Evonik Degussa Corporation, and the various companies, who provided support, facilities and materials for this research. I take this opportunity to express my gratitude to my parents, my wife and our families, for their constant motivation, support and advice. 1 1. Introduction Refractories containing chrome ore (mainly FeCr2O4 with Al2O3 and other oxide impurities) and magnesia form complex spinel during heat treatment as a solid solution of 6 spinel phases: MgO?Cr2O3, MgO?Al2O3, MgO?Fe2O3, FeO?Cr2O3, FeO?Al2O3, FeO?Fe2O3. Such formulated refractory materials resist hydration (e.g. due to water vapour present in air and combustion gases) and corrosion of periclase, a cubic phase of magnesium oxide that is the major component of basic refractories. Unfortunately, one of the various oxidation states that chromium ion can have in cyclic oxidizing/reducing atmosphere is an environmental concern. During oxidation it may form Cr6+ in the presence of alkali and alkaline-earth metal oxides at temperatures greater than 1500?C. Hexavalent chromium can form carcinogenic water-soluble compounds especially leading to contamination of fresh water resources. This has led to development of chrome-free refractories for steel, cement and glass furnaces, coal gasifiers etc. Further, the fabrication of magnesia-chrome ore based refractory ceramics requires heat treatment at temperatures in excess of 1730?C, which results in high energy consumption and greenhouse gas emissions during the manufacturing process. Magnesia-chrome bricks are used for non-ferrous smelters due to their corrosion resistance against fayalite slags rich in iron oxide that results in better performance as compared to the available chrome ore-free magnesia-based bricks. The service-life in the smelters compensates for the greenhouse gases consumed during manufacture as 2 lesser repairs and shutdown of the hot converter leads to lower energy consumption. Also, in the non-ferrous metal production processes, there has been no detectable amount of Cr6+ as the temperature is below 1350?C and the slags contain small amounts of alkali and alkaline-earth metal oxides. Although magnesia-chrome ore based refractories are still used currently, there may be changes in the furnace conditions or environmental concerns at a later date, which demands research for suitable alternatives through fundamental studies on the system. Aluminium oxide, similar to chromium oxide, is a refractory oxide that forms magnesium aluminate (MA) spinel with magnesium oxide and has been successfully used in the steel and cement industry, where greater than 66 wt. % Al2O3 is required. Therefore, the current research explores the possibilities of processing chrome-free spinels with 60 wt. % magnesia. The alumina-magnesia system can be processed and fired at lower temperatures as compared to the chrome-bearing systems, thus decreasing the greenhouse gas emission during the manufacturing process. This project reviewed various oxides in their interaction and bonding with magnesia grains. The review led to the understanding that magnesia-alumina system can be successfully used for preparing alumina-rich refractories, based on the dissolution of alumina in MA spinel. The bonding process is similar to chrome ore containing refractories, and creates bonds with magnesium oxide by direct reaction and mutual dissolution of the solid phases, namely periclase and spinel. In order to obtain good bonding of the periclase grains and strength of the fired magnesia-complex spinel 3 based refractories using chrome ore, the mix of oxides has to be fired at high temperature for densification during sintering. During cooling, the non-stoichiometric spinel precipitates from the periclase grains and bonds to the adjacent grains resulting in a reaction bonding along with precipitation around the grains. In order to achieve better bonding in chrome free spinel system, marked difference in solubility of different spinel compounds in magnesia with temperature would result in precipitation and formation of bonds with the other grains during cooling. During initial investigation, tetravalent cation oxides like tin dioxide (SnO2) and titanium dioxide (TiO2) as additives in the sintering of magnesia-alumina based ceramics have shown enhanced solubility of magnesia in spinel solid solutions as compared to pure MA spinel. Theoretical analysis of the position of the cations in spinel structure indicated that tetravalent ions create inversion in position of the bivalent ions in spinel structure. This positional inversion is similar to the behaviour exhibited mainly by Fe3+ and to a smaller extent by Cr3+ in the complex magnesia-chrome spinel phase. This led to the hypothesis that inducing inversion in spinel will increase the solubility of spinel in magnesia. In the current work, the theoretical inversion in the complex spinel solid solution is calculated based on a modular approach for inversion in the pure composition of the individual spinel phases (MgO?Cr2O3, MgO?Al2O3, MgO?Fe2O3, 2MgO?SnO2, 2MgO?TiO2) hypothesising that all the ions are in their most stable state. The values for inversion in the complex spinel solid solution are used to calculate the lattice parameter and compared to the experimental values. 4 This work further explored size grading of the initial materials and change in the process parameters with the objective that reaction bonding on the surface of the periclase grains may be created at lower firing temperatures, which will reduce the greenhouse gas emissions. Solid oxides that react with magnesia at temperatures below 1400 ?C forming spinel structures can bond the periclase grains through dissolution during heating and re-precipitation during cooling in the firing process. This could lead to development of chrome-free refractories fired at temperatures lower than 1500 ?C and their disposal would be eco-friendly as compared to magnesia-chrome refractories. The current research contributes to the preparation of such chrome-free aggregates and novel binding system for making both refractory bricks and castables with firing temperatures below 1550 ?C. The results show that simultaneous synthesis and sintering of the magnesia-complex spinel phase could indeed lead to development of chromium-free binding phase for basic castables, with properties comparable to magnesia-chrome bricks. 5 2. Literature review The magnesia-chrome bricks used in non-ferrous converters are manufactured through fusion of magnesia and chrome ore and its subsequent grinding to form aggregates (180 ? 4000 ?m) having periclase (MgO) and complex spinel solid solution (Mg,Fe2+)(Al,Fe3+,Cr3+)2O4 as the primary phases. These aggregates are subsequently mixed with finer particles of magnesia and chrome ore and pressed to shape in order to achieve one-step firing process at temperatures above 1700 ?C. There are two major factors that have led to further research in the refractory lining of non-ferrous converters: (1) the greenhouse gases produced during the high temperature synthesis of the refractory materials and (2) the disposal of spent linings that have oxides of chromium as they may contain water-soluble Cr6+ based carcinogens. In order to understand the processes involved in the manufacture of chrome-free basic refractories and their ability to provide service life in pyro-metallurgical applications, studies of the different aspects involved in manufacture and service have been identified as: 1. Sintering of pure magnesium aluminate spinel 2. Structure of different types of spinel 3. Crystal structure determination through refinement of X-ray diffraction pattern 4. Effect of additives on the sintering of MA spinel 6 5. Sintering and bonding of magnesia in the presence of MgAl2O4 6. MA spinel formation through physico-chemical methods and its effect in sintering 7. Hydration of magnesia 8. Refractory degradation in simulated industrial conditions 2.1. Sintering of pure magnesium aluminate spinel The synthesis of stoichiometric MA spinel (MgAl2O4, Error! Reference source not found.) from the oxides is solid-state diffusion controlled [1] and accompanied by 7.8 % volume expansion. This can be calculated on the basis of specific gravities of corundum, periclase and MA spinel. The specific gravity of magnesia is 3.58 g.cm-3 [2], that of alumina is 3.98 g.cm-3 [3] and that of MA spinel is 3.58 g.cm-3 [4]. The molecular weights of magnesia, alumina and MA spinel are 40.31, 101.96 and 142.27 g/mole respectively. The initial volume of the materials considering the formation of one gram mole of MA spinel is 11.17 cm3 MgO and 25.63 cm3 of Al2O3, a total volume of 36.80 cm3, which results in a final product volume of 39.66 cm3. Hence the volume expansion in creating spinel is 7.8 % from the initial volume of the compounds. The result of (a) volume expansion during spinel structure formation from constituent oxides and that (b) the rate controlling step is through diffusion results in low rate of spinel structure formation and the associated volume expansion does not result in 7 closure of pores. The use of aluminium and magnesium hydroxides to form MA spinel [5] by calcination at 1400 ?C for 1 h does not completely form the spinel structure. Figure 2.1: Phase diagram of MgO-Al2O3 [Adapted from B. Hallstedt, Thermodynamic Assessment of the System MgO?Al2O3, Journal of the American Ceramic Society, 75 (6), 1992, pp. 1497-1507, with permission from John Wiley and Sons, Reference 6] From MgO-Al2O3 phase diagram shown in Figure 2.1, it is seen that alumina solubility in MA spinel is almost negligible at 1100 ?C (1373 K) and increases at 1800 ?C (2073 K) to form spinel solid solution where 1.5 moles of Al2O3 is soluble in one mole of 8 MgAl2O4 spinel, whereas only 0.16 moles of MgO is soluble in one mole of MgAl2O4 spinel. The cubic structures at 1800 ?C indicate that the solid solution in the MgO-Al2O3 system is a non-stoichiometric spinel. The formation of solid solution during firing and precipitation bonding during cooling is more pronounced for alumina based refractories. Further, the phase diagram indicates that the formations of compounds are with fusion in excess of 1700 ?C (1973 K), indicating reaction sintering temperature higher than 1750 ?C. The usage temperature (1250 ?C) of the refractories is less than 0.7 times the fusion temperature (2100 ?C), hence the mechanical properties of these refractory compounds are not affected significantly by the temperature of use. The available data from the literature [7, 8] show that flexural strength at 1300 ?C (~160 MPa) is greater than 75 % of values at room temperature (~175 MPa) for stoichiometric and alumina-rich MA spinel based refractories. Sarkar et al [5] reported that after calcination at 1400 ?C, the precursor hydroxides did not completely form MA spinel. These were then mixed with additives (co-precipitated MA spinel, TiO2, Cr2O3, B2O3, V2O5). They were shaped by uniaxial pressing at 100 MPa with organic binder (poly-vinyl alcohol) dissolved in water, which burnt out during the initial stage of firing, and the role of additives in reducing the MA 9 spinel formation temperature was studied as a function of additive concentration. However, there has been no data on the role of multiple additives in formation of spinel structure, and spinel solid solutions with more than four oxides have not been reported. The phase diagrams in the ternary systems including iron oxide, tin oxide and titanium oxide are based on non-stoichiometric oxides, and have not been included because the heat treatment in air and after spinel solid solution formation are not reported in the literature and the effect of synthesis, sintering and further reaction in presence of any spinel phase has not been explored. 2.2. Structure of spinel Spinel structure, Figure 2.2 [9], consists of a cubic close packed (CCP) lattice of thirty-two oxygen ions with one eighth of the sixty-four tetrahedral voids and half of the thirty two octahedral voids are occupied by cations, resulting in a total of twenty four cations in the spatial arrangement of eight formula units. For this kind of structure, the representation of the lattice is done by making the bivalent cation as A, trivalent cation as B and the bivalent anion as X, having the structural formula as A8B16X32 (Figure 2.2), or as a formula unit denoted by AB2X4. 10 Figure 2.2: Structure of spinel [Reprinted from: A. Kalendov?, D. Vesel? and J. Brodinov?, Anticorrosive spinel-type pigments of the mixed metal oxides compared to metal polyphosphates, Anti-Corrosion Methods and Materials, 51 (1), 2004, pp.6 ? 17, with permission from Emerald Group Publishing Limited 10] Since 8 tetrahedral sites and 16 octahedral sites are filled up by cations, there are two circumstances regarding the position of cations. In the case of all 8 tetrahedral sites being filled up by bivalent cations, then all sixteen octahedral sites are occupied by trivalent cations making the structure a normal spinel. When all 8 tetrahedral sites are occupied by trivalent cations and sixteen octahedral sites equally shared by the bivalent and trivalent cations, the structure is called inverse spinel because of an inversion in 11 position of all the bivalent cations, which occupy octahedral sites rather than all eight tetrahedral sites. However, due to the presence of vacancies and movement of cations in the lattice, it has been observed that perfect normal or inverse spinels generally do not exist [11]. The modification in spinel structure by substitution has an important role to play in the material properties. For example, when a different bivalent cation substitutes a host bivalent cation, that is, without change in spinel structure or the internal spatial positions, there is very little modification in the physical properties, as for example between MgAl2O4 and a solid solution of MgAl2O4 and ZnAl2O4 [12]. However, it is seen that increase in atomic number for any substitution wherein the lattice parameters do not vary much has an effect of increasing the melting point as is the case with trivalent substitution of aluminium by chromium. A 10 mole % substitution of Al3+ by Cr3+ increases the fusion temperature from 2135 ?C to greater than 2200 ?C [13]. Substitutions by other cations can change the properties altogether as they result in an imbalance of either charge or cations. The effect may be on the thermal conductivity with the creation of defect structures generated through the balance of the charge. When such ions are present in small quantities, they may play a major role in the initial phase of material preparation, thus acting as sintering aids when they enter the structure. The substitutions in the structure are written in chemical form using the Kr?ger-Vink notation system. Through this system, the position of substitution cation and its effect on the crystal can be expressed. 12 Kr?ger-Vink notation is used to denote defect species in crystals [14]. The substitutions in the spinel structure are depicted through chemical representation as proposed by Kr?ger and Vink using a set of conventions used to describe electrical charge and lattice position of the ions in the lattice. For example, when titanium replaces aluminium in the MA spinel, the same can be written as AlAlx + Ti?4+? Al?3+ + TiAl? Reaction 2.2.1 This charge can be balanced by a magnesium ion replacing the aluminium ion. AlAlx + Mg?2+? Al?3+ + MgAl/ Reaction 2.2.2 Here the subscripts denote the ion being replaced and ??? indicates the free ion. Similarly, the superscripts are used to denote the charge of the free ion and charge imbalance, if any, in the case of one ion being replaced by another. The nomenclature for denoting the electron deficiency in the structure is through a superscript dot ???, the superscript cross ?x? denotes a neutral position and the superscript slash ?/? denotes a positive charge. Two of these electron deficient sites can be balanced by the addition of an extra oxygen ion or removal of a magnesium ion from the structure to be in a chemically balanced state. 13 O?2- ? Oi// where ?i? denotes an interstitial position Reaction 2.2.3 However, the possibility of a large oxygen ion with ionic radius 1.4 ? to go to the interstitial position a cubic cell of oxygen needs more space or distortion of the lattice compared to creating a vacancy in the cation site. Magnesium ion may be the first to be displaced in this mechanism because of its higher ionic radius (0.72 ?) [15] compared to aluminium (0.54 ?) [16]. Since titanium (0.61 ?) [17] and chromium (0.52 ?) [18] ions have more electrons in the outermost shell and can form coordinate as well as electronic bonds with the oxygen ions, they are more preferred to the octahedral positions as compared to both magnesium and aluminium ions. However, excess aluminium ion is less likely to be displaced from its position when a titanium ion takes its place in a MA spinel because of the ratio of its charge to ionic radius, and the magnesium then displaced to maintain charge balance can react further with two aluminium ions to form the spinel structure. The creation of localised defects near the titanium ions can enhance the sintering and lower down the melting point of this MA spinel with titanium ion replacements because of the substitution defects thus formed in the structure. MgMgx ? Mg?2+ + VMg?? Reaction 2.2.4 14 Substitution of bivalent and trivalent ions in their own site does not create any charge imbalance. However, when chromium ion substitutes the aluminium ion, the chromium ion with partially filled ?d?-orbital can share the electrons in the outermost shell of oxygen and at the same time create ionic bonds. The resulting spinel structure has better properties compared to MA in terms of the melting point, mechanical properties and chemical inertness. Iron can be present both in bivalent and trivalent state and thus substitute the magnesium and aluminium ions in the spinel lattice. However, the tendency of iron to change its valence state and react with the molten mass and carbon monoxide generated during pyro-metallurgical reduction [19] makes it unsuitable as a large-scale component of refractory systems. MgMgx + Fe?2+ ? Mg?2+ + FeMgx Reaction 2.2.5 AlAlx + Fe?3+ ? Al?3+ + FeAlx Reaction 2.2.6 15 2.2.1 Preferred position of Mg2+ ions The Mg2+ ions result in inert configuration, similar to the electron cloud of Ne. The position of the Mg2+ ions is thus dependent on the site preferences of other cations present in the spinel phase. 2.2.2 Preferred position of Fe2+ ions The bands caused by spin-allowed transitions of Fe2+ in fourfold coordination display distinct polarization dependencies [20]. The group theory approach to the polarization properties is more transparent in the case of the octahedral coordination compared to the tetrahedral one. In the case of octahedral coordination, the distorted sites are modelled to lower symmetry states and observed that all transitions are forbidden in a regular octahedron. In the case of tetrahedral coordination, the opposite situation exists. In a regular tetrahedron, all transitions are symmetry-allowed, except for the square-planar coordination. This is an extreme example of such distortion leading to a number of new induced symmetry elements, including a center of symmetry, that cause all transitions of IVFe2+ to be forbidden. In the absence of ?d?-orbital effects from other ions, Fe2+ ions form a normal spinel structure. 16 2.2.3 Preferred position of Al3+ and Ti4+ ions In the case of both Al3+ and Ti4+, the inert configuration is obtained. The charge factor [21] determines the position of the ions and both Al3+ and Ti4+ occupy the octahedral sites in case any bivalent cation is present. 2.2.4 Preferred position of Cr3+ ions Cr3+ is fully ordered at octahedral sites. This arrangement results in a three-dimensional network of corner-sharing tetrahedral [22] in which Cr have a six-fold coordination by oxygen. The dominant magnetic exchange interactions are antiferromagnetic due to the direct overlap of the t2g orbitals (dxy, dyz, dzx) of the nearest neighbouring Cr3+ (3d3) ions. 2.2.5 Preferred position of Fe3+ and Sn4+ ions Fe3+ ions have 3d5 outer shell electrons, or a half-filled d orbital that results in high spin [23] along the degenerate planes. Similarly, Sn4+ with completely filled 4d10 would occupy the tetrahedral site because of the degenerate or non-symmetric planes. These result in both Fe3+ and Sn4+ ions to occupy the tetrahedral site in the spinel 17 structure, and resulting in displacement of bivalent cations from the tetrahedral site as in a normal spinel structure, that results in the formation of inverse spinel structure. In the case of having other cations that could share the ?d?-electron through exchange mechanisms [20], the amount of inversion may vary. 2.3. Crystal structure determination through refinement of X-ray diffraction pattern X-ray diffraction (XRD) technique is primarily used to determine the phases in any material based on Bragg?s Law [24]. The powder diffraction technique has led to the development of database of identified materials, such as the International Centre for Diffraction Data's Powder Diffraction File [25] (PDF) or the Cambridge Structural Database [26] (CSD). The database provides information about known materials, which can be compared to newer structures using a full pattern analysis technique called the Rietveld [27] method. The crystal structure is used to generate a theoretical diffraction pattern that can be compared to the observed data. A procedure to obtain the least squares minimizes the difference between the calculated pattern and each point of the observed pattern by adjusting parameters like the unit cell dimensions and angles. There are available softwares that can be used to do the Rietveld method for analysis of changes in crystal structure due to external effects like temperature, pressure, 18 fabrication method or intrinsic effects like changes in the composition from pre-identified material in the PDF or CSD. JADE software has been reported for identifying crystal structure of metastable compounds in the MA spinel-alumina system [28], or even to analyse perovskite and pyrochlore modifications of Pb2MnReO6 [29]. This shows the versatility of the software to analyse multiphase systems and determine the crystal structure. 2.4. Effect of additives on the sintering of magnesium aluminate spinel Different additives, which would reduce the MA spinel synthesis temperature, are based on their ability to form intermediate compounds at low temperature that would result in the synthesis of spinel structures at low temperature (T < 1500 ?C). The addition of ZnO [5] improves densification, but its role in decreasing the synthesis temperature or increasing the solubility of the oxides in the MA spinel is not verified. The role of Y2O3 [30] in densification and flexural strength improvement for alumina-rich and stoichiometric MA spinel but did not affect magnesia-rich compositions, indicating that it might have increased the solubility of alumina in MA spinel. In the case of both Y2O3 and ZnO, there is no change in valence with temperature that would result in intermediate compounds as in the case of oxides of iron. TiO2 addition [7] below 2 wt. % to magnesia and alumina fired at 1550 ?C ? 1650 ?C resulted in the MA spinel phase 19 without any effect of TiO2 additions as its presence was limited to the grain boundary regions with impurities like calcium oxide and silica that form low melting phases. Figure 2.3: Variation of bulk density for different additives in MA spinel [Reprinted from R. Sarkar, S. K. Das and G. Banerjee, Effect of additives on the densification of reaction sintered and presynthesised spinels, Ceramics International, 29 (1), 2003, pp. 55-59, with permission from Elsevier, Reference 5] The addition of co-precipitated MA spinel [5] was reported to have no effect on densification compared to the sintering without additive, as the calcined co-precipitate did not induce vacancy in spinel lattice structure that increases the reactivity. V2O5 and B2O3 had a detrimental effect on the density as they might have enhanced the grain growth and coarsening before densification and formed low melting phases with the 20 MgO?Al2O3 system. Cr2O3 increased the density as it can substitute Al3+ in the spinel lattice and make a solid solution with MA spinel. The authors have reasoned this to increasing diffusivity and mass transfer caused by the substitution. The increase in the densification was highest for TiO2 additive (Figure 2.3) reporting that dissolution of TiO2 and ex-solution of Al2O3 in spinel structure enhanced the mass transport for densification by creating cation vacancy that enhanced cation movement. Sarkar et al [7] used magnesia and alumina for preparing MA spinel and TiO2 as additive (0-2 wt. % in the magnesia-alumina mixture) and mixed them in a ball mill for 15 minutes before attrition milling it for 4 h. The milled powders were isostatically pressed at 175 MPa to form pellets (? 25 mm, h 10 mm) and bars (60 mm ? 6 mm ? 6 mm), then dried at 110 ?C and fired at 1550 ?C, 1600 ?C and 1650 ?C with 2 hours soaking at maximum temperature. Phase analysis of all the different batches sintered at different temperatures determined only the MA spinel phase, and titania showed no effect on the X-ray diffraction pattern. They studied the densification pattern and reported that addition of TiO2 had a beneficial effect on the density and mechanical properties for sintering at 1550 ?C for 0.5 wt. % addition in the mix composition. The use of titanium dioxide as additive in MA spinel shows only a marginal improvement in the cold modulus of rupture (from 175 MPa to 190 MPa), but the values at 1300 ?C decrease for additions above 0.5 wt. %, to 160 MPa. They further reported that the EDS analysis showed that titanium dioxide was present in the grain boundary 21 regions with impurities like calcium oxide and silica that form low melting phases. This shows that liquid phase sintering is an important parameter in the densification when TiO2 is used as an additive with industrial grade raw materials. Sarkar et al [30] studied the role of yttrium oxide in densification of the stoichiometric and magnesia rich (MgO:Al2O3 ratio of 2:1) MA spinel and reported that it promotes densification of the MA spinel as well as improves the high temperature flexural strength. The stoichiometric MA spinel showed poor densification at 1550 ?C (bulk density of 2.2 g.cm-3), and Y2O3 addition of 4 wt. % improved densification (bulk density increased to 2.8 g.cm-3). Magnesia rich MA spinel showed higher densification, and Y2O3 addition resulted in a compact structure with bulk density 3.4 g.cm-3 and 95% relative density at 1550 ?C. The modulus of rupture at 1300 ?C, for samples fired at 1600 ?C, showed low value, which improves significantly with the addition of Y2O3. In the case of hot strength of magnesia rich MA spinel is much higher (85 MPa) than for the stoichiometric MA spinel (38 MPa), which the authors attribute to angular grain shape and presence of free periclase as second phase, as observed in phase analysis and microstructure. For the magnesia rich batch, the addition of Y2O3 increases the hot strength (105 MPa) but the improvement is not as significant (80 MPa) as that of the stoichiometric batch. 22 Ghosh et al [8] studied the effect of ZnO additives and reported that it favours densification but there is a decrease in the strength from 175 MPa at room temperature to 100 MPa at 1300 ?C. The authors also reported that 1.25 wt. % ZnO goes into the solid solution at 1600 ?C, and the best results for modulus of rupture are obtained for 1 wt. % ZnO additive. The use of B2O3 as additive for MA spinel formation can lower the sintering temperature. Bhattacharya et al [31] varied the B2O3 content from 0.5 to 10 wt. % and used calcined magnesia and alumina powders for the fabrication of MA spinel. After mixing and pressing, the samples were fired at 900 ?C and 1100 ?C for 3 h. The use of boron oxide causes the formation of MA spinel at 900 ?C, and the maximum spinel structure formation at this temperature is observed for 1.5 wt. % additive. Further increasing the temperature to 1100 ?C increases the amount of spinel structure formed, and decreases the amount of the borate. This can be attributed to glass formation as the borate may form a glass with alumina and impurities. The mechanism of spinel structure formation is by diffusion through a liquid phase wherein the liquid phase dissolves Mg2+ and increases the mobility thus allowing it to then further react with alumina to form MA spinel at a lower temperature as the alumina reacts preferentially with magnesia. At 1100 ?C, the alumina is also dissolved in the borate but it then reacts with magnesia and the MA spinel crystallizes out of the liquid phase. Higher than 1.5 wt. % boron oxide causes borate formation and this hinders the formation of spinel 23 structure. The formation of glassy phase and its role in the properties of the spinel phase thus formed is an important parameter in the recrystallized MA spinel formed. Quenard et al [32, 33] prepared hot pressed samples using the urea-nitrate combustion method to prepare nano-powders of a mix of spinel and zirconia. However, zirconia was reported to be present as a discrete secondary phase unlike titania. The secondary phase improved the mechanical properties of the ceramic, through micro-cracking and stress induced transformation in zirconia. It is also seen that the modulus of rupture at 1300 ?C increased from 85 MPa to 110 MPa with zirconia content increased to 20 wt. %. However, ZrO2 [34] did not react with alumina, magnesia and MA spinel, remaining as a secondary phase. The crystal structure of a normal spinel compound like MgAl2O4 consists of bivalent cations in tetrahedral site and trivalent cations in octahedral sites. Inverse spinel structure as in MgFe2O4 consists of all the bivalent cations in octahedral site. In the case of solid solution of different spinel compounds, the presence of both normal spinel and inverse spinel will result in bivalent cations distributed in both tetrahedral and octahedral sites as in the case of the spinel solid solution formed by the reaction of magnesia with chrome ore, which contains oxides of iron, aluminium and chromium. A study of the complex spinel phase, (Mg,Fe2+)(Cr,Fe3+,Al)2O4, in the magnesia-chrome bricks that is currently used in non-ferrous refractory linings on the basis of 24 structure [21] shows that the preferential position of different cations have a role in the high temperature stability of spinel solid solution and this may be the one of the reasons for lower reactivity of the fayalite type slag rich in oxides of iron. Theoretical calculations [21, 35] on the high temperature order-disorder in the position of the cations in the spinel structure have been studied in order to understand solid solutions of different spinel compounds. The experimental findings have shown that at high temperature, there is an equilibrium distribution of the occupancy of tetrahedral and octahedral sites by bivalent cations where the spinel phase is most stable and the reported value of ~33% inversion [35] results in a structure of high thermodynamic stability even at elevated temperature. At high temperature, the thermal excitation of the electrons and the expansion of the bonds result in cations not occupying their lowest energy position. This results in the material being more reactive at high temperature. Complex spinel phase can be less reactive as the cations are in thermodynamic stable state at high temperature, thus resulting in such phases to be inert as compared to both normal and inverse spinel. With respect to the use of tetravalent oxides [36], the aim is to obtain a combined effect with a bivalent cation to maintain the charge balance similar to that of two aluminium ions. Tetravalent cations may also form another cubic structure - perovskite (ABO3) [37], which has cations in both the tetrahedral and octahedral sites, similar to that of spinel structure [9], although the number of tetrahedral and octahedral site occupancies and their arrangements are not similar. The formation of perovskite structure is based mainly 25 on the ratio of bivalent to tetravalent cations and hence in a magnesia-rich system this phase may not be formed. The cations have preferential site occupancy at room temperature, which may be attributed to their electronic configuration and ionic size. On the basis of previous researches, it can be summarized that the additives to MA spinel result in the following phenomena: 1. Titanium dioxide is reported to be the most effective for densification of MA spinel through solid state route, but the modulus of rupture at 1300 ?C decreases from 180 MPa with increasing titania content to 160 MPa for 2 wt.% TiO2 addition. 2. Boron oxide decreases the temperature of MA spinel formation from 1500 ?C to 900 ?C. 3. Use of zirconia increases the strength of the MA spinel, but zirconia remains as a secondary phase that does not enhance in spinel formation. 2.5. Sintering and bonding of magnesia in the presence of MgAl2O4 In situ formed MA spinel-based magnesia refractories [38] can be processed with the addition of an alumina that reacts with the magnesia matrix before or during sintering, where a strong bond between MA spinel grains and the magnesia matrix is formed (Figure 2.4). The initial MA spinel formation has been reported to occur around 26 the periphery of the alumina particles and then extends towards the particle centre. This results in a better bond of the peripheral spinel and has hollow core. Figure 2.4: SEM of magnesia-5 wt. % in-situ MA spinel [Reprinted from C. Aksel, P. D. Warren and F. L. Riley, Magnesia-spinel microcomposites, Journal of the European Ceramic Society, 24 (10-11), 2004, pp. 3119-3128, with permission from Elsevier, Reference 38] However, the strength and high temperature fracture characteristics are dependent on the level of impurities in magnesia, and also on the distribution of secondary phases. If the alumina used is granular, the MA spinel formation may take more time and expand during service, but the use of smaller alumina particle size will increase the rate of spinel structure formation in the diffusion controlled reaction, thus improving the densification as well as occupy the grain boundary regions. 27 In Figure 2.4, the 5 wt% of MA spinel, the bright phase is at the grain boundary region. Here, the alumina used was A16SG of Alcoa having a purity of 99.8 % and average particle size of 0.5 ?m and magnesia was calcined to 1300?C to get average particle size of 0.5 ?m. However, when the alumina content was increased to 20 wt. %, the spinel structure formation was not restricted to the grain boundary only, but this structure also appeared as intra and inter-granular phases in the magnesia matrix [38]. Figure 2.5: SEM of magnesia-20 wt. % in-situ MA spinel [Reprinted from C. Aksel, P. D. Warren and F. L. Riley, Magnesia-spinel microcomposites, Journal of the European Ceramic Society, 24 (10-11), 2004, pp. 3119-3128, with permission from Elsevier, Reference 38] The flexural strength at ambient temperature decreases from 225 to 175 MPa with increasing alumina/ MA spinel content to 30 wt. % and this may be caused because of 28 internal cracks and lower strength of the MA spinel particles (fracture surface energy of ~4-7 J.m-2) as compared to the magnesia particles (fracture surface energy of ~14-15 J.m-2). Figure 2.6: SEM of magnesia-20wt% preformed MA spinel [C. Aksel, P. D. Warren and F. L. Riley, Magnesia-spinel microcomposites, Journal of the European Ceramic Society, 24 (10-11), 2004, pp. 3119-3128, with permission from Elsevier, Reference 38] Figure 2.5 shows that the MA spinel particles (the bright phase) close some of the pores in the grain boundary region or convert them to closed pores, especially by their presence in triple points. On the other hand, these MA spinel grains can also be observed within the magnesia grains as inclusions. The grain size of magnesia is also 29 smaller as compared to Figure 2.4 but the dispersion of spinel structure in the magnesia matrix is not uniform. The pre-formed MA spinel is reported not to bond well with the magnesia grains (Figure 2.6). It has been reported [38] that preformed MA spinel in a magnesia matrix has a peripheral gap of 1 to 4 ?m around the MA spinel grain. This gap can allow the molten metal and slag to penetrate through, thus reducing the corrosion resistance of the magnesia based refractory system. During cooling, when all the grains contract, there is very little precipitation that can bond the grains together resulting in a peripheral gap that is more pronounced in Figure 2.6 than in Figure 2.5. MA spinel particles have been added in various proportions to MgO in order to improve its thermal shock resistance. The reason for the improvement in thermal shock resistance is linked to the difference in thermal expansion coefficient between magnesium oxide (mean value ~13.5?10?6 ?C?1) and MA spinel (~7.6?10?6 ?C?1). This difference causes micro-cracking, around spinel grains during cooling from fabrication temperatures in excess of 1600 ?C (Figure 2.7). However, the flexural strength is affected (i.e. decreases from ~240 MPa to ~105 MPa for 30 wt. % MA spinel) but the use of 0.5 ?m alumina particles that form spinel phase before/during sintering shows better strength (~180 MPa) compared to 3 ?m preformed MA spinel particles. The strength of the in-situ formed MA spinel based magnesia composite does not reduce with increasing alumina content beyond 20 wt. % of MA spinel [38]. The use of 30 very fine alumina that reacts to form in-situ MA spinel forms also a better bond than pre-synthesized MA spinel although it is seen from the microstructure that the grain size of spinel phase formed from 0.5 ?m alumina particles is almost similar to the grain size of the 3 ?m MA spinel powder. Figure 2.7: Flexural strength of magnesia-spinel composites as a function of MA spinel content [Reprinted from C. Aksel, P. D. Warren and F. L. Riley, Magnesia-spinel microcomposites, Journal of the European Ceramic Society, 24 (10-11), 2004, pp. 3119-3128, with permission from Elsevier, Reference 38] This may be related to better bonding of the in-situ MA spinel formation as compared to the no bonding in the preformed MA spinel and also to the volume expansion during spinel structure formation that causes it to fill up the pores. However, if the MA spinel 31 particles formed during sintering are not small, the volume expansion may result in cracks and thus affect the flexural strength. In order to achieve better bonding of periclase grains by the spinel structure in the matrix, oxides that form the spinel structure dissolve in magnesia at high temperature and precipitate on cooling such that they bond the interface of two periclase grains. In order to achieve better bonding, the seeding effect [39] of fine preformed MgAl2O4 spinel in magnesia-rich compositions has resulted in densification at 1600 ?C. However, the presence of direct bonding [40] could not be established as compared to magnesia-chrome bricks. The low corrosion resistance of periclase- MA spinel bricks may be because of the low solid solubility of both magnesia and alumina in MA spinel at high temperature as compared to magnesia and chrome ore, thus not yielding the desired bonding of the grains even in dense specimens. The particle size of magnesia grain affects [41] its diffusion in fine alumina, resulting in fine alumina being bonded to the periphery of the magnesia grain, but the linear expansion behaviour shows that smaller magnesia particles result in a gradual volume expansion and MA spinel formation at lower temperature. The mismatch in thermal expansion between periclase and MA spinel results in tensile stresses on the periclase grains during cooling, which may cause cracking of the periclase grains. The result of using large magnesia grains is that it results in cracking of the grains during cooling, thus exposing periclase to the surface, making the magnesia-normal spinel refractory susceptible to corrosion and hydration. 32 2.6. Magnesium aluminate spinel formation through physico-chemical methods and its effect on sintering A recent approach to reduce the sintering temperature of spinel based refractories has been through chemical methods. Through this process, low temperature reactions and formation of intermediates can result in low temperature synthesis and reaction sintering of MA spinel. Stoichiometric MA spinel formed by sol gel [42] results in lower synthesis temperature (T< 900 ?C) with the presence of free periclase. Chemical methods for formation of MA spinel at low temperatures include combustion synthesis [43, 44], co-precipitation [45, 46, 47], spray-pyrolysis [48, 49], plasma spraying [50], mechano-chemical synthesis [51] and combination of the methods [52, 53, 54]. In all these methods, however, the ratio of precursor volume to the final product is large and the shrinkage from precursor to dense material results in at least two firing stages, thereby compounding the manufacturing parameters. Nano-oxides and complex organo-metallic compounds formed during the initial heat-treatment reduce the diffusion path and decompose respectively, resulting in the spinel structure forming reaction and densification at lower temperature. A comparative study of hot pressed specimens of MA spinel and zirconia- MA spinel composites [32, 33] prepared from nano-powders of a mix of MA spinel and zirconia using the urea-nitrate combustion method showed that zirconia was present as a discrete secondary phase. The secondary phase improving the mechanical properties 33 of the ceramic, through micro-cracking and stress induced transformation in zirconia. The usage of precursor nano- MA spinel powder through the above-mentioned methods that are cost hindering are not used for manufacture of refractories, although some sols are used to create initial strength. Nano-particles [55] have a high surface to volume ratio and thus have a high driving force for both reaction and sintering. Densification and grain growth mechanisms during sintering of nano-materials have shown that they can agglomerate and coalesce at relatively lower temperature [56] resulting in coarsening of the grains that are responsible for closed pores and decreased reactivity. However the use of chemical methods for large-scale production of nano-oxides in refractories is generally not feasible. If a stable sol can be prepared such that when used in small quantities, it will act as a green binder and precursor for reaction sintering. Through these mechanisms, the periclase grains have direct bonding with the spinel structure thus created by the reactive sol. In the case of alumina rich spinel based castables, the use of sol gel spinel precursor as a binder and spinel structure former [57] wherein the solubility of alumina in spinel phase results in precipitation of alumina on cooling to room temperature similar to precipitation of complex spinel from magnesia-chrome bricks [58]. 34 The role of nano-materials is evident even in the case of fired magnesia-chrome ceramics, where vacuum impregnation with sol precursors [59] had a better corrosion resistance than non-impregnated bricks. Sols forming nano Cr2O3 formed better bonds to the exposed periclase grains as compared to sols forming nano MgCr2O4. Both variants of sols showed that impregnated bricks had lower apparent porosity and reduced mean pore diameter, making them candidates for improved corrosion resistance as compared to similar non impregnated direct-bonded magnesia-chrome ceramics. 2.7. Hydration of magnesia The preparation of both bricks and castables require water in order to shape. In the case of basic bricks containing free magnesia, the hydration of magnesia [60] can occur. This is caused by the reaction of magnesia with water that occurs during mixing and water vapor during the drying process, or even from the moisture in the environment, resulting in the formation of brucite, Mg(OH)2. The volume expansion due to this formation, on the basis that one mole of periclase (11.17 cm3 calculated from 40.31 g with specific gravity 3.58 g.cm-3 [2]) would react with one mole of water to form one mole of brucite (17.84 cm3 as calculated using 58.31 g with specific gravity 3.25 g.cm-3 [61]), is calculated to be 59.7 %. The standard free energy change due to reaction of 1 mole of water with 1 mole of magnesia at room temperature in air is -926 35 KJ [62] that decreases to zero at 350 ?C, which is considered as brucite decomposition temperature [63]. There are two main reported mechanisms for the kinetics of magnesia hydration, namely the ?shrinking core model? [64] and the dissolution-precipitation of magnesium oxide [65]. In the ?shrinking core model?, magnesia hydrates in two steps [64]. The initial formation of brucite layer at the grain boundaries without scope for free expansion [64, 66, 67], creates stress in the structure. As the hydration reaction proceeds, the forces created by the expansion are able to break grain boundaries, and result in smaller particles. Thus, both the surface area and the reaction rate increases exponentially until stable structure is attained. At this stage, the reaction rate decreases, and it is mainly influenced by water diffusion through the hydroxide layer [64]. This hydration mechanism does not have intermediate steps in the reaction, and there is no dissolution before the hydrated phase is formed. The second mechanism for magnesia hydration is generally directed at obtaining specific hydroxide morphology through initial dissolution. This step is followed by a slow nucleation process which precedes the precipitation of hydrated phases. This ?induction period? remains till the hydrate nuclei are formed. After the induction period, rapid precipitation results in crystal growth [65]. The mechanism based on the dissolution and precipitation of MgO particles has been assessed through different kinetic models and intermediate steps for the reaction have been proposed, namely (a) MgO-alkaline oxide acts as electron donor in water 36 (b) OH- anions are adsorbed on the positively charged surface (c) OH- anions are then desorbed from the surface, releasing magnesium ions (d) Ions concentration creates super-saturation, at which point the hydroxide starts to precipitate on the oxide surface. Considering the above mechanisms for the hydration of magnesia, in order to inhibit the formation of brucite, it is necessary to decrease the rate of one or both of the two following processes; the dissolution (a-c) or precipitation (d). This expansion will result in tensile stresses within the grains and result in initial micro-cracking, thus providing more magnesia to react with water and form more brucite. This will result in extensive cracking and dusting [68, 69] through the bricks. However, using magnesia as a hydraulic binder, where controlled brucite precipitation leads to enhanced mechanical strength of the castables [70]. This would result in a positive technological and economic impact due to its performance and use in making unshaped refractories. 2.8. Refractory degradation in simulated industrial conditions Basic refractories are used in the non-ferrous converters where the fayalite slag [58] reacts with periclase as FeO can form a solid solution with MgO and Fe2O3 can react to form MgFe2O4. The conditions within the converter can change between oxidizing and reducing during the operations resulting in cyclic oxidation and reduction of iron oxide 37 result in volume expansion and contraction of the solidified slag that penetrates through the pores in the refractory. This may result in stress induced cracks in the slag penetrated brick further increasing the slag penetration leading to greater corrosion and erosion of the refractory lining. Due to cyclic fluid oxidation and reduction processes at 1100 to 1350 ?C in the converter, in the presence of both liquid and gaseous reaction compounds and their high energy motion, result in non-laminar fluid flow effects and the molten slag and gases enter through pores in the refractory lining, thus reacting with the refractory and making it weaker [71] and susceptible to damage. The microstructure of chrome ore based basic refractories have, on the grains and grain boundaries of chromite and complex spinel grains, precipitation of a complex spinel structure, which is thermodynamically stable [35]. This precipitation occurs as the solid solution on the grain surface becomes saturated during cooling after the soaking period in the firing schedule. As the chrome ore contains iron oxide, it also decreases the solubility of iron oxide in the spinel matrix. Although the corrosion resistance of magnesia-chrome bricks is better than other refractories [58], their disposal has a potential hazard if the chromium compounds get oxidized and react with alkali and alkaline earth metal cations to form hexavalent chromium compounds. These compounds are soluble in acid rain and the dissolved hexavalent chromium is reported to have carcinogenic effects [72]. Chrome based refractories in iron and steel [73] as well as in cement [74] manufacturing processes has been replaced with alumina- MA spinel type refractories due to their easier disposal and 38 similar service life. In the case of non-ferrous converters, the use of such compositions would result in increased corrosion as the slag reacts with alumina and MgAl2O4. The corrosion resistance of periclase-MgAl2O4 spinel refractories in static conditions is poor as compared to magnesia-chrome refractories containing ~60 wt. % magnesia [58], which has been attributed to the low bonding of the periclase by MgAl2O4 spinel. The region of MA spinel precipitation show good resistance to corrosion but does not bond as well as the complex spinel in magnesia chrome refractories. 2.9. Summary ? The fabrication of periclase-MA spinel ceramics from constituent oxides as a replacement for magnesia-chrome refractories has not been successful to-date because of the diffusion-controlled reaction along with the associated volume expansion during the spinel structure formation. Moreover, the MA reacts with fayalite slag unlike the complex magnesia-chrome ore based spinel structure. ? The magnesia-chrome ore based complex spinel is used as refractory lining because of established manufacturing technology and its corrosion resistance to fayalite slags in non-ferrous converters as a result of the following: - The stability of inversion phases in the complex spinel structure, providing improved corrosion resistance. 39 - The ability to withstand the furnace conditions without degradation of the thermo-mechanical properties, which increases the service life. ? The difficulty in disposal of the spent refractory is due to the environmental concern with soluble chrome compounds, especially when Cr6+ is formed. Additionally, the possibility of Cr6+ formation during sintering and/or use of these refractories will be an increasing concern with tightened environmental guidelines. ? There are no studies substantiating the use of tetravalent oxide cations as replacement for chrome ore that can form basic refractories for use in non-ferrous converters, but these may be suitable replacements bearing into consideration that: - The tetravalent cations with both bivalent and trivalent cations can form complex spinel solid solutions similar to mag-chrome spinel. - The solid solutions are stable to temperature, pressure and chemical conditions making them reasonable substitutes for magnesia-chrome ore based ceramics. ? The X-ray diffraction technique and analyses of the diffraction patterns can provide data for calculations of the crystal structure of spinel solid solution, which can be compared to the theoretical inversion and lattice parameter. 40 ? The use of small quantities of nano-powders in processing of the refractories may be advantageous because of the following: - They can reduce the sintering temperature and diffusion distance. - There is anticipated increase in the bonding between the grains resulting in improved mechanical properties and chemical resistance. - The firing temperature can be reduced, so as to reduce the amount of grain coarsening of the periclase phase resulting in smaller fraction of closed pores. ? The development of basic ceramics (including bricks and castables) using brucite as a precursor to magnesia formation can be advantageous because of the following: - They can create reactive magnesia that reduces the sintering temperature. - The volume contraction from decomposition of brucite while forming periclase can be compensated by volume expansion due to spinel formation from the oxides. 41 3. Scope and objectives Magnesia-chrome refractory ceramics used in the non-ferrous industry for lining furnaces and convertors have two main phases, namely periclase (formula: MgO) and spinel (formula: (Mg,Fe)2+(Cr,Al,Fe)3+2O4). In the current work, the reduction of firing temperature and chromium bearing compounds are targeted. The cement and lime industries already use refractory ceramics containing corundum (formula: Al2O3) and MA spinel (formula: MgAl2O4). The spinels containing chrome ore in magnesia based refractory ceramic systems (formula: (Mg,Fe)2+(Cr,Al,Fe)3+2O4) were replaced by alumina because of the possibility of formation of Cr6+ at temperatures above 1400 ?C in the presence of alkali and alkaline earth metal cations in the slag. Chrome ore (formula: (Cr, Al, Fe3+)2O3) containing refractories bond with magnesium oxide (MgO) by reaction of periclase and complex spinel. In order to obtain sufficient bonding and thus mechanical strength (e.g. compressive strength > 20 MPa at room temperature), the state-of-the-art refractory oxides have to be fired at temperatures of 1730 - 1850 ?C for spinel forming and sintering. During cooling, the complex spinel precipitates from the periclase grains along grain boundaries, strengthening the structure. In order to maximize this effect, the solubility of complex spinel phase in magnesia during firing is enhanced through formation of solid solutions. In this research, the tetravalent ions in spinel structure that create inversion in position of the bivalent ions, similar to the behaviour exhibited by the Fe3+ ions in magnesia-chrome spinel 42 phase are explored in order to achieve chrome-free basic refractories that can be fired at temperatures below 1500 ?C, as opposed to the current technology of firing above 1750 ?C. The two salient features of this work are (1) reduction of greenhouse gases by reducing the firing temperature, and (2) ease of disposal of spent refractory by replacing chromium in the system by non-carcinogenic compounds. (1) This work tests the hypothesis that inducing inversion in spinel will increase the solid solubility range of the periclase and spinel phases, and that this phenomenon could lead to development of chrome-free refractories for non-ferrous smelters. (2) The work further attempts to demonstrate that such chrome free refractory ceramics fired below 1500?C is comparable to the current state-of-art magnesia-chrome refractory ceramics fired at temperatures in excess of 1730?C. The specific engineering science objectives of the project are as follows: 1. Investigation of the role of different spinel forming oxides that induces solid solution of the complex spinel and periclase phases similar to the effect of MA spinel and corundum in the magnesia-alumina system. This study is aimed at investigating the formation of periclase and spinel phases during the synthesis process. The spinel formers under consideration include MgO, Al2O3, Fe2O3, SnO2 and TiO2. 43 2. Analysis of the role of tetravalent cations (Me4+) and ferric ions in forming complex spinel with MgO and Al2O3 and its comparison with the complex spinel (Mg,Fe)2+(Cr,Al,Fe)3+2O4 in re-bonded fused magnesia-chrome ore based refractory ceramics with regard to the calculated inversion in the spinel structure. 3. Study of the simultaneous synthesis and sintering (through the formation of intermediate phases at lower temperatures) of oxide constituents so that periclase and complex spinel phase are formed in dense ceramic structure. The application of this study is to form binding systems including periclase and complex spinel for refractory ceramics including bricks (Section 4.4) and castables (Section 4.5) that will not contain chromium ions and can be manufactured at lower firing temperatures. The refractory ceramics developed using MgO-Al2O3-FeOx-Me4+O2 system to form the binding system in this current study are compared to the currently used rebonded fused grain magnesia-chrome refractories used in the non-ferrous industry. The basic refractory ceramics developed with in-house chrome-free binding system using magnesia-chrome ore based fused aggregates is compared with the current standard magnesia-chrome refractories in terms of their performance in a bottom blown oxygen convertor through post trial analyses (Section 4.6). 44 4. Methodology 4.1. Calculation of theoretical inversion in complex spinel The spinel structure consists of a cubic close packed (CCP) lattice of thirty-two oxygen ions with one eighth of the sixty-four tetrahedral voids and half of the thirty two octahedral voids are occupied by cations, resulting in a total of twenty four cations in the spatial arrangement of eight formula units (Figure 2.2, Section 2.2). For this kind of a structure, the representation of the lattice is done by marking the bivalent cation as A, trivalent cation as B, tetravalent cation as D, and the bivalent anion as X, which makes up a structural formula of A8B16X32 or A16D8X32, or as a formula unit denoted by AB2X4 and A2DX4. A spinel solid solution containing all the cations is represented by A1+zB2-2zDzX4, based on the solid solution containing (1-z) moles of AB2X4 and z moles of A2DX4. Based on the available literature data (refer to Section 2.2), the preferred position of the cations in a complex spinel structure was ascertained for calculating the inversion in spinel solid solution as described below: Al3+ ions occupy the octahedral site because of charge factor and empty ?d? shell, which results in formation of normal spinel structure with Mg2+ in tetrahedral site [21]. 45 Cr3+ ions occupy the octahedral site because of charge factor and ?d? shell being filled at lower energy (dxy, dyz, dzx) with symmetric O2- ions, in accordance with the crystal field theory [22]. This results in the formation of normal spinel structure with Mg2+ in the tetrahedral site. Fe3+ ions mainly occupy the tetrahedral site, resulting in the bivalent ion being displaced to the octahedral site, thus forming inverse spinel [11, 23]. The Fe3+ ions are attributed with high spin and ?d? shell is filled at lower energy in non-symmetric positioning of O2- in the z-direction. This results in the formation of inverse spinel structure with Mg2+ in the octahedral site. However, the associated phenomenon of active super-exchange linkage in the Fe3+-O2--Fe3+ can account for 3% of Fe3+ being present in the octahedral site, which result in a spinel with 97% inversion [20]. Ti4+ ions occupy the octahedral site because of charge factor and empty ?d? shell. This results in formation of normal spinel structure with Mg2+ in tetrahedral site [21]. Sn4+ ions predominantly occupy the tetrahedral site because of filled ?d? shell. However, the completely filled ?4d? shell in spite of the empty ?5s? influences the site occupancy and 5% of the Sn4+ is assumed to be present in the octahedral site. This results in the bivalent ion being displaced by 95% of the Sn4+ ions to the octahedral site, resulting in the formation of inverse spinel structure [20]. 46 On the basis of the above assumptions, the theoretical inversion in complex spinel solid solution was calculated by combining the literature data on the preferred position of different cations [11, 13, 20, 21, 22, 23] in spinel solid solution. Various researchers [22, 35, 36] have estimated the inversion in spinel phase based on crystallographic studies at different temperatures and pressures. In the current calculations, we have combined their methods and used the predominantly occupied sites of different cations in spinel solid solution to estimate the inversion in spinel using the following steps: (1) AT = B / 2 + D - (0.97 ? B5 / 2) - (0.95 ? D10) Equation 4.1.1 (2) AO = A - AT Equation 4.1.2 (3) % Inversion = (AO / A) ? 100 % Equation 4.1.3 Where, A = Total moles of bivalent cations in the spinel phase (Mg2+) B = Total moles of trivalent cations in the spinel phase (Al3+; Cr3+; Fe3+) D = Total moles of tetravalent cations in the spinel phase (Ti4+; Sn4+) AT = Bivalent Ions in Tetrahedral Site of the spinel phase AO = Bivalent Ions in Octahedral Site of the spinel phase B5 = Cations with half-filled d-orbital (Fe3+) in the spinel phase D10 = Cations with completely filled d-orbital (Sn4+) in the spinel phase 47 The inversion in the complex spinel solid solution predicted through the above calculations was used to design the composition of the magnesia rich complex spinel based ceramic systems with the inversion parameters tailor-made to the magnesia-chrome ore based refractories that are currently used in the industry. The macro for calculating predicted inversion and calculations for values of A, B and D from wt. % of oxides is shown in Appendix A. For example, the predicted inversion of MTSF8 (Table 4.5) containing 60 wt. % MgO, 26 wt. % Al2O3, 8 wt. % Fe2O3, 3.5 wt. % SnO2 and 2.5 wt. % TiO2 is calculated as follows: Considering 100 g of MTSF8, Moles of Al3+: 0.510 (B) Required moles of A2+ ions: 0.255 Moles of Fe3+: 0.100 (B & B5) Required moles of A2+ ions: 0.050 Moles of Sn4+: 0.023 (D & D10) Required moles of A2+ ions: 0.046 Moles of Ti4+: 0.031 (D) Required moles of A2+ ions: 0.062 Total moles of A2+ required: 0.413 (A) Available in the form of Mg2+: 1.489 Hence, spinel and periclase are the two phases formed. From Equation 4.1.1, AT is calculated as 0.288 From Equation 4.1.2, AO is 0.125 Hence, the predicted % inversion is 30, using Equation 4.1.3. 48 4.2. Complex spinel preparation and characterization The oxide constituents in the compositions were designed on the basis of spinel solid solution formation and investigated for the role of tetravalent cation oxides in forming complex spinel with Mg2+ as the only bivalent cation in the system. The starting compositions were calculated on molar basis as illustrated in Appendix B so that the final molar composition of the complex spinel could be established with different partial fractions of the Mg2D4+O4 phases in the Mg1+zB2-2zDzX4 complex spinel solid solution. In this part of the work, oxides used had lower than 0.1 wt. % impurities, with particle size 0.5 to 1.4 ?m and supplied by Fisher Scientific (Appendix C). These samples were prepared by dry ball milling with alumina balls for 30 minutes in order to attain homogeneous mixing. Homogenised powder was then separated from the alumina balls and mixed with aqueous solution of organic binder before pressing the cylindrical pellets (15 mm height by 25 mm diameter) at 0.25 T/cm2. The pressed specimens of different compositions were dried overnight at 110 ?C and fired in air at 1400 to 1800 ?C with soaking at the maximum temperature for 3 h. The furnace was programmed for firing with heating rate of 10 ?C/min to 1000 ?C and then 5 ?C/min to the maximum temperature. After soaking the furnace was cooled naturally to room temperature, before the samples were removed and characterized for their phase content, density and microstructure. In order to ascertain the formation of non-stoichiometric complex spinel, the compositions were divided into different categories, as outlined below: 49 4.2.1. Category I compositions Stoichiometric complex spinel-forming oxide constituents were combined as shown in Table 4.1 and Table 4.2. The temperature of spinel solid solution formation and variation in the spinel lattice parameter, calculated from the whole pattern fit Reitveld analysis of X-ray diffraction data, was ascertained. The titania-containing samples are coded as ST#, where # stands for the calculated mole % of Mg2TiO4 spinel in the MgAl2O4-Mg2TiO4 solid solution. Similarly, the tin-containing samples are coded as SS#, in the same nomenclature. Table 4.1. Weight percent compositions of stoichiometric spinel mixes with TiO2 Constituent Oxides ST0 ST2 ST4 ST6 ST8 ST10 ST15 MgO (wt. %) 28.3 28.8 29.3 29.8 30.3 30.8 32.0 Al2O3 (wt. %) 71.7 70.1 68.5 66.9 65.3 63.7 59.7 TiO2 (wt. %) 0.00 1.1 2.2 3.3 4.4 5.5 8.3 True density (g.cm-3) 3.58 3.58 3.58 3.58 3.58 3.58 3.57 Predicted inversion parameter (%) 0 2 4 6 7 9 13 50 Table 4.2. Weight percent compositions of stoichiometric spinel mixes with SnO2 Constituent Oxides SS0 SS2 SS4 SS6 SS8 SS10 SS15 MgO (wt. %) 28.3 28.5 28.8 29.0 29.1 29.3 29.8 Al2O3 (wt. %) 71.7 69.4 67.1 64.9 62.8 60.7 55.7 SnO2 (wt. %) 0.0 2.1 4.1 6.1 8.1 10.0 14.5 True density (g.cm-3) 3.58 3.61 3.64 3.67 3.69 3.72 3.79 Predicted inversion parameter (%) 0 4 7 11 14 18 25 4.2.2. Category II compositions Complex spinel forming oxides with excess magnesia, which was kept constant throughout at 60 wt%, as shown in Table 4.3 and Table 4.4, could give insight to the amount of different tetravalent oxides that needed to be incorporated for good densification (less than 5 % open porosity) and formation of complex spinel structure (complete solid solution of spinel phase as determined through X-ray diffraction analyses). This was hypothesised to result in simultaneous synthesis and sintering through initial formation of metastable perovskite (MgMe4+O3) in magnesia-rich compositions, and we were the first to report this phenomenon, refer to Section 5.1. It 51 was also hypothesized to result in precipitation-bonding caused by dissolution of trivalent and tetravalent metal oxides in periclase structure at high temperatures of 1500 to 1800 ?C, and re-precipitation during cooling similar to the behaviour exhibited during firing of currently used basic refractories containing magnesia and chrome ore as the main constituents [40], refer to Sections 6.1 and 6.4. Table 4.3. Weight percent compositions of magnesia-rich spinel mixes with TiO2 Constituent Oxides MT0 MT2 MT4 MT6 MT8 MT10 MT15 MgO (wt. %) 60.0 60.0 60.0 60.0 60.0 60.0 60.0 Al2O3 (wt. %) 40.0 39.4 38.7 38.1 37.4 36.8 35.1 TiO2 (wt. %) 0.0 0.6 1.3 1.9 2.6 3.2 4.9 True density (g.cm-3) 3.58 3.58 3.58 3.58 3.58 3.58 3.58 Predicted inversion parameter (%) 0 2 4 6 7 9 13 52 Table 4.4. Weight percent compositions of magnesia-rich spinel mixes with SnO2 Constituent Oxides MS0 MS2 MS4 MS6 MS8 MS10 MS15 MgO (wt. %) 60.0 60.0 60.0 60.0 60.0 60.0 60.0 Al2O3 (wt. %) 40.0 38.8 37.7 36.6 35.4 34.3 31.7 SnO2 (wt. %) 0.0 1.2 2.3 3.4 4.6 5.7 8.3 True density (g.cm-3) 3.58 3.60 3.62 3.63 3.65 3.66 3.70 Predicted inversion parameter (%) 0 4 7 11 14 18 25 4.2.3. Category III compositions The amount of iron oxide for fabricating ceramics with MgO-Al2O3-Fe2O3-Me4+O2 complex spinel was based on inversion in the complex spinel (17 to 30 %) resembling that calculated for magnesia-chrome ore based refractory ceramics (Table 5.7) as in Section 4.1 (refer to the calculation example and macro in Appendix A). Table 4.5 shows the composition of basic refractory ceramics in periclase ? chrome-free complex spinel systems. 53 Table 4.5. Weight percent compositions of magnesia-rich spinel mixes Constituent Oxides MTSF0 MTSF4 MTSF8 MgO (wt. %) 60.0 60.0 60.0 Al2O3 (wt. %) 34.0 30.0 26.0 TiO2 (wt. %) 2.5 2.5 2.5 SnO2 (wt. %) 3.5 3.5 3.5 Fe2O3 (wt. %) 0.0 4.0 8.0 True density (g.cm-3) 3.63 3.67 3.72 Predicted inversion parameter (%) 17 24 30 4.2.4. Category IV compositions Use of nano-sized oxide powders, with high surface to volume ratio and thus higher driving force for both reaction and sintering, was studied to better understand their effect on temperature of reaction-sintering. In these Category IV compositions, industrially manufactured nano-powders were incorporated in the mixes to increase diffusion into magnesia to encourage precipitation of complex spinel-forming precursors. It was hypothesized that the increased diffusion through nano-powders would perform similar 54 to the precipitation of complex spinel from magnesia in magnesia-chrome ore based bricks. The roles of the nano-oxides in MgO-Al2O3-Fe2O3-Me4+O2 mixes were analysed on the basis of phase formation and physical properties, in comparison to similar composition without nano-oxides. For this analysis, three systems were studied through substitution of either TiO2 and/or Al2O3, while keeping the compositions as shown in Table 4.5. The nano alumina (Aeroxide Alu C, crystallite size 13 nm, surface area 100 ? 15 m2.g-1) and nano titania (Aeroxide TiO2 P25, crystallite size 21 nm, surface area 50 ? 15 m2.g-1) were procured from Evonik Degussa Corporation. A comparison between systems (a) N1 and N2 in Table 4.6 would depict the effect of different amounts of nano-oxides, and (b) N2 and N3 would show the effect of bi-modal nano-powder distribution and its effect on phase formation and physical properties. Table 4.6. Weight percent substitution by nano-oxide powders in magnesia-rich spinel mixes Nano-substitution N1 N2 N3 Al2O3 (wt. %) 0.0 0.0 1.0 TiO2 (wt. %) 1.0 2.6 1.6 55 The phase analyses of sintered specimens were done using powder diffraction data collected through Rigaku X-Ray Diffraction (XRD) and whole pattern fitting of XRD patterns using JADE software [75]. The fired samples were powdered in agate mortar and pestle in order to mount it in a powder specimen holder for Rigaku X-ray diffractometer. The detector was programmed to scan through diffraction angle (2?) of 5 to 90? at scan rate of 2? per minute with step size of 0.02? and 0.6 s dwelling time. X-ray diffraction data of samples were analysed using techniques of peak search match and results were correlated using the Whole Pattern Fitting (WPF) method based on Reitveld refinement for lattice parameter calculation using JADE, the X-ray diffraction data analysis software. Density measurements were done using the ASTM standard method of boiling in water [76]. The fired samples were labelled and then weighed in a top loading electronic balance and dry weight was recorded. These were then immersed in water in sealed container connected to vacuum pump and samples were thus boiled in water at room temperature for 5 h in vacuum. Samples were then weighed for their suspended weight in water. Wet samples are gently tapped with a moistened cloth and then weighed again for their weight while saturated in water. The microstructure of specimens fired at different temperatures were observed in Hitachi Environmental Scanning Electron Microscope (SEM) after polishing a section of 56 selected specimens mounted in resin and the grains and binding system were analysed in terms of their shape, size and entrapped pores. 4.3. Hydration of the aggregates Due to effects of hydration in the systems containing reactive magnesia, that are used for making basic refractories, and subsequent brucite formation during mixing and drying, aggregates were checked for their hydration resistance. In order to assess water absorbed due to the hydration, the different size fraction of aggregates were dried and calcined for weight loss, due to hydration in storage conditions. The magnesia-chrome grains (Table 4.7) prepared by sintering (C1 and C2) were from Refratechnik Steel GmbH China Department, Liaoning, China and that by fusion (C3) were from Washington Mills, Niagara Falls, ON, Canada. Aggregates were kept in contact with moisture by immersion in water for 48 h (based on the time for which castables are kept in mould and air dried before being dried) and then dried at 110?C for 24 h in order to investigate the weight gain due to formation of brucite. 57 Table 4.7. Composition and calculated inversion of magnesia-chrome aggregates Composition (%) C1 C2 C3 MgO 57.4 54.2 57.6 Cr2O3 23.0 27.9 20.0 Al2O3 6.5 5.2 7.9 Fe2O3 12.0 10.0 12.9 CaO 0.6 0.6 0.7 SiO2 0.5 1.1 0.9 Predicted inversion parameter (%) 25 18 27 Chrome-free aggregates (different size grades, namely 3-6 mm, 1-3 mm, 0-1 mm, and fines below 180#) were also prepared, based on outcomes in Chapter 5, by crushing fired cylinders (diameter 50 mm, height 25 mm) made similarly as in Section 4.2 using composition MTSF8, in order to investigate if these were suitable for large production and fabrication of bricks and castables. 58 4.4. Preparation and characterization of bricks using aggregate and binder systems Based on the preliminary results of the binder systems studies, the bricks were prepared using grains and fines with binding systems (MTSF4 and MTSF8) envisaged for large specimens. Sintered magnesia (97% purity, Baymag Inc, Exshaw, AB, Canada), and magnesia-chrome grains prepared by both sintering (Refratechnik Steel GmbH China Department, Liaoning, China) and fusion (Washington Mills, Niagara Falls, ON, Canada) process were procured from the industries as in parenthesis, and used as the coarse fraction for characterizing bricks. Through this method, characterisation of binding system in terms of bonding with the grains could be analysed. Bricks were prepared by dry balling fines with alumina balls for 30 minutes and then separated from balls. Further mixing was carried out in a laboratory paddle mixer (Hobart model N50) with aggregates for 5 minutes and then with aqueous solution (8 wt. %) of organic binder (polyvinyl alcohol and/or methocel) for a minute before pressing. These were fired at 1450 ?C with soaking at maximum temperature for 5 h after being dried overnight at 110 ?C. Furnace was programmed for firing with a heating rate of 8 ?C/min to 800 ?C and then 5 ?C/min to maximum temperature. After soaking, the furnace is cooled naturally to room temperature. Fired samples (Figure 4.1 shows a 230 mm ? 230 mm ? 76 mm pressed refractory ceramic after thermal treatment) were characterised for densification using the ASTM 59 standard method of boiling in water [76]. They were weighed in a top loading electronic balance and dry weight was recorded. These were then immersed in water in sealed container connected to vacuum pump and the samples were thus boiled in water at room temperature for 5 h in near vacuum. Samples were then weighed for their suspended weight in water. Wet samples are gently tapped with a moistened cloth and then weighed again for their weight while saturated in water. Figure 4.1. Pressed sample fired at 1450 ?C, 5 h Modulus of rupture (MOR) was measured by 3 point bending method using 176 mm ? 25 mm ? 25 mm (7? ? 1? ? 1?) bars. Values for MOR were calculated from the value of the load (P) from Instron universal testing machine based on the ASTM standard for 60 testing with cross-head speed of 2.54 mm/min [77]. Formula used for this calculation is: MOR = (3 ? P ? l) ? (2 ? b ? d2), where l is the span between two supports, b is breadth and d is depth of fractured specimen. Compressive strength (CCS) was measured from the broken half bar further cut to 25 mm (1?), thus resulting in 25 mm (1?) cube. Cubes were crushed by load plate and maximum load (P) from Instron universal testing machine were used to calculate compressive strength. Based on ASTM standard for compressive strength [77], applied loading rate was 0.24 MPa. The formula used for compressive strength CCS calculation was CCS = P ? (b1 ? b2), where b1 and b2 are breadths of the specimen parallel to the pressing plates. 4.5. Preparation and characterization of vibratory castables using aggregate and binder systems Vibratory castables were prepared using novel in-house binding system. Fused magnesia-chrome grains (C3) was procured from industry to be used as coarse fraction for castables. Through this, large samples could be prepared so that they could also be installed for after-use analysis of spent refractory in industrial convertor. 61 Chemical analyses of the aggregates are listed in Table 4.7 and they were sieved in order to get 4 size fractions in order to get castables with existing multi-modal system developed in-house [78]. Through this method, characterisation of binding system in terms of bonding with grains could be analysed. Castables were prepared by mixing fines with coarse fraction of aggregates for 10 minutes and then further mixing was carried out in a laboratory paddle mixer with other size fractions of aggregates for 5 minutes. Samples were then wet mixed for 5 minutes after adding deionized water and cast using vibration table. Figure 4.2: Flow table used for castables 62 Castables were then cured in ambient temperature at 100% relative humidity for 24 h and then de-moulded and allowed to cure for another 24 h in humid condition before being dried at 60 ?C for 10 h followed by further drying at 110 ?C for 16 h. Appendix E shows various mixes tested in order to attain the desirable flow and casting properties for the optimized compositions, particle size distributions and drying conditions for the plant trials. Flows of various green castable compositions were measured using flow cone according to the ASTM standard [79]. Wet mixed castable was poured in flow cone [? (100-70) ? 50 mm] and vibrated (Figure 4.2) with large end up till the upper surface became smooth and evened out with upper surface of cone. This was then placed centred on flow table with smaller end up and tapped down 15 times in 9 s. Increase in average base diameter of castable from its initial base diameter was expressed as the percent flow value. Maximum working time was measured on the basis of the ability to shape castable either by hand or on vibration table. Initial setting time was measured when vibration would result in no change in shape or finger impression on castable. These were fired at 1450 ?C with soaking at the maximum temperature for 5 h. The furnace was programmed for firing with heating rate of 8 ?C/min to 800 ?C and then 5 ?C/min to maximum temperature. After soaking at maximum temperature, furnace was cooled naturally to room temperature. 63 The fired vibratory castable samples (e.g. Figure 4.3) were characterised similar to the pressed bricks in Section 4.4. Densification studies were done using the ASTM standard method of boiling in water [76]. The MOR was measured by 3 point bending method using 176 mm ? 25 mm ? 25 mm (7? ? 1? ? 1?) bars. Values for MOR were calculated from load (P) recorded from Instron universal testing machine based on ASTM standard [77]. The CCS was measured from (a) cast 25 mm (1?) cubes, and (b) broken half bars further cut to 25 mm (1?), thus resulting in 25 mm (1?) cube. Cubes were crushed by load plate and maximum load (P) from Instron universal testing machine was used to calculate compressive strength according to ASTM standard for compressive strength. Figure 4.3: Cured and dried vibratory castables after firing at 1450 ?C, 5 h 64 The microstructure of vibratory castables fired at 1450 ?C, 5 h were observed for the aggregate and binding system in Hitachi Environmental Scanning Electron Microscope (SEM) after polishing a section of fractured specimens, which was mounted in resin, and were analysed in terms of their shape, size, interface structure and entrapped pores. 4.6. Characterization of the trial refractory ceramics containing chromium-free binder system and comparison with magnesia-chrome bricks After the cast half-bricks were fired and characterized, they were sent for installation in a Bottom Blown Oxygen Converter (BBOC) at Teck Cominco (Trail, BC, Canada) and used comparatively with regular standard industrial bricks. The bricks faced the following highly variable environment: (i) temperature cycling from 900 to 1300 ?C, (ii) molten slag rich in variable amounts of PbO (> 25 wt. %), Bi2O3 (> 30 wt. %), FeO (> 5 wt. %) with small amounts of Sb2O3, CaO, SiO2, (iii) molten metal including Pb, Ag, Sb and Bi, and (iv) cyclic shift in gaseous environment from slightly reducing to oxidizing when hot air is blown from the bottom. After use for 40-45 days and shut-down of BBOC, characterization of both types of spent refractories was carried out to compare and understand the performance of developed refractory ceramic with chromium-free binder system. Dimensions of the bricks, as received, were measured for the trial half-brick (UT1) and for the re-bonded fused grain magnesia-chrome ore based standard 65 industrial brick (F15). The longer of the two UT1 half-bricks was selected for the comparative analyses. F15 brick was cut in a half and selected for the post-mortem analyses. Selected wedge-shaped bricks were trimmed to obtain rectangular shapes, which were sectioned longitudinally with wet brick saw, perpendicular to the hot face (i.e. the face that is directly affected by and in contact with the melt). Similarly, cold face is the face that is furthest away and in contact with insulation or shell of convertor. Three 25 mm thick sections were obtained from UT1 brick; the two lateral ones were used to test mechanical and physical properties, and the one in the middle was used to test the microstructure by SEM, and the mineralogical phases by XRD and elemental compositions by EDS. The side slices were cut parallel to the cold face to obtain 25 x 25 x 118 mm bars from UT1 and 20 x 20 x 120 mm bars from F15. These specimens were tested for modulus of rupture (MOR) with the pressing direction along to the hot face. The MOR was measured by 3 point bending method based on the ASTM standards [77]. As observed from the CCS values for samples in Section 4.5, there was no significant difference for cast cubes and cubes cut from samples for MOR tests. Hence, after testing for MOR, these specimens were further cut parallel to hot face to obtain 25 mm and 20 mm cubes respectively for compressive strength (CCS) tests. Cubes were crushed by load plate and the maximum load (P) from Instron universal testing machine 66 were used to calculate compressive strength, based on ASTM standards [77]. Remaining pieces, after cutting cubes, were used for absorption tests using ASTM standard method of boiling in water [76]. Two 10 mm wide bars were also cut longitudinally from the central portion of the middle slice, and used for SEM, EDS and XRD. Phase analyses of the sintered specimens were done using powder diffraction data collected through Rigaku X-Ray Diffraction (XRD) and whole pattern fitting of the XRD pattern using JADE software. Fired samples were powdered in order to mount it in powder specimen holder for Rigaku X-ray diffractometer. X-ray diffraction data of the samples were used to identify the phases in the spent refractory by matching it to diffraction data files. Microstructures at different positions of spent refractory, based on distance from the hot face, were observed in Hitachi Environmental Scanning Electron Microscope (SEM). The samples for microstructural analyses were prepared by polishing a section after mounting in resin. The binding systems were analysed in terms of their shape, size and entrapped pores. Elemental distributions of metal and slag that penetrated into the refractory during the operation of BBOC campaign were mapped from hot face to cold face of the used refractory. 67 5. Development of dense ceramic structure as binder systems for refractories In this chapter of the thesis, a study of the lattice parameters with different spinel forming cation oxides used in conjunction is correlated to their densification behaviour. The target of this study is to achieve a dense ceramic that could serve as a potential chromium free binder system for basic refractories. As the spinel phase in magnesia-chrome ore based refractories is a complex spinel containing mainly Mg2+, Al3+, Cr3+ and Fe3+, this section reports the results of synthesizing and characterizing well-defined complex spinel solid solutions in order to understand the role of firing temperature on their lattice parameters and densification (the processing and characterization methods details of these materials is reported in Section 4.2). Through selective optimisation of cation substitutions using MgAl2O4 spinel as the basis structure, it is envisaged that dense binder system could be developed, so that it can be used with aggregates for making bricks and castables. 5.1. Role of tetravalent oxides in forming complex spinel with MgO and Al2O3 The formation of spinel solid solutions using tetravalent cation oxides was envisaged through shift of the 2? values for the [311] spinel peaks at its maximum intensity indicating changes in the lattice parameters for both tetravalent cation additives (TiO2 and SnO2) as shown in Table 5.1 and Table 5.2. The additives are depicted as 68 calculated mole percent of the spinel phase Mg1+zAl2-2zDzO4, where D denotes the tetravalent (Sn or Ti) cation, for ease of comparative analysis. For example, 10 mole % TiO2 substitution forms Mg1.1Al1.8Ti0.1O4 spinel solid solution (s.s.) phase. The instrument was calibrated using quartz crystal as the internal standard and the 2? values were accurate to ? 0.04 degrees at 26.65 degrees. Table 5.1. Spinel peak shift with addition of TiO2 and SnO2 in stoichiometric spinel compositions Mole % Mg2TiO4/Mg2SnO4 0 4 8 15 Max 2? (TiO2) 36.88 36.92 36.96 37.02 Max 2? (SnO2) 36.84 36.76 36.74 36.72 Table 5.2. Spinel peak shift with addition of tetravalent cation oxide additions in magnesia-rich compositions Mole % Mg2TiO4/Mg2SnO4 0 4 8 15 Max 2? (TiO2) 36.70 36.80 36.80 36.92 Max 2? (SnO2) 36.78 36.62 36.46 36.60 69 Appendix D shows the instrumental errors for 2? values and in estimation of lattice parameters. In the case of using titania, there was decrease in the lattice parameter. The lattice parameter increased with increasing the amount of Sn4+ ions, which can be attributed to larger radius of Sn4+ cation. Figure 5.1: XRD patterns of stoichiometric spinel samples containing Mg2SnO4 fired at 1450 ?C, 3h The effect of magnesia dissolving in spinel (Figure 5.1) was observed through the decrease in lattice parameter or increase in the 2? values (Table 4.1) of the complex spinel solid solution (CSSS) phase for the TiO2 based compositions. 70 However, for the SnO2 based compositions, spinel peaks decreased in intensity (Figure 5.2) at ~31 and ~37 degrees for spinel as compared to the peak at ~42 degrees for periclase. The lattice parameter increases, showing that there have been different crystallographic arrangements formed or dissolved in periclase forming solid solution of magnesia rich compositions. The increase in intensity of secondary spinel peak at ~17 degrees along with the peak at ~18 degrees indicated the presence of two spinel phases, rather than a spinel solid solution, after firing at 1450 ?C. Figure 5.2: XRD patterns of magnesia-rich spinel samples fired at 1450 ?C 71 Table 5.3. Effect of additives on bulk density (g/cm3) of stoichiometric compositions Firing Temperature Percent Additive 0 2 4 6 8 10 15 1400 ?C SnO2 1.48 1.51 1.52 1.54 1.57 1.61 1.69 1400 ?C TiO2 1.48 1.50 1.51 1.53 1.56 1.57 1.60 1450 ?C SnO2 1.56 1.57 1.57 1.57 1.63 1.65 1.69 1450 ?C TiO2 1.56 1.58 1.56 1.57 1.63 1.65 1.69 Table 5.4. Effect of additives on % true density of stoichiometric compositions Firing Temperature Percent Additive 0 2 4 6 8 10 15 1400 ?C SnO2 41.4 41.9 41.8 42.1 42.4 43.4 44.6 1400 ?C TiO2 41.4 41.8 42.2 42.8 43.7 44.0 44.7 1450 ?C SnO2 43.7 43.6 43.2 42.9 44.0 44.4 44.7 1450 ?C TiO2 43.4 44.1 43.7 44.0 45.5 46.3 47.4 72 Table 5.5. Effect of additives on bulk density (g/cm3) of magnesia-rich compositions Firing Temperature Percent Additive 0 2 4 6 8 10 15 1400 ?C SnO2 2.41 2.37 2.34 2.30 2.28 2.31 2.32 1400 ?C TiO2 2.40 2.37 2.33 2.29 2.28 2.29 2.34 1450 ?C SnO2 2.69 2.62 2.59 2.57 2.57 2.56 2.60 1450 ?C TiO2 2.70 2.61 2.57 2.54 2.55 2.56 2.59 Table 5.6. Effect of additives on % true density of magnesia-rich compositions Firing Temperature Percent Additive 0 2 4 6 8 10 15 1400 ?C SnO2 67.2 65.9 64.7 63.3 62.5 62.9 62.6 1400 ?C TiO2 67.1 66.0 65.0 64.0 63.8 64.1 65.3 1450 ?C SnO2 75.0 72.7 71.7 70.8 70.5 69.8 70.2 1450 ?C TiO2 75.3 72.8 71.7 70.8 71.1 71.4 72.3 In case of using TiO2 and SnO2 additives, bulk density was lower for stoichiometric formulations as compared to magnesia rich formulations (Table 5.3; Table 5.4; Table 73 5.5; Table 5.6). This could be attributed to the fact that formation of spinel is a process involving increase in volume and hence did not sinter during the diffusion-controlled synthesis as the low rate of spinel formation did not result in closure of pores. In spite of the formation of magnesia-rich spinel for magnesia-rich compositions, there was free magnesia that could sinter and also dissolve spinel that precipitated on cooling, thus increasing the bulk density. Figure 5.3: The effect of TiO2 additive [as mole % Mg2TiO4 in spinel solid solution] on the apparent specific gravity (ASG) of stoichiometric (S) and magnesia rich (M) spinel compositions fired at 1400 and 1450 ?C. 74 The average apparent specific gravity (with the overall measurement error estimated at ? 0.03 g.cm-3) indicated that stoichiometric spinel samples have a minor increase in the values at 1400?C for TiO2 additions (Figure 5.3) as compared to SnO2 additive (Figure 5.4). This could be attributed to the fact that Mg2TiO4 has a lower true density compared to Mg2SnO4. Figure 5.4: The effect of SnO2 additive [as mole % Mg2TiO4 in spinel solid solution] on the apparent specific gravity of stoichiometric (S) and magnesia rich (M) spinel compositions fired at 1400 and 1450 ?C. The apparent specific gravity data at 1450?C showed that TiO2 addition resulted in higher values for stoichiometric spinel, similar to magnesia rich spinel compositions at 75 6% and higher addition of TiO2. In case of SnO2 additions, the apparent specific gravity data for magnesia rich compositions fired at 1400?C resembled closely the values for stoichiometric compositions fired at 1450?C and they increased with increasing the firing temperature. However at 15% SnO2 addition, it was seen that the apparent specific gravity was higher for compositions fired at 1400?C as a result of high water absorption, which may be caused either due to the volume expansion during spinel formation or because of hydration of magnesia. Figure 5.5: The effect of additives SnO2 and TiO2 [as mole % Mg2TiO4 / Mg2TiO4 in spinel solid solution] on the apparent porosity (AP, vol. %) of magnesia rich spinel compositions fired at 1400 and 1450 ?C. 76 The average apparent porosity (Figure 5.5) (maximum deviation of ? 0.4 % from the average value) of the magnesia-rich spinel samples showed similar trend of increasing porosity for TiO2 and SnO2 additives. They enhanced the spinel formation and thus created a volume expansion during synthesis. At 6-8% addition of tetravalent cation oxides, the apparent porosity was highest. It also marked the maximum change in the lattice parameters for spinel (Table 5.1 and Table 5.2) indicating that solid solution formation did not assist sintering in a single stage firing. As the above results could not explain the formation of spinel at low temperatures, compositions were fired at 1600 ?C to 1800 ?C, in order to ascertain the reactive spinel formation. The dimensions of the unit cell of spinel solid solution (Figure 5.6) (angular error in XRD data ? 0.04 degrees, combined refinement error including the 2? correction factor and pattern fitting ? 0.02 ?) varied with the composition suggest that solid solutions had different amount of cations at different mix compositions. The pattern fitting of XRD data for samples fired at 1600?C to 1800?C resulted in unit cell dimensions for a single spinel phase that were greater than the unit cell dimensions observed for the composition without SnO2 additive. The spinel solid solution showed decrease in the lattice parameter with increase in firing temperature, unlike analysis for no additive. Compositions with 8 mole % Mg2SnO4 showed inflexion in lattice parameters. Increasing the SnO2 content resulted in large reduction in the size of spinel unit cell as firing temperature was changed from 1700?C to 1800?C (considering the confidence interval of ? 0.02 ? for Reitveld analysis using JADE software). The unit cell 77 dimensions increased with SnO2 content for the samples fired at 1700?C, but had inflexions at 8 mole % Mg2SnO4 in spinel solid solution when fired at 1600?C and 6 mole % for 1800?C firing. The change in unit cell dimension from 4 to 6 mole % Mg2SnO4 was comparatively the highest over all temperature ranges. It was assumed that the increase in lattice parameter was a result of substitution defects in the spinel structure as Mg2SnO4 has a larger unit cell than MgAl2O4. Figure 5.6: Variation of unit cell dimensions with mole % Mg2SnO4 in spinel solid solution fired for 3 h (lines joining the points are drawn for easier readability) The unit cell length of spinel solid solution increased with TiO2 content (Figure 5.7) and decreased with firing temperature, similar to the result for SnO2. Compositions 78 containing TiO2 had complete spinel solid solution formation. The unit cell size increased for all compositions with increase in temperature from 1500?C to 1600?C, except for 8 mole % Mg2TiO4. This composition showed a gradual decrease in the size of unit cell, unlike the compositions with lower TiO2 that had almost the same unit cell size for firing at 1600?C to 1800?C. Figure 5.7: Variation of unit cell dimensions with mole % Mg2TiO4 in spinel solid solution; samples fired for 3 h at temperatures indicated in the figure The 10 mole % Mg2TiO4 showed increase in the size of unit cell from 1500?C to 1700?C, followed by reduction at 1800?C and 15 mole% had an increase unit cell with increasing temperature. The formation of perovskite at low temperatures in the initial stage might have increased the reactivity of Mg2+ ions into the system resulting in the formation of bonding during reaction sintering to form spinel. The increased lattice 79 parameter of spinel solid solution as compared to MgAl2O4 was assumed to be a result of defect structure caused by the Ti4+ ion that forms Mg2TiO4, with a larger unit cell. The substitution of Al2O3 by SnO2 showed an increase in the open porosity (Figure 5.8) (maximum deviation from the mean values +0.6; -0.5) for specimens fired at 1800?C, accompanied by a decrease in the density. The apparent specific gravity increased with incremental increase in SnO2 content to 4 mole % Mg2SnO4 in spinel solid solution as calculated from the oxide composition. Further increase in SnO2 content reduces apparent specific gravity. Figure 5.8: Physical properties of pellets vs. mole % Mg2SnO4 in spinel solid solution fired at 1800?C, 3 h 80 The true density of Mg2SnO4 spinel (4.77 g.cm-3) is greater than both magnesia (3.59 g.cm-3) and MA spinel (3.58 g.cm-3) and theoretically, the pellets containing SnO2 in the spinel solid solution would have a higher specific gravity. The reduction in apparent specific gravity might have been because of a phase of lower true density or due to closed pores within the grains. The increase in open porosity led to the hypothesis that SnO2 alone is not a good sintering additive, but it might be thermodynamically more stable because of its ability to form complex spinel structure [21, 28]. Figure 5.9: Physical properties of pellets vs. mole % Mg2TiO4 in spinel solid solution fired at 1800?C, 3 h 81 Figure 5.9 shows the decrease in apparent specific gravity with increasing TiO2 content. The open porosity was at minimum value (4.2 vol. %) at 8 mole % Mg2TiO4 in the spinel phase. Further TiO2 addition increased the open porosity. The decrease in apparent specific gravity might be related to the true density of Mg2TiO4 (3.54 g.cm-3) and the creation of closed pores within the grains. The open porosity at 6 to 8 mole % Mg2TiO4 in spinel solid solution had apparent porosity less than 5%. The microstructure and combined effect of tetravalent and trivalent cations are discussed in Section 6.1. The variation of spinel lattice parameters (Figure 5.7) from 6 to 10 mole % Mg2TiO4 in the spinel solid solution and its decrease with increase in firing temperature gave optimum results in terms of achieving desired apparent porosity (Figure 5.9) for forming a ceramic matrix. This signifies that the defect structures formed during heating to the maximum temperature resulted in increased unit cell dimensions of the spinel solid solution, which led to the compensation of volume expansion during spinel formation at higher temperature (during soaking at maximum temperature). The composition with 6 mole % Mg2SnO4 had the same variation in the lattice parameter of spinel and might have more closed pores than those with lower SnO2 content. 5.2. Calculations of inversion in spinel solid solutions The compositions of the complex (Mg2+, Fe2+)O?(Cr3+, Fe3+, Al3+)2O3 spinel solid solution in different magnesia-chrome bricks were estimated from their chemical 82 analysis and used to calculate the inversion using oxide composition (Section 4.1) based on the literature (Section 2.2) as shown in Table 5.7. Table 5.7. Composition and calculated inversion of magnesia-chrome bricks Brick 1 Brick 2 Composition (%) MgO 59.0 54.2 Cr2O3 21.0 27.9 Al2O3 7.0 5.2 Fe2O3 11.5 10.0 CaO 0.5 0.6 SiO2 0.5 1.1 Calculated inversion of spinel (%) 25 18 On the basis of the inversion values, chromium-free compositions were formulated as in Table 4.5 to replicate the position of bivalent cations in the spinel structure of magnesia-rich basic refractories using trivalent cation oxides and tetravalent cation oxides along with MgO as the spinel formers in order to achieve complex spinel solid 83 solution with similar calculated inversion parameter. The predicted lattice parameters based on the structure are shown in Table 5.8. Table 5.8. Calculated inversion and lattice parameter of the spinel solid solution in the chrome-free mixes Composition MgO ? 60 wt. % Al2O3 ? 34 wt. % TiO2 ? 2.5 wt. % SnO2 ? 3.5 wt. % Fe2O3 ? 0 wt. % MgO ? 60 wt. % Al2O3 ? 30 wt. % TiO2 ? 2.5 wt. % SnO2 ? 3.5 wt. % Fe2O3 ? 4 wt. % MgO ? 60 wt. % Al2O3 ? 30 wt. % TiO2 ? 2.5 wt. % SnO2 ? 3.5 wt. % Fe2O3 ? 4 wt. % Calculated Inversion (%) 17 24 30 Calculated Lattice parameter (?) 8.144 8.367 8.496 Experimental Lattice parameter (?) 8.100 8.136 8.160 Reitveld WPF range of inversion (%) 9 ? 17 15 ? 21 21 ? 27 These were then correlated with the Whole Pattern Fitting Reitveld Analysis and the experimental values are lower than that for the calculated values, which may be due to the formation of non-stoichiometric magnesia-rich spinel and due to some cations occupying their non-preferred sites in the presence of other metal ions. The inversion in 84 the complex spinel structure is not just dependent on the chemistry of the spinel phase, as the calculated values in Table 5.8 show, but also depends on factors like the time and temperature of thermal treatment, heating and cooling cycles, electric and magnetic fields. 5.3. Effect of Fe2O3 in forming complex spinel in MgO-Al2O3-SnO2-TiO2 mixes The iron oxide addition even at 4 wt. % level resulted in complete spinel formation at 1350 ?C as in Figure 5.10 (red markers for complex spinel solid solution (CSSS), blue markers for periclase). This was attributed to a combination of factors like formation of metastable phases at 1060 ?C in the MgO-SnO2-Fe2O3 system [80] that promoted reaction with TiO2 and Al2O3 in creating substitution defects (Ti4+ for Sn4+ and Al3+ for Fe3+) that result in diffusion through the structure. These phenomena reduced the average diffusion path, unlike the case for formation of MgAl2O4 spinel, which acts as a barrier in further reaction of MgO and Al2O3 [5]. The refinement of the XRD data of samples fired at 1450 ?C (Table 5.9) showed that the increasing iron oxide content results in formation of spinel solid solution of larger lattice parameters than that without iron oxide. This indicated that low temperature reaction of Fe2O3 and SnO2 with MgO in forming spinel during heating activated the formation of spinel solid solution by forming reactive non-stoichiometric defect spinel at lower temperature. 85 Figure 5.10: XRD pattern of magnesia-rich spinel sample containing 4 weight % Fe2O3 fired at 1350 ?C, 3 h Table 5.9. Unit cell dimensions [?] of complex spinel for mixes fired at 1450 ?C Ferric oxide [weight %] Mixes with 2.5 wt. % TiO2 and 3.5 wt. % SnO2 0 8.100 4 8.136 8 8.160 86 Figure 5.11: Effect of iron oxide on the densification of magnesia-alumina-titania-tin oxide mixes (solid symbols denote bulk density; hollow symbols denote open porosity) It was also observed that iron oxide has a marked effect on the densification pattern of MgO-Al2O3-SnO2-TiO2 mix in decreasing the open porosity and increasing the bulk density, as in Figure 5.11, showing that the presence of iron oxide leads to improved reaction sintering as compared to mixes without iron oxide. The addition of iron oxide (8 wt. %) in the mixes resulted in a spinel solid solution with open porosity of 5% in the mixes, showing that it could be used as an additive. It is also expected that the solubility of iron oxide in the spinel solid solution could be reduced by incorporation of iron oxide in the complex spinel. This would increase the corrosion resistance of the spinel phase to fayalite slag, which is rich in oxides of iron, 87 similar to the case of complex spinel in ma gnesia-chrome bricks, as this complex spinel would have decreased solubility in the slag. The use of additives that undergo valence changes under reducing conditions and form intermediate compounds at low temperatures have shown to improve sintering when used in combination [40]. 5.4. Role of nano-powders in complex spinel systems The presence of nano-oxides in the initial mix could result in seeding effect and reactive sintering forming spinel bonds [39]. This could influence the microstructure and result in lesser open porosity, as well as increase the synthesis of spinel solid solution formation, and as a result the sintering temperature could be reduced. For the mixes containing alumina, titanium dioxide, tin dioxide and magnesia, it was seen that replacement of titania sub-micron powder by nano-titania dioxide (with average crystallite size 21 nm, surface area 50 ? 15 m2.g-1; source Evonik Degussa Corporation) resulted in improved densification at 1350 ?C, but the same effect was not observed for samples fired at 1450 ?C. At 1450 ?C, the use of nano-powders did not have any effect on the densification and sub-micron sized oxide powders had the same result as nano-powders. Incorporation of different amounts of iron oxide showed different densification behaviour with nano-titania additions at 1350 ?C as can be observed Figure 5.12 and Figure 5.11. 88 Figure 5.12: Densification pattern of magnesia-alumina-tin oxide-ferric oxide-nano titania mixes sintered for 3 hrs at different temperatures (solid symbols denote bulk density; hollow symbols denote open porosity) Figure 5.13: Densification pattern of magnesia-alumina-titania-tin oxide-ferric oxide-nano titania mixes sintered for 3 hrs at different temperatures (solid symbols denote bulk density; hollow symbols denote open porosity) 89 Figure 5.14: Densification of magnesia-alumina-titania-tin oxide-ferric oxide-nano alumina-nano titania mixes sintered for 3 hrs at different temperatures (solid symbols denote bulk density; hollow symbols denote open porosity) There was negligible difference in the apparent porosity with or without the use of nano-powders for the mixes fired at 1450 ?C containing 8 wt. % Fe2O3 as observed in Figure 5.11, Figure 5.12, Figure 5.13 and Figure 5.14. In the case of incorporation of 4 wt. % iron oxide, both the bulk density and apparent specific gravity was highest for the mix containing both nano-titania and nano-alumina as in Figure 5.14. One of the possible reasons for this is due to lower volume of closed pores. The correlation with the unit cell dimension of spinel, calculated values of which are shown in Table 5.10, as lower lattice parameter for the same ratio of the oxide constituents indicate higher true density of the complex spinel. In the case of the latter, the use of nano-oxides reduced the diffusion path for spinels, possibly resulting in metastable spinel being formed during the initial phase of reaction sintering that stabilize during the latter stage of sintering. 90 Table 5.10. Unit cell dimensions [?] of complex spinel for mixes fired at 1450 ?C Ferric oxide [weight %] Mixes with nano TiO2 completely replacing TiO2 powder Mixes with nano TiO2 partially replacing TiO2 powder Mixes with nano TiO2 and nano Al2O3 as partial replacements 0 8.102 8.104 8.116 4 8.141 8.145 8.141 8 8.157 8.154 8.136 From the densification results, it was observed that trivalent and tetravalent transition metal cations, that occupy the tetrahedral sites, had a marked influence on the densification pattern and improved sintering at relatively low temperature of 1450 ?C. The use of either tin oxide or iron oxide as inverse spinel formers in different mixes individually did not result in major improvement in either spinel formation or densification, but a combination of the two resulted in improved reactivity and densification. This is because their combination results in low temperature defect cubic structure [80] that transformed into a spinel at higher temperature. The reaction or solid solutions formed of tin dioxide and ferric oxide with magnesia that initially promoted the spinel forming reaction or the reduction-oxidation behaviour of both the cations result [35] in complex single phase spinel solid solution formation at low temperature. 91 The single phase of complex spinel solid solution during heating or in the initial stage of sintering stabilised itself during the soaking at maximum temperature. This, as compared to the MgAl2O4-Mg2TiO4 spinel solid solution, showed similar pattern on firing at 1600 to 1800 ?C (Section 5.1) combined with the effect of complex spinel formation. The ascertained ratios of different constituent cations that form stable complex spinel at high temperature were based on calculations as described in Section 5.2. 5.5. Comparison of chrome-free mixes to chrome-based mixes The physical properties shown in Table 5.7 depicted improved sintering of the different mixes from 1350 to 1450 ?C with iron oxide and nano-oxides in the mixes (Section 4.2.3; details on these mixes processing and compositions are presented in Section 4.2). The open porosity for samples containing chromia (Table 5.11), fabricated to understand whether Cr2O3 also exhibited showed a decrease of ~37% (~42% to ~25%) on increasing the firing temperature from 1350 to 1450 ?C. In the case of chrome-free mixes, the decrease in open porosity of ~80% (from ~29% to ~4%) showed that iron oxide results in improved sintering along with increased reactivity as also confirmed through the X-ray diffraction analysis. Less than 5% open pores also indicated that reaction sintering for spinel formation and direct bonding of the spinel to magnesia might have been possible for chromium-free mixes. The highest change in open porosity could be attributed to the presence of both tin oxide and iron oxide in the 92 mix, which has resulted in bulk density being ~90% of the true density and less than 10 % total porosity for the composition. Table 5.11. Physical properties of mixes containing nano-oxides Firing temperature Mixes with chrome (MTSFC8) Mixes without chrome (MTSF8) Bulk density [g.cm-3] Open porosity [%] Total porosity [%] Bulk density [g.cm-3] Open porosity [%] Total porosity [%] 1350 ?C 2.22 42 43 2.47 29 34 1450 ?C 2.79 25 29 3.40 4 8 The addition of nano-alumina (with average crystallite size 13 nm, surface area 100 ? 15 m2.g-1; source Evonik Degussa Corporation) into the mixes containing chromia did not show any increase in the bulk density or decrease in open porosity, showing that it alone did not increase the sinterability of the mixes. Even the densification data for addition of nano-titania did not show any improvement in the sintering of chromia-based mixes at both 1350 and 1450 ?C. It was further seen that nano oxide based compositions with lower open porosity also had a low apparent specific gravity, indicating that the samples had pores that were either closed or too fine for the water to 93 penetrate them. The other reason for low apparent specific gravity is that spinel structure formation resulted in a decrease in the true density of the specimens. 5.6. Summary 1. The content of tetravalent cation oxides for formation of low temperature spinel solid solution phases was analysed and found to be 3.5 wt. % SnO2 and 2.5 wt. % TiO2 for compositions containing 60 wt. % MgO; this composition was subsequently used to fabricate low-temperature complex spinel based binder system. 2. The amount of ferric oxide additive (4 wt. % and 8 wt. %) was determined based on the high temperature inversion ratio in complex spinel systems (Section 2.2) and calculated to replicate the inversion (between 17 and 30 %) in the spinel phase of currently used chromium bearing basic refractories. 3. The nano-oxides did not reduce the sintering temperature. However, the lattice parameter of these complex spinel systems showed similar pattern, which is a decrease in lattice parameter with increasing firing temperature, to that of the metastable phases in MgAl2O4-Mg2TiO4 systems that led to better densification. 94 4. Dense complex spinels in the system MgO-Al2O3-Fe2O3-SnO2-TiO2, with water absorption less than 1%, could be fabricated through simultaneous synthesis and sintering at temperatures below 1500 ?C. The compositional range of the spinels is as follows: (Wt. % MgO: 60, Al2O3: 25-35, Fe2O3: 3-9, SnO2: 3-4, TiO2: 2-3). These spinels could be used for development of both aggregates and binder systems for chromium-free basic refractory ceramics. 95 6. Development of refractory ceramics using in-house processed binder systems In Chapter 5 of the thesis, dense ceramic matrix was formed through simultaneous synthesis and sintering of compositions in the MgO-Al2O3-Fe2O3-SnO2-TiO2 complex spinel systems. It was also observed that nano-powders did influence the structure of these complex spinels thus formed, possibly by creating defect structures that led to change in lattice parameters (Table 5.9), but did not aid in reducing the firing temperature. In this section of the thesis, the dense ceramic systems' microstructure is studied to ascertain (a) if the inhibition of grain growth occurred, and (b) if the magnesia was covered by spinel solid solution. These were targeted in order to have better resistance to hydration and reaction with FeO rich environments. The properties of the refractory bricks and castables containing in-house binder systems with either (i) periclase grains, (ii) sintered or fused magnesia-chrome ore grains, or (iii) aggregates prepared with the same composition as the binder system were compared to regular magnesia-chrome ore based bricks that are used in non-ferrous industry, so that bricks and castables based on one of these systems could be fabricated for plant trial. 6.1. Microstructure of the binder systems The microstructure of the optimized compositions reported in Section 5.1 in the systems MgO-Al2O3-TiO2 and MgO-Al2O3-SnO2 are shown in Figure 6.1 and Figure 6.2 96 respectively. It was observed that dense structure with low porosity was formed in the MgO-Al2O3-TiO2 system, confirming the open porosity data plotted in Figure 5.9. The brighter regions, indicating spinel solid solution (as confirmed through EDX), were around the boundaries of the darker regions, which indicate periclase grains, for both MgO-Al2O3-TiO2 and MgO-Al2O3-SnO2 systems. This indicates that grains were bonded by the formation and precipitation of complex spinel solid solution, similar to that in magnesia-chrome ore based bricks [40]. Figure 6.1: Scanning electron micrograph of MgO-Al2O3-TiO2 fired at 1800 ?C, 3 h The microstructure with TiO2 incorporation (Figure 6.1) showed that the grain growth was inhibited through the formation of spinel solid solution that created a barrier 97 between the adjoining periclase grains, similar to that for alumina rich compositions [38]. However, in the case of SnO2 addition as observed in Figure 6.2, grain growth was observed and the dissolution and re-precipitation of spinel solid solution led to the periclase grains being encapsulated in the secondary spinel phase, similar to reaction bonding in magnesia-chrome ore based refractory ceramics [35]. Both of these phenomena as seen for TiO2 addition (Figure 6.1) and SnO2 addition (Figure 6.2) are required in order to improve the sintering and to improve the resistance of the periclase grains to hydration and reactions with molten metal and slag during use in pyro-metallurgical environments. Figure 6.2: Scanning electron micrograph of MgO-Al2O3-SnO2 fired at 1800 ?C, 3 h 98 Selected compositions in the MgO-Al2O3-Fe2O3-SnO2-TiO2 complex spinel system were further analysed for in-situ reaction that would result in a spinel solid solution with inversion similar to that of magnesia-chrome ore based refractories. The in-situ reaction resulted in a finer matrix [38] and open porosity below 5 volume % at firing temperature of 1350 ?C as shown in Figure 6.3. The complete formation of spinel at 1350 ?C, Section 5.3, and the fine grain size could be attributed to various factors, which led to the hypothesis that a sintering mechanism different from the predominant reaction sintering for formation of spinel was achieved. Figure 6.3: Scanning electron micrograph of MgO-Al2O3-Fe2O3-SnO2-TiO2 fired at 1350 ?C, 3 h 99 The salient differences of this simultaneous synthesis and sintering process from reaction sintering were as follows: (a) The formation of metastable phases, for example, perovskite at 1060 ?C [80] and complex spinel with interstitial defects, as shown in Figure 5.4 and Figure 5.6, below the firing temperature, further leading to the development of desired phase during densification (b) The reaction of metastable phases, to form the desired spinel phase promoted the sintering, as the shrinkage from metastable phases [80] was compensated by volume expansion due to the formation of spinel that resulted in (a) reduction in tensile stresses that lead to crack formation, and (b) inhibition of grain coarsening, both of which were observed during reaction sintering of MgO-Al2O3 spinel [38]. 6.2. Fabrication of refractory ceramics by pressing with complex spinel binder The specimens were pressed (refer to Section 4.4 for materials and processing details) using periclase (P) and sintered magnesia-chrome ore (C1) grains as aggregate and binder systems MTSF4 and MTSF8 (Section 4.2.3). The bulk density of periclase being lower than mag-chrome spinel, accounted for lower bulk densities of P-MTSF4 and P-MTSF8 as compared to C1-MTSF4 and C1-MTSF8 in Table 6.1. The marked 100 increase in the density and decrease in open porosity, with the binding system MTSF8 as compared to binding system MTSF4 indicated that the ratio of bivalent ions in the tetrahedral and octahedral sites of the spinel lattice affect the physical properties of the system. Table 6.1. Physical properties of the fired samples Material Bulk density (g.cm-3) Open porosity (%) Apparent specific gravity (g.cm-3) Flexural strength (MPa) P-MTSF8 (P1) 3.24 14.6 3.79 6.8 P-MTSF4 (P2) 3.04 17.5 3.68 4.4 C1-MTSF8 (C1) 3.73 17.9 4.56 5.7 C1-MTSF4 (C2) 3.49 21.1 4.42 3.2 The decrease in the apparent specific gravity indicated the presence of closed pores in mixes with MTSF4 as compared to samples fired with the binder MTSF8. The larger difference in physical properties, with magnesia-chrome ore based aggregates, which contain preformed spinel, showed that these reactions occur predominantly on the surface of magnesia, thereby indicating that the diffusion through an interfacial layer retards the spinel formation. The flexural strength of the fired samples indicated higher 101 strength for the mixes with fused magnesia (P type) aggregate. This confirmed that the simultaneous synthesis and sintering promoted better bonding. In the case of C type aggregate, the seeding effect of spinel might be attributed to better bonding. Only the P-MTSF8 and C1-MTSF8 samples were of comparable flexural strength to the commercially available mag-chrome bricks (6 to 15 MPa). The effect of simultaneous synthesis and sintering, through formation of metastable phases that resulted in reaction-bonding, was similar to the direct-bonded basic bricks rich in magnesia. 6.3. Hydration of aggregates The surface cracking in some pressed ceramics and the need to develop castables for making different wedge shapes led to hydration studies of the grains and magnesia fines. The mag-chrome (C2) grains and fines showed considerable hydration within 24 h period (Table 6.2) indicating that they would create brucite, with the associated volume expansion and tensile stresses after casting or pressing, which could crack the shapes during drying. In comparison, the C1 mag-chrome grains and fines showed lesser hydration within 24 h period as compared to the C2, but they would also result in volume expansion and tensile stresses after casting or pressing. The fused mag-chrome grains C3, as compared to C1 and C2 grains that were sintered, showed no hydration in 24 h showing 102 that they could be used to make pressed and cast ceramics without having any volume expansion that could develop tensile stresses or cracking in the formed shapes. Table 6.2. Hydration results of mag-chrome aggregates Material % Drying Weight Loss % Weight Loss on Calcination % Weight Gain by Hydration C1 3-6 mm 0.01 0.07 0.09 1-3 mm 0.01 0.05 0.15 0-1 mm 0.02 0.07 0.34 -180# 0.06 0.19 1.66 C2 3-6 mm 0.06 0.08 0.09 1-3 mm 0.11 0.26 0.10 0-1 mm 0.12 0.38 0.09 -180# 0.13 0.43 0.76 C3 0.01 0.01 0.00 103 In order to test the hydration resistance of the in-house prepared chromium free dense ceramic and its potential use as aggregates, when prepared in bulk, pressed cylinders (? 50 mm, h 45 mm) were fired, crushed, and screened to have all size fractions as required for pressing or casting. These crushed grains were also tested for hydration resistance and found to be equivalent (Table 6.3) to fused mag-chrome grains (C3). This also indicated that the free magnesia in the system had precipitated spinel, which created a better hydration resistance. This was in accord with the castable microstructure in Section 6.4 (Figure 6.5 and Figure 6.6), showing that the binder system forms a precipitation layer even on fused mag-chrome grains. Table 6.3. Hydration results chrome-free aggregates Material % Drying Weight Loss % Weight Loss on Calcination % Weight Gain by Hydration 3-6 mm 0.00 0.00 0.00 1-3 mm 0.00 0.01 0.00 0-1 mm 0.00 0.01 0.00 -180# 0.00 0.03 0.01 104 6.4. Development of vibratory castables with complex spinel binder systems Three types of magnesia powders, (a) light-burnt, (b) dead-burnt and (c) sintered, were added to the mixes with magnesia-rich aggregate so that the spinel formers could react both with magnesia in the magnesia-rich aggregate and the magnesia powder to create reaction bonding during the firing process (refer to Appendix C for characteristics of magnesia powders). The use of hydrated grains was envisaged for better reaction during the firing process. The particle size distributions used for all mixes were based on a previous research [78] as shown in Figure 6.4. Figure 6.4: Particle size distribution of the castable with modified Andreassen coefficient of ~ 0.23 [78] 105 The flow (%), working time (minutes) and setting time (minutes), with different grades of magnesia and hydrated aggregates, measured as per ASTM standard method described in Section 4.5, are presented in Table 6.4. This showed that the use of hydrated magnesia-rich aggregates led to poor flow (a minimum of 20 % flow [78] is desired for vibration casting) and very low working time. A minimum working time of 30 minutes was required to cast a 230 mm ? 115 mm ? 90 mm sample, in order to make a cast sample for trial in the industrial conditions. Table 6.4. Flow and setting time of castables with different types of magnesia Type of magnesia Flow (%) HT (minutes) VT (minutes) ST (minutes) Sintered 7 5 7 8 Dead-burnt - - <4 4 Light-burnt - - - <4 The hydrated grains resulted in increased alkalinity and fast setting, thus not having sufficient time for casting. Further, the sintered magnesia having the best results showed that dead-burnt and light-burnt magnesia were prone to hydration and found unsuitable for castables, even though they were more surface-reactive to form better 106 bonds during the firing process. The flow, working time and setting time with different additives are presented in Table 6.5. This showed that the use of hydrated magnesia-rich aggregates led to poor flow and low working time. Table 6.5. Flow and setting time of castables with different additives Additive Wt. % addition Flow HT VT ST Sodium poly-methyl acrylate (SPMA) 0.05 - - - <4 SPMA Citric Acid 0.05 0.02 2 4 4 5 SPMA Boric Acid 0.05 0.02 - - - <4 SPMA Oxalic Acid 0.05 0.02 5 4 4 5 Sucrose 0.05 - - - <4 The hydrated grains resulted in alkalinity of water, fast hydration and do not give sufficient time for casting. The (a) pH based setting retarders like SPMA with or without organic acids, which neutralize the alkalinity of castables containing MgO, due to formation of Mg(OH)2 and chelates, and (b) pH neutral organic solute like sucrose that 107 reduces the ionization of water and forms chelates with Mg(OH)2 did not show any marked improvement with hydrated aggregates and the mixes were almost set in mixer. Table 6.6. Flow and setting time of castables with different dispersants Dispersant (%) HA-1 HA-2 HB-2 HB-3 ADW1 0.5 0.75 - - ADS3 0.5 0.75 - - FS10 - - 0.2 - FS60 - - - 0.2 Results Flow (%) - 3 13 15 HT (mins) - 4 6 7 VT (mins) - 4 7 10 ST (mins) <4 5 10 11 The flow, working time and setting time with different dispersants used for MA spinel based castables are presented in Table 6.6. The long chain organo-metallic compounds 108 (brand names: ADW1 and ADS3, procured from: Almatis Inc., USA), used industrially for MA spinel based castables, did not result in any increase in setting time and the mixes almost set in the mixer. The polymerised long chain carboxylate molecule based dispersants (brand name: Castament?, type: FS10, FW10 and FS60, procured from: BASF, Germany) that aid in dispersion and retard the setting had been tried unsuccessfully, but they had created a small increase in working time. Table 6.7. Flow and setting time of castables with different dispersants Dispersant (%) NA-1 NA-2 ADW1 0.50 0.75 ADS3 0.50 0.25 Results Flow (%) 3 6 HT (mins) 8 15 VT (mins) 10 20 ST (mins) <15 25 109 The hydrated aggregates resulted in coalescence of the fines, as could be observed from flow of less than 10 %, in spinel structure forming binder system and thus decreased the setting time, making them unsuitable for further characterization. In the case of using non-hydrated aggregates, the flow and setting time (Table 6.7) showed that varying the content of ADW1 and ADS3 long chain organic carboxylate additives could increase the working and setting time. Table 6.8. Flow and setting time of castables with different dispersants [UNITECR 2011, Ref. 81] Dispersant (%) NB-1 NB-2 NB-3 NB-4 FS-10 0.1 0.2 - 0.1 FW-10 - - - 0.1 FS-60 - - 0.2 - Results Flow (%) 14 20 17 8 HT (mins) 30 40 45 25 VT (mins) 35 45 50 30 ST (mins) 50 60 65 40 110 The results with polycarboxylate based organic dispersants (Table 6.8; Table 6.9) showed that they can be used in basic castables to cast shapes for further characterization. The magnesia-rich aggregates and binder systems resulted in just sufficient working time for casting shapes and a flow level of ~20% showed that the water content and particle size distribution were already optimized in a previous research [78]. Table 6.9. Physical properties of castables after drying and firing [UNITECR 2011, Ref. 81] Type Temperature (?C) Bulk Density (g.cm-3) Open Porosity (%) NB-2 75 3.81 14 816 3.75 19 1400 3.63 14 NB-3 75 3.64 15 816 3.69 18 1400 3.60 14 111 On the basis of the workability, compositions NB-2 and NB-3 were cast in bars (7??1??1?) and cubes (7??1??1?) to test the properties of cured castables after drying at 75?C for 16 h and firing at 816 ?C and 1400 ?C. The results of physical characterization (Table 6.9) showed that the NB-2 has a decreasing bulk density with increase in firing temperature. The decrease in both BD and AP with increase in firing temperature from 816 ?C to 1400 ?C indicated the formation of spinel as the specific gravity of spinel is less than a mechanical mixture of its constituent oxides. This is in accord with our previous results showing that the binder system could be fired at 1400 ?C to form a dense matrix. Table 6.10. Mechanical properties of castables after drying and firing [UNITECR 2011, Ref. 81] Type Temperature (?C) MOR (MPa) CCS (MPa) NB-2 75 5.9 11.8 816 3.0 3.6 1400 21.4 29.3 NB-3 75 6.4 13.1 816 3.2 3.3 1400 13.8 21.2 112 The mechanical properties of the castables were measured in terms using three-point bending test on bars (7??1??1?) for modulus of rupture and compressive test on cubes (1??1??1?). The castables (Table 6.10) NB-2 had higher green strength and the decreasing density with increase in firing temperature led to controlled volume expansion and thus better fired strength than NB-3. The decrease in both MOR and CCS on firing at 816 ?C indicated the removal of binder and hydraulic bonds leading to lower strength and the firing at 1400 ?C led to the formation of complex spinel that bonds to the grains and improved mechanical properties for both mixes. Figure 6.5: Scanning electron micrograph (magnification ? 50) of NB-2 fired at 1450 ?C for 5 h with elemental analysis of aggregate surface [UNITECR 2011, Ref. 81] 113 The microstructure (Figure 6.5) of the castables NB-2 using fused magnesia-chrome ore aggregates (C3) with the in-house developed novel binder system showed that the mag-chrome grains were surrounded by binding system. It was also observed that the binding system was dense with closed pores around it. The elemental distribution identified through EDS on the fused aggregate showed titanium, which was expected, along with tin, only when cations from the binder system would diffuse in the fused aggregate and result in bonding due to dissolution and re-precipitation. Figure 6.6: Scanning electron micrograph (magnification ? 1000) of NB-2 fired at 1450 ?C for 5 h [UNITECR 2011, Ref.81] The micrograph with magnification ? 1000 (Figure 6.6) indicates a differentiation between the binder phase (lighter shade) to the aggregate (darker), the pores being the 114 darkest. The precipitation of the binder phase was identified on the mag-chrome aggregates and also between the aggregates, thus supporting the EDS results on the aggregate and confirming that the binder system creates precipitation of complex spinel phase on the fused aggregate after firing at 1450 ?C. 6.5. Summary 1. The water absorption was less than 1% for magnesia-rich complex spinel forming compositions, fired at 1450 ?C, indicating that it was similar to fused magnesia-chrome ore based aggregates. 2. The in-house fabricated (by pressing) and crushed dense ceramics, used to make different size fractions of aggregates, showed excellent resistance to hydration, similar to the fused magnesia-chrome aggregate (C3), indicating the superiority of developed system to available sintered basic bricks and aggregates. 3. The bricks, fabricated through hydraulic pressing, using the developed binding system with three types of aggregates (sintered periclase, fused mag-chrome, and sintered in-house developed chrome-free types) had physical properties comparable to the basic bricks that are used as convertor linings. 115 4. Different organic additives for the binder system were studied in order to have flow and working time suitable for vibration casting, which could be used for making castables with comparable properties to the standard magnesia-chrome bricks. This was then used to scale-up in order to fabricate a brick shape that could be installed in the Bottom Blown Oxygen Converter (BBOC) of Teck Metals Ltd., in order to analyse its potential life in service and usability in industrial environment. 116 7. Characterization of trial ceramics containing the new developed chromium-free binder system [*] 7.1. Properties of cast bricks for the plant trial Based on the results of physical characterization (Section 6.4) and scaling up of select composition to make a standard brick size, the castables were installed in a bottom blown oxygen convertor (BBOC) with service life of 40-45 days in the environment as described in Section 4.6 at Teck Metals Ltd., Trail Operations, in Trail, BC used for cupellation of silver. The bricks were used in two campaigns, first of which was from mid-May to 28 June 2012 and the second from early September to October 11, 2012. The bricks for post mortem characterisation were made to specified dimensions [115 x 230 x (89 - 65) mm], fired at 1450 ?C and characterized. The properties of the trial bricks, average values and range as shown in Table 7.1, were comparable to regular rebonded-fused grain (RFG) magnesia-chrome brick (BCF-F15, by RHI International) with bulk density 3.28 g.cm-3, apparent porosity 15.5% and CCS 50 MPa. Table 7.1. Properties of cast half wedge bricks, fired at 1450 ?C, 5 hours BD (g.cm-3) AP (%) WA (%) ASG (g.cm-3) MOR (MPa) CCS (MPa) 3.25 ? 0.05 14.9 ? 1.1 4.6 ? 0.4 3.82 ? 0.03 8.8 ? 2.0 43.1 ? 8.5 [*Parts of this chapter are quoted from a private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] 117 7.2. Visual observation and volume loss for the trial ceramics Visual observations of the spent bricks have comparable and contrastable wear patterns, suggesting that some of the corrosion mechanisms could be very similar for the industrial brick and our trial brick. The morphology of both the bricks at hot face also indicated the same. Figure 7.1: Cross-section of UT1 brick, showing slag penetration and distribution of pores and cracks [private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] Corrosion occurred through dissolution both of the matrix and of magnesia-chrome grains, into molten slag. The hot faces of the bricks were very coarse and consisted 118 mainly of slag with silver inclusions, as seen in the optical microscopy overview macro-graphs in Figure 7.1 and Figure 7.2 (the microstructural details of these bricks are illustrated in SEM micrographs later in this chapter). Figure 7.2: Cross-section of F15 brick, showing slag penetration and distribution of pores and cracks [private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] In some areas, intact magnesia-chrome grains were not observed, so it is concluded that they were dissolved in the slag. However, the shape of used bricks did not indicate a high degree of corrosion, but rather a combination of thermal spalling and corrosion leading to cracking and fall out of the brick. 119 The dimensions of UT1 half-brick (made by casting with the procured fused magnesia-chrome aggregate and the developed novel chromium-free binder system), as received after use in BBOC, was 115 x 154 x (89-76) mm. As the initial dimensions of this half-brick were 115 x 230 x (89-65) mm; hence ~33% of the length was lost during the trial. The same measurements on the currently used mag-chrome brick (F15) showed 72% remaining length; the 28% loss in length was only slightly lower than for UT. This was a level of normal corrosion over a campaign, for both bricks, as the losses were comparable to other campaigns. The volume loss for F15 was 21.8 % and 26.5 % for UT1; this could be due to a combination of positioning in the convertor, thermal spalling and dissolution, rather than to dissolution alone. During the second campaign with six half bricks for trial, the losses were similar at 37 % indicating comparable performance. Slag penetration in UT1 brick was limited to about 1 mm (reaction layer). However, at about 38 mm from the hot face, there were a few cracks parallel to the hot face and impregnated with the slag. The presence of these parallel cracks was indicative of spalling due to microstructural changes by porosity reduction or reaction with solid and liquids. In contrast, the cracks at the cold face were mostly perpendicular to the hot face. Cross-section showed that pores were evenly distributed, with 1-3 mm diameter, due mainly to the casting process used rather than the standard pressing technique for making magnesia-chrome bricks. However, the pores at the cold face were filled with metals like silver and lead, while at the hot face slag remained mostly in the cracks. 120 In contrast, thickness of the reaction layer in F15 brick was about 3-4 mm, which indicated a higher degree of corrosion than in UT1. The slag could be found even past the center of the brick, at 90 mm from hot face. Similar to UT1, the cracks at the hot face were parallel to it, even at 30 mm into the spent brick, and they were perpendicular to the hot face at the cold face. 7.3. Comparative variation of physical properties of the ceramics Water absorption test was performed in order to determine bulk density, apparent porosity and water absorption at different distances from hot face through sampling as described in Section 4.6; an overall average was calculated for each brick type. For UT1 brick, apparent porosity, water absorption (Figure 7.3) and bulk density (Figure 7.4) remained relatively constant with distance from the hot face and apparent porosity was determined to be about 5 % throughout the brick?s thickness, which was observed to be similar to another campaign consisting of 3 bricks with a range of 4-7% apparent porosity . 121 Figure 7.3: Distribution of water absorption and apparent porosity along the bricks perpendicular to the hot face [private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] 122 Figure 7.4: Distribution of bulk density (g.cm-3) along the bricks perpendicular to the hot face [private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] As observed in SEM images (Section 7.4), molten slag, including Ag, penetrated through the full thickness of brick, most likely along the grain boundaries. This caused relatively low apparent porosity and water absorption (1.2 %) of the spent bricks. The bulk density showed slight decrease when moving from the hot face towards the cold face, implying that slag penetration was reduced towards the cold face. Apparent 123 porosity of spent F15 brick also did not change considerably from hot face to cold face, and an average value was 1.1 %, about 5 times lower than that for UT1. Both UT1 and F15 have bulk densities higher than of unused bricks, suggesting penetration of heavy metals such as lead and silver into the pores. However, pores in UT1 were relatively larger than pores in F15 structure (i.e. 20-500 ?m in UT1 vs 20-200 ?m in F15), while both had similar apparent porosity of 12-16 vol % in unused condition. Comparing the average bulk densities of UT1 and F15, and considering bulk density of pre-trial bricks, as well as their pore size distribution, slag impregnation was observed to be more severe in F15 than in UT1. This showed that chromium-free binder system provided a better bonding with firing at 1450 ?C as compared to the industrial magnesia-chrome bricks, fired in excess of 1730 ?C. 7.4. Comparison of the microstructures of the ceramics Composite images of the cross-sections containing the hot face are presented in Figure 7.5 at low magnification, to observe different penetration profiles into both types of bricks. 124 Figure 7.5: Composite image of the analysed UT1 and F15 bricks in the slag zone Elemental distribution of the main constituents of solid deposits in the cross-sections of the bricks, between the hot and cold faces is presented in Figure 7.6. The EDS analyses were performed on three positions (edges and middle of the sample) at every 10 mm, at 50x magnification, on the whole area, and average values were used for variation plots. Typical variation of these measurements is within 0.2 - 0.8 wt. %. 125 Figure 7.6: Distribution of Ag, Sb, Pb and Bi from hot face to cold face for the used UT1 and F15 bricks 126 Elemental copper was determined at low concentration of 0.5% in the solidified slag. Silver penetrated the whole thickness of both bricks and it was present as inclusions of various dimensions, as large as ~1 mm2 in the hot region, but generally below 0.4 mm2 (mostly below 0.2 mm2) through the thickness of the bricks. Ag concentration was higher towards cold face (2-4 wt. % at 90-150 mm in UT1 and 3-5 wt. % at 50-150 mm in F15). Average values along the whole thickness were 1.9 wt. % for UT1 and higher, 3.2 wt. % for F15. Sb content was zero at 10 mm from hot face for UT1 and 30 mm for F15. Pb penetrated both bricks throughout their thickness, and it was calculated at average values of 5.6 wt. % for both. Bi was also found in significant amounts in both bricks through the whole thickness; average value was 9.6 wt. % in UT1 and slightly higher, 10.7 wt. % in F15. Micrographs of the cracks at the hot face are presented in Figure 7.7. There are only a few cracks in UT1, parallel to the hot face and maximum 2 ?m in width. Figure 7.7: Micrographs showing cracks from hot face of both the used bricks 127 The hot-face region of F15 had more cracks, most of them parallel to the hot face, but also few were perpendicular to it. These cracks were also much wider than in UT1, at maximum 10 ?m, but most of them were 3-4 ?m. The backscattered (BSE) images of the cross-sections of the UT1 at different distances from HF (Figure 7.8) showed that UT1 brick was fully penetrated. Between 100 mm and the CF (150 mm), penetrating pattern was very similar to the pattern seen at 90 mm. Impregnating material was mainly composed of antimony (Sb), lead (Pb), bismuth (Bi) and silver (Ag), with small amounts of Cu (~0.5 wt. %). Two distinct areas could be identified in the microstructure: a ?slag zone?, with reactions between the refractory and slag at the slag/refractory interface, and the rest of the brick, called ?penetrating zone?. It was observed that the pores and some of the grain boundaries were the main path for the penetration of the corrosive material into the brick. The variation in the concentration of the elements forming the slag revealed that silver was the only element that did not react with the magnesia-chrome grains. In contrast, the other elements in the slag diffused into the magnesia-chrome grains and dissolved the grains of the brick, thus increasing the porosity and further slag penetration. According to spot EDS analyses of transient layer, Sb first reacted with Ca, which then dissolved magnesia and chromia. As a consequence, magnesia-chrome grains would be ultimately dissolved into slag or would just be dislodged into slag, leaving the brick structure. 128 Figure 7.8: Microstructure of used UT1 brick from the hot face to 90 mm towards the cold face (MCr is fused magnesia-chrome aggregate) A similar mechanism of dissolution was observed for magnesia grains. The EDS analyses revealed that the Sb-rich part of the slag was also rich in Ca, suggesting a 129 reaction between SbO2 and CaO. Since Ca was mostly accumulated in magnesia grain boundaries, slag penetrated into the grains through grain boundaries. Magnesia grains were first separated from their grain boundaries, transferred to slag and then dissolved in it. Presence of slag at the grain boundaries inside magnesia-chrome grains supported the high penetration of Pb and Bi into magnesia-chrome grains. It appeared that Bi and Pb could be dissolved in periclase grains. The other important feature was limited interaction between precipitated complex spinels inside magnesia-chrome grains structure and slag, suggesting that complex spinel structures are more corrosion resistant against Pb and Bi-rich slags. The limited interaction indicates the stability of a particular degree of inversion [35, 36] in complex spinel structure that retards further reaction. It is hypothesized that the slag components including Bi and Pb were dissolved selectively in magnesia and damaged the grain structure. Micrographs of cross-section through used commercial magnesia-chrome brick (F15), at different distances from the HF are shown in Figure 7.9. Similar to UT1, F15 was also totally impregnated, with a similar distribution from 0 mm of the hot face to the cold face. While grain boundaries were also affected by the slag, the grains were corroded more severely than in UT1. The distribution of slag elements (Figure 7.6) along penetration direction showed significant differences in corrosion patterns of the two bricks: for instance, no transient layers were observed at slag/brick interface of F15, suggesting dissolution of brick constituents into slag. For UT1, formation of a transient layer suggested that slag components diffused into the grains (magnesia-chrome and 130 magnesia), while microstructure of used F15 shows that grains strongly reacted with slag, and dissolved into it. Figure 7.9: Microstructure of used F15 brick from the hot face to 90 mm towards the cold face (MCr is fused magnesia-chrome aggregate) 131 The EDS analyses indicated that brick constituents were selectively dissolved into slag; Ca, Al, and Fe found in the Sb-rich portion of slag implied the dissolution of spinels into slag. Ag did not react with any refractory materials and remained as a metal in microstructure. Bi and Pb formed a solid-solution with each other; that was not a reactive phase and seemed that reacted only with small amounts of Ca present. As a result of suggested interactions, Mg-rich regions and Sb and Ca-rich regions were formed in spent F15?s microstructure. 7.5. Mineralogical composition of the ceramics by X-ray diffractometry The XRD results for UT1 showed Pb and Bi penetrating through the brick?s thickness, Figure 7.10. Lead and bismuth were identified as Bi24Pb2O40 (B) at the hot face and cold face. Bismuth was also found as Bi2O3 (A), and lead was also found as lead silicate at hot face, but no lead silicate was identified at cold face. Brick components were identified as periclase (MgO) (C) and a complex spinel, magnesiochromite (MC) (Mg,Fe)(Cr,Al)2O4. For F15 brick, Fig. 7.5.2, lead was identified as Pb2Cr0.5Sb1.5O6.5 (Y) throughout the thickness of brick, implying dissolution of the brick into molten slag. Bismuth was identified as Bi2O3 and Bi24Pb2O40 through the thickness of brick. 132 Figure 7.10: X-ray diffraction patterns of used UT1 brick at hot and cold faces [private communication by H. Zargar,C. Oprea, and G. Oprea in September 2012, Ref. 82] Comparison between brick components in UT1 (Figure 7.10) and F15 (Figure 7.11) showed that concentration of (Mg,Fe2+)(Cr,Al,Fe3+)2O4 at hot face was higher than in F15, indicating low corrosion of mag-chrome grains in UT1. It showed that the most prevalent F15 brick component at hot face was MgO, verifying the corrosion mechanism was that grains strongly reacted with slag, and dissolved into it. 133 Figure 7.11: X-ray diffraction patterns of the used F15 brick at the hot and cold faces [private communication by H. Zargar,C. Oprea, and G. Oprea in September 2012, Ref. 82] 7.6. Comparative variation of mechanical properties of the ceramics Overall moduli of rupture (Figure 7.12) for UT1 brick were relatively constant through the thickness of the brick, with an average of 29.5 MPa. The highest modulus of rupture for the brick was 34.0 MPa at hot face. For F15 brick, overall modulus of rupture was 39.8 MPa and declined from hot to cold face; this was due to decrease in apparent 134 porosity from hot face. Similar results were obtained in the second trial with 3 UT bricks, indicating that the variation in the results were ~ 5%. Figure 7.12: Variation of modulus of rupture from hot to cold face of used UT1 and F15 bricks [private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] 135 For UT1, compressive strength (Figure 7.13) was 83 MPa at hot face and decreased to 50 MPa towards cold face. The trend for F15 brick was also decreasing values towards cold face, from 87 MPa at hot face to 70 MPa at 135 mm from hot face. The mechanical properties correlated well with the physical properties and microstructural analyses results. Figure 7.13: Variation of cold crushing strength across the brick, perpendicular to the hot face of the used UT1 and F15 bricks [private communication by H. Zargar, C. Oprea, and G. Oprea in September 2012, Ref. 82] 136 While Figure 7.12 and Figure 7.13 plot the average results of the MOR and CCS tests, the range of error due to instrumental errors and sampling for the tests were ? 2 MPa, which was also verified with deviation from the average values for the tests. 7.7. Summary The cast trial brick UT1 outperformed the standard F15 brick, in regard to the issues important for life in service in the convertor. Namely, we have observed for UT1 less reacted brick material by the slag, smaller cracks at the hot face indicating that there was lower thermal spalling, less grain boundary penetration by slag, lower impregnation by the slag as indicated by high porosity after impregnation as compared to F15 and verified by lower MOR and CCS values. The volume of the brick lost by corrosion, thermal spalling or other factors was lower (21.8%) for F15 brick in Trial 1 as compared with UT1 (26.5%) and comparable during Trial 2 at 37 % despite the fact that the other properties proved that UT1 had better after-use properties than F15. That could have been due to the initial better macrostructure generated by dry pressing of F15 as compared to casting for UT1, which generally assumes larger pores and weaker texture, despite better behavior in use due to better microstructure and mineralogical composition. 137 Larger amounts of slag impregnating the F15, compared with UT1, demonstrated higher vulnerability of this brick, due to reactions with impregnated slag, mostly Bi2O3, Sb2O3 and PbO/PbO2, when convertor temperature exceeded the regular process temperature of 1250 ?C. In regard to the micro-structural changes due to interactions between the brick phases and penetrated slag, the UT1 showed a transient layer at slag-refractory grain interfaces, as opposed to F15, that showed strong reaction between grains and slag, which would speed up the refractory corrosion or even cause a total failure. UT1 had a slightly higher initial porosity and very different ceramic matrix than F15, resulting in superiority of UT1 versus F15 in regard to cracking due to thermal shock. The cracks parallel to hot face were smaller in width and fewer for UT1 than F15 and those on and perpendicular to hot face were practically inexistent for UT1, demonstrating a superior microstructure for this environment, having a lot of thermal cycling due to process conditions, or thermal shocks due to un-predicted operational reasons. Comparative analyses showed less affected physical properties, micro- and macro-structures due to slag penetration and corrosion for UT1 than F15, which indicate that a brick made with chromium-free binding system and fused magnesia-chrome grain fired at 1450 ?C has longer life in service than the classical magnesia-chrome brick, of re-bonded type with fused grains, fired above 1730 ?C. 138 8. Conclusions In this work, the understanding of complex spinel structures and the mechanisms of their synthesis and sintering has been established to achieve chromium-free basic refractory systems, which could be fired at lower temperatures than the currently used chrome-ore refractories. The reduction or absence of chromium-bearing compounds in the refractories for non-ferrous converters would result in safer disposal of the spent refractories. The sintering temperatures decrease from greater than 1730?C (as is the case with direct bonded magnesia-chrome bricks) to less than 1500?C, would result in lower greenhouse gases produced during this energy intensive firing process. The two major findings of this research leading to chromium-free refractory system were as follows: a. The metastable complex spinel solid solution phases having defect structure with higher unit cell volume, as calculated from the Whole Pattern Fit Rietveld refinement of XRD data, could be achieved during thermal treatment in air. Formation of these phases, in the system Mg1+x(Al,Fe)2-2x(Ti,Sn)xO4, compensates for the volume expansion during the complex spinel formation, unlike reaction sintering where volume expansion is not compensated, leading to simultaneous synthesis and sintering process. 139 b. The inversion in chromium free complex spinel could be tailor-made for the usage conditions, in order to achieve the inversion index values of 17 ? 30 % as in the current state-of-art magnesia-chrome ore based bricks. The procedure for achieving the specific inversion index in these complex spinels involved the following steps: (a) fix the MgO content to 60 wt. %; (b) investigate the optimum amounts of SnO2 and TiO2 that would result in lower temperature of densification (b) to calculate the amount of Fe2O3 based optimum amounts of SnO2 (3.5 wt. %) and TiO2 (2.5 wt. %) to get inversion in the range of 17 to 30 %. Through the integration of these two major findings, the hydration resistance of chrome-free systems fired at 1450 ?C was similar to that of the rebounded fused grain magnesia-chrome bricks with firing temperature greater than 1730 ?C. The developed novel complex spinel system was subsequently used as a binder system for basic refractories with chromium-based aggregates and for development of chromium-free aggregates. The current research reached its goal of developing a chromium-free refractory ceramic, particularly addressing the needs of the non-ferrous industry, where magnesia-chrome ore based refractory ceramics are currently in use. The evidence of improved service life during two plant trials has led to patent application by UBC-UILO (#12-013). 140 8.1. Techniques for fabrication of basic chrome-free complex spinel refractory ceramics 1. Optimum content of tetravalent cation oxides (SnO2 and TiO2) for the formation of low temperature spinel phases at 1350 ?C was determined to be 8 mole % and 6 mole% respectively. Based on these data, the novel MgO-Mg1+x(Al,Fe)2-2x(Ti,Sn)xO4 binder system was fabricated and fired at 1450 ?C. This novel binder system had the following characteristics: bulk density 3.40 g.cm-3; water absorption 0.8 vol. %; 92 % relative density. 2. The amount of ferric oxide additive (3 to 9 wt. %) was based on the inversion of the complex spinel phase (17% to 30%) in the currently used chromium bearing basic refractories. The open porosity of these dense ceramic systems in the MgO-Al2O3-Fe2O3-SnO2-TiO2 system was less than 1 vol. % making it similar to the fused magnesia-chrome ore based systems. The dense ceramic compositions were subsequently used for development of both aggregates and binder systems. 3. The in-house fabricated and crushed dense ceramics in the MgO-Al2O3-FeOx-Me4+O2 system that have 17 - 30% inversion (wt. % MgO: 60, Al2O3: 25-35, Fe2O3: 3-9, SnO2: 3-4, TiO2: 2-3), were used to make aggregates in the size range of 0.1 to 6.0 mm. A minimum of three size fractions of aggregates are used for preparing bricks and castables, and were prepared by crushing, grinding and sieving fired cylindrical samples 141 (? 50mm; h 25 mm) to get size fractions of 0.1 to 1.0 mm, 1.0 to 3.0 mm and 3 to 6 mm. These aggregates showed excellent resistance to hydration, i.e. less than 0.1 % weight gain on being submerged in water for 48 h. This level of hydration resistance is similar to hydration resistance of the fused magnesia-chrome aggregates (C3), indicating the superiority of developed system to the currently available sintered basic bricks and aggregates. 4. The castable refractories developed with our novel binder system, using fused mag-chrome type aggregate, had physical properties (such as bulk density 3.25 g.cm-3; open porosity 15 vol. %; water absorption 4.6 vol. %; MOR 8.8 MPa; CCS 43.1 MPa) comparable to the standard bricks fired at 1450 ?C. Consequently, this process was subsequently scaled up to make wedge shapes for installation in the Bottom Blown Oxygen Converter (BBOC) of Teck Metals Ltd. 8.2. In service performance of developed refractory ceramics in the BBOC through comparative analyses after usage 1. The developed castable (UT) had a slightly higher initial porosity (15 vol. %) and different matrix (composed of MgO-Mg1+x(Al,Fe)2-2x(Ti,Sn)xO4) than the regularly used magnesia-chrome brick (F15). The UT castable demonstrated the superiority versus F15, in regard to cracking due to thermal shocks. The cracks parallel to the hot face were significantly smaller in width (i.e. 2 ?m vs. 4 ?m) as observed from the 142 microstructure of the spent refractory at the hot face. This confirmed a superior microstructure of UT and thus its suitability for this environment with significant component of thermal cycling leading to thermal shock. 2. The volume of the bricks lost by corrosion, thermal spalling or other factors like position in the convertor was lower (21.8%) for the F15 brick compared with the UT (26.5%) during the first trial. In the second trial with 3 sets of bricks for UT, the volume losses were similar (37%). 3. The microstructural changes due to interactions between the brick phases and the penetrated slag depicted that UT had a transient layer at the slag-refractory grain interfaces. This transient layer was marked by precipitated spinel and absence of Sb and Bi, which impeded the reaction of the periclase and mag-chrome grain. In contrast, F15 showed strong reaction between grains and molten slag, which would speed up the corrosion or could even cause a total failure of the brick. 4. The physical properties, micro- and macrostructure of the spent refractory ceramics, developed with in-house fabricated novel binder system and fused magnesia-chrome aggregate, indicated that these trial bricks were less affected by the thermal cycling from 900 to 1300 ?C. They were also less affected by the rapidly changing oxidizing and reducing conditions, slag penetration and corrosion, as compared to the currently used magnesia-chrome ore based refractory ceramics. 143 9. Future work 1. It was reported that complete spinel formation was achieved at 1350 ?C without using nano-powders and that the nano-powders affected the lattice parameters of the spinel solid solution. The decrease in lattice parameter for compositions containing iron, fired at 1450 ?C, with increase in content of nano-oxides needs to be investigated. 2. A novel simultaneous synthesis and sintering method was proposed based on the analysis of densification and lattice parameter results in the MgO-Al2O3-FeOx-Me4+O2 system. Investigation with other combinations of cations of different valence states would confirm the validity of this sintering method for other systems. 3. It was found that hydration behaviour of in-house developed sintered basic chrome-free aggregates prepared by simultaneous synthesis and sintering was similar to that of fused magnesia-chrome ore aggregates. Preparation of these aggregates for scaling up and the possibility of making refractory bricks and castables need to be analysed. 4. The vibratory castable refractory based novel binder system outperformed current state-of-art rebonded fused grain magnesia-chrome ore based refractory ceramics during use in the Bottom Blown Oxygen Convertor of Teck Metals, Trail, BC. Industrial trials in such convertors with chrome-free basic refractories could give an insight into the service-life of chrome-free refractories. 144 5. Our trials with cast refractory ceramics had the same service life as the industrial bricks. It seems plausible that reduced greenhouse emissions and energy savings have been achieved. However, life cycle analyses in different convertors are necessary to confirm these reductions as there are various energy resources used even in the raw materials to manufacturing continuing to usage and disposal. 145 References 1. R.D. Maschio, B. Fabbri and C. Fiori, Industrial applications of refractories containing magnesium aluminate spinel, Industrial Ceramics, 8 (2), 1988, pp. 121-126. 2. CAS Registry ID: 1309-48-4. 3. CAS Registry ID: 1344-28-1. 4. CAS Registry ID: 12068-51-8. 5. R. Sarkar, S. K. Das and G. Banerjee, Effect of additives on the densification of reaction sintered and presynthesised spinels, Ceramics International, 29 (1), 2003, pp. 55-59. 6. B. Hallstedt, Thermodynamic Assessment of the System MgO?Al2O3, Journal of the American Ceramic Society, 75 (6), 1992, pp. 1497-1507. 7. R. Sarkar and G. Bannerjee, Effect of addition of TiO2 on reaction sintered MgO-Al2O3 spinels, Journal of the European Ceramic Society, 20(12), 2000, pp. 2133-2141. 8. A. Ghosh, S. K. Das, J. R. Biswas, H. S. Tripathi and G. Banerjee, The effect of ZnO addition on the densification and properties of magnesium aluminate spinel, Ceramics International, 26 (6), 2000, pp. 605-608. 9. C. B. Carter and M. G. Norton, Ceramic Materials: Science and Engineering, Springer New York, 2007, pp. 101-102. 146 10. http://www.emeraldinsight.com/fig/1280510101008.png 11. G. Ottonello, Energetics of multiple oxides with spinel structure, Physics and Chemistry of Minerals, 13 (2), 1986, pp. 79-90. 12. M.A. Petrova, G.A. Mikirticheva, A.S. Novikova and V.F. Popova, Spinel solid solutions in the systems MgAl2O4?ZnAl2O4 and MgAl2O4?Mg2TiO4, Journal of Materials Research, 12 (10), 1997, pp. 2584-2588. 13. I.-H. Jung, S. Decterov and A. D. Pelton, Thermodynamic Modeling of the MgO?Al2O3?CrO?Cr2O3 System, Journal of the American Ceramic Society, 88 (7), 2005, pp. 1921-1928. 14. C. B. Carter and M. G. Norton, Ceramic Materials: Science and Engineering, Springer New York, 2007, pp. 183-184. 15. CAS Registry ID: 7439-95-4. 16. CAS Registry ID: 7429-90-5. 17. CAS Registry ID: 7440-32-6. 18. CAS Registry ID: 7440-47-3. 19. K. Mondal, H. Lorethova, E. Hippo, T. Wiltowski and S.B. Lalvani, Reduction of iron oxide in carbon monoxide atmosphere?reaction controlled kinetics, Fuel Processing Technology, 86 (1), 2004, pp. 33-47. 147 20. G. R. Rossman and M. N. Taran, Spectroscopic standards for four- and fivefold-coordinated Fe2+ in oxygen-based minerals, American Mineralogist, 86 (7-8), 2001, pp. 896-903. 21. R.J. Hill, J.R. Graig and G.V. Gibbs, Systematics of the Spinel Structure Type, Physics and Chemistry of Minerals, 4 (4), 1979, pp. 317-339. 22. L. Ortega-San-Mart?n, A. J. Williams, C. D. Gordon, S. Klemme and J. P. Attfield, Low temperature neutron diffraction study of MgCr2O4 spinel, Journal of Physics: Condensed Matter, 20 (10), 2008, 104238 (5pp). 23. A. F. Holleman, E. Wiber and N, Wiberg (eds.), Inorganic Chemistry, p.1191, Academic Press, California, U.S.A., 2001. 24. C. Hammond, The Basics of Crystallography and Diffraction (3rd ed.), pp.196-198, Oxford University Press, New York, U.S.A., 2009. 25. http://www.icdd.com/ 26. http://www.ccdc.cam.ac.uk/ 27. H. M. Rietveld, A profile refinement method for nuclear and magnetic structures, Journal of Applied Crystallography, 2 (2), 1969, pp. 65-71. 28. T. R. Hinklin and R. M. Laine, Synthesis of metastable phases in the magnesium spinel?alumina system, Chemistry of Materials, 20 (2), 2008, pp. 553-558. 148 29. K. Ramesha, L. Sebastian, B. Eichhorn, and J. Gopalakrishnan, Perovskite and pyrochlore modifications of Pb2MnReO6:? synthesis, structure, and electronic properties, Chemistry of Materials, 15 (3), 2003, pp. 668-674. 30. R. Sarkar, H. S. Tripathi and A. Ghosh, Reaction sintering of different spinel compositions in the presence of Y2O3, Materials Letters, 58 (16), 2004, pp. 2186-2191. 31. G. Bhattacharya, S. Zhang, M. E. Smith, D. D. Jayaseelan and W. E. Lee, Mineralizing magnesium aluminate spinel formation with B2O3, Journal of the American Ceramic Society, 89 (10), 2006, 3034-3042. 32. O. Qu?nard, Ch. Laurent, A. Peigney and A. Rousset, Zirconia-spinel composites. Part I: synthesis of powders and dense materials, Materials Research Bulletin, 35 (12), 2000, pp. 1967-1977. 33. O. Qu?nard, Ch. Laurent, A. Peigney and A. Rousset, Zirconia-spinel composites. Part II: mechanical properties, Materials Research Bulletin, 35 (12), 2000, pp. 1979-1987. 34. R. Lodha, A. Ghosh, B. Mukherjee and G. N. Agrawal, Zirconia?magnesium aluminate spinel composite, American Ceramic Society Bulletin, 85 (6) ,2006, pp. 9201-9203. 149 35. G.B. Andreozzi, S. Lucchesi, H. Skogby and A.D. Giusta, Compositional dependence of cation distribution in some synthetic (Mg,Zn)(Al,Fe3+)2O4 spinels, European Journal of Mineralogy, 13 (2), 2001, pp. 391-402. 36. B. Lavina, G. Salviulo and A.D. Giusta, Cation distribution and structure modelling of spinel solid solutions, Physics and Chemistry of Minerals, 29 (1), 2002, pp. 10-18. 37. A. S. Bhalla, R. Guo and R. Roy, The perovskite structure - a review of its role in ceramic science and technology, Materials Research Innovations, 4 (1), 2000, pp. 3-26. 38. C. Aksel, P. D. Warren and F. L. Riley, Magnesia-spinel microcomposites, Journal of the European Ceramic Society, 24 (10-11), 2004, pp. 3119-3128. 39. D. Mohapatra and D. Sarkar, Preparation of MgO-MgAl2O4 composite for refractory application, Journal of Materials Processing Technology, 189 (1-3), 2007, pp. 279-283. 40. K. Gotod and W. E. Lee, The \"direct bond\" in magnesia chromite and magnesia spinel refractories, Journal of the American Ceramic Society, 78 (7), 1995, pp. 1753-1760. 41. M. A. L. Braulio, L. R. M. Bittencourt and V .C. Pandolfelli, Magnesia grain size effect on in situ spinel refractory castables, Journal of the European Ceramic Society, 28 (15), 2008, pp. 2845-2852. 150 42. J. Parmentier, M. Plouet and S. Vilminot, Influence of the sol-gel synthesis on the formation of spinel MgAl2O4, Materials Research Bulletin, 33 (11), 1998, pp. 1717-1724. 43. I. Ganesh, B. Srinivas, R. Johnson, B.P. Saha and Y.R. Mahajan, Effect of fuel type on morphology and reactivity of combustion synthesized MgAl2O4 powders, British Ceramic Transactions, 101 (6), 2002, pp. 247-256. 44. I. Ganesh, R. Johnson, G. V. N. Rao, Y. R. Mahajan, S. S. Madavendra, B. M. Reddy, Microwave-assisted combustion synthesis of nanocrystalline MgAl2O4 spinel powder, Ceramics International, 31 (1), 2005, pp. 67-74. 45. R. J. Bratton, Coprecipitates yielding MgAl2O4 spinel powders, American Ceramic Society Bulletin, 48 (8), 1969, pp. 759-762. 46. G. Gusmano, P. Nunziante, E. Traversa and G. Chiozzni, The mechanism of MgAl2O4 spinel formation from the thermal decomposition of coprecipitated hydroxides, Journal of the European Ceramic Society, 7 (1), 1991, pp. 31-39. 47. J. K. Popovic, N. Miljevic and S. Zec, Spinel formation from coprecipitated gel, Ceramics International, 17 (1), 1991, pp. 49-52. 48. D.J. Janackovic, V. Jokanovic, I. Petrovic-Prelevic and D. Uskokovic, Synthesis of spinel powders by the spray pyrolysis method, Euro Ceramics V (Part-1), 1997, pp. 197-200. 151 49. Y. Suyama and A. Kato, Characterization and sintering of Mg?Al spinel prepared by spray pyrolysis technique, Ceramics International, 8 (1), 1982, pp. 17-21. 50. G. Lallemand, S. Fayeulle and D. Treheux, Fabrication process of spinel powder for plasma spraying, Journal of the European Ceramic Society, 18 (14), 1998, pp. 2095-2100. 51. D. Domanski, G. Urretavizcaya, F. J. Castro, and F. C. Gennari, Mechanochemical synthesis of magnesium aluminate spinel powder at room temperature, Journal of the American Ceramic Society, 87 (11), 2004, pp. 2020-2024. 52. G. Ye, G. Oprea and T. Troczyski, Synthesis of MgAl2O4 spinel powder by combination of sol-gel and precipitation processes, Journal of the American Ceramic Society, 88 (11), 2005, pp. 3241-3244. 53. T. Shiono, K. Shiono, K. Miyamoto, and G. Pezzotti, Synthesis and characterization of (MgAl2O4) spinel precursor from a heterogeneous alkoxide solution containing fine MgO powder, Journal of the American Ceramic Society, 83 (1), 2000, pp. 235-237. 54. G. Ye and T. Troczynski, Mechanical activation of heterogeneous sol-gel precursors for synthesis of MgAl2O4 spinel, Journal of the American Ceramic Society, 88 (10), 2005, pp. 2970-2974. 152 55. J. R. Groza and R. J. Dowding, Nanoparticulate materials densification, Nanostructured Materials, 7 (7), 1996, pp. 749-768. 56. D.-M. Liu, Adsorption, rheology, packing, and sintering of nanosize ceramic powders, Ceramics International, 25 (2), 1999, pp. 107-113. 57. A. Banerjee, S. Das, S. Misra, S. Mukhopadhyay, Structural analysis on spinel (MgAl2O4) for application in spinel-bonded castables, Ceramics International, 35 (1), 2009, pp. 381-390. 58. V. Petkov, P.T. Jones, E. Boydens, B. Blanpain, P. Wollants, Chemical corrosion mechanisms of magnesia-chromite and chrome-free refractory bricks by copper metal and anode slag, Journal of the European Ceramic Society, 27 (6), 2007, pp. 2433-2444. 59. Y.-Y. Deng, H.-Z. Wang and H.-Z. Zhao, Influence of chrome-bearing sols vacuum impregnation on the properties of magnesia-chrome refractory, Ceramics International, 34 (3), 2008, pp. 573-580. 60. P. Lauzon, J. Rigby, C. Oprea, T. Troczynski, G. Oprea, Hydration studies on magnesia-containing refractories, in Proceedings of UNITECR? 2003, 2003, pp. 54-57. 61. CAS Registry ID: 1309-42-8. 62. E. Zen and J.V. Chernosky Jr., Correlated free energy values of anthophyllite, brucite, clinochrysotile, enstatite, forsterite, quartz, and talc, American Mineralogist, 61 (11-12), 1976, pp. 1156-1166. 153 63. N. H. Brett and P. J. Anderson, Mechanism of decomposition of brucite, Transactions of the Faraday Society, 63, 1967, pp. 2044-2050. 64. A. Kitamura, K. Onizuka, K. Tanaka, Hydration characteristics of magnesia, Taikabutsu Overseas, 16 (3), 1995, pp. 3-11. 65. S.D. Rocha, M.B. Mansur, V.S.T. Ciminelli, Kinetics and mechanistic analysis of caustic magnesia hydration, Journal of Chemical Technology and Biotechnology. 79 (8), 2004, pp. 1-6. 66. A. Yoschida, T. Nemoto, A. Kaneyasu, Evaluation method for hydration resistance of magnesia fine powder and effect of B2O3 content in magnesia raw materials, in: Proceedings of UNITECR? 2003, 2003, pp. 21-30. 67. R.A. Wogelius, K. Refson, D.G. Fraser, G.W. Grime, J.P. Goff, Periclase surface hydroxylation during dissolution, Geochimica et Cosmochimica Acta 59 (9), 1995, pp. 1875-1881. 68. G.K. Layden, G.W. Brindley, Kinetics of vapor-phase hydration of magnesium oxide, Journal of the American Ceramic Society, 46 (11), 1963 pp. 518-522. 69. A. Kitamura, K. Onizuka, K. Tanaka, Hydration characteristics of magnesia, Taikabutsu Overseas 16 (3), 1995, pp. 3-11. 70. R. Salom?o, L.R.M. Bittencourt, V.C. Pandolfelli, A novel magnesia based binder (MBB) for refractory castables, Refractories Manual, 58, 2009, pp. 21?24. 154 71. J. R. Donald, J. M. Toguri and C. Doyle, Surface interactions between fayalite slags and synthetic spinels and solid solutions, Metallurgical and Materials Transactions B, 29 (2), 1998, pp. 317-323. 72. C. Pellerin and S. M. Booker, Reflections on Hexavalent Chromium: Health Hazards of an Industrial Heavyweight, Environmental Health Perspectives, 108 (9), 2000, pp. A402-A407. 73. A. Tsuchinari, H. Okamoto and Y. Yokoyama, Chrome-free refractory for vacuum degassers, Proceedings of UNITECR?95, 2, 1995, pp. 333-340. 74. D.J. Griffin, Chrome-free generation, World Cement, 27 (3), 1996, pp. 88-92. 75. http://www.materialsdata.com/ 76. ASTM C830-00 (2011): Standard Test Methods for Apparent Porosity, Liquid Absorption, Apparent Specific Gravity, and Bulk Density of Refractory Shapes by Vacuum Pressure. 77. C133-97 (2008) e1: Standard Test Methods for Cold Crushing Strength and Modulus of Rupture of Refractories. 78. G. Ye, (2005), Reactive processing of ceramic binding systems for refractory castables, Ph.D. Thesis, The University of British Columbia, Vancouver BC, Canada. 155 79. C1445-07 (2007): Standard Test Method for Measuring Consistency of Castable Refractory Using a Flow Table. 80. R. M. Tyson and L. L. Y. Chang, The Systems ZnO-Fe2O3-SnO2 and MgO-Fe2O3-SnO2 at 1060?C, Journal of the American Ceramic Society, 64 (1), 1981, pp. C-4?C-6. 81. R. Lodha, H. Zargar, T. Troczynski and G. Oprea, Magnesia rich complex spinel bonded basic castables, Proceedings of the 12th Biennial Conference on Refractories, UNITECR 2011, Japan, 153-156. 82. H. Zargar, C. Oprea and G. Oprea, Characterization of Chromium-Free Spinel-Bonded Trial Bricks Used in the BBOC at Teck Metals Ltd., Trail Operations? Part II: Used bricks, UBCeram Private Communication, September 2012. 156 Appendix A Calculation for predicted inversion in spinel solid solution For a composition containing 60 wt. % MgO, 26 wt. % Al2O3, 8 wt. % Fe2O3, 3.5 wt. % SnO2 and 2.5 wt. % TiO2, the macro below will assign the following values: Al: 0.510 Cr: 0.000 Fe: 0.100 Sn: 0.023 Ti: 0.031 In the next step, the macro calculates the stoichiometric value for Mg in spinel solid solution using ?Mg = Al / 2 + Cr / 2 + Fe3 / 2 + Sn * 2 + Ti * 2? Mg: 0.413 Subsequently D5 and D10 are assigned to Fe and Sn, respectively and amount of bivalent cations in normal spinel position (at tetrahedral site) is calculated using ?NorS = Al / 2 + Cr / 2 + Fe3 / 2 + Sn + Ti - 0.97 * D5 / 2 - 0.95 * D10? NorS: 0.288 Since the remaining bivalent cations are at octahedral site, subtracting the amount of bivalent cations at tetrahedral site from total number of bivalent cations gives the amount of bivalent cations in octahedral site (inverse spinel) and inversion as: ?InvS = Mg - NorS\" and ?CompS = 100 * InvS / (NorS + InvS)? InvS: 0.125 CompS: 30 157 Sub TextBox1_Click() Dim Al, Cr, Fe3, Sn, Ti, Mg, Fe2 As Double Dim D5, D10, NorS, InvS, CompS As Double Al = Worksheets(\"Sheet1\").Range(\"Al2O3\") * 2 / 101.96 Cr = Worksheets(\"Sheet1\").Range(\"Cr2O3\") * 2 / 151.99 Fe3 = Worksheets(\"Sheet1\").Range(\"Fe2O3\") * 2 / 159.69 Sn = Worksheets(\"Sheet1\").Range(\"Sn1O2\") / 150.71 Ti = Worksheets(\"Sheet1\").Range(\"Ti1O2\") / 79.87 Mg = Al / 2 + Cr / 2 + Fe3 / 2 + Sn * 2 + Ti * 2 D5 = Fe3 D10 = Sn NorS = Al / 2 + Cr / 2 + Fe3 / 2 + Sn + Ti - 0.97 * D5 / 2 - 0.95 * D10 InvS = Mg - NorS CompS = 100 * InvS / (NorS + InvS) Worksheets(\"Sheet1\").Cells(12, 2).Value = CompS End Sub 158 Appendix B Calculations from molar to weight percent Molecular weight of MgO: 40.3 Molecular weight of Al2O3: 102.0 Molecular weight of SnO2: 150.7 Molecular weight of TiO2: 79.9 Sample calculation for Category I Compositions (Section 4.2.1) ST10 consists of 10 mole % Mg2TiO4 and 90 mole % MgAl2O4 For 100 moles of spinel phase, the mass of MgO: (2 ? 10 + 90) ? 40.3 = 4730 g TiO2: 10 ? 79.9 = 799 g Al2O3: 90 ? 102.0 = 9180 g Total mass = 14709 g Converting to weight percent, MgO: 100 ? 4730 ? 14709 = 32.2 %; Similarly, TiO2 = 5.4 %; And Al2O3 = 62.4 % 159 Sample calculation for Category II Compositions (Section 4.2.2) MS10 consists of 10 mole % Mg2TiO4 and 90 mole % MgAl2O4 in the spinel phase; MgO content is fixed at 60 wt. % For 100 moles of spinel phase, the mass of SnO2: 10 ? 150.7 = 1507 g Al2O3: 90 ? 102.0 = 9180 g Total mass = 10687 g Converting to weight percent, the total contribution from Al2O3 and SnO2 is 40 wt. % SnO2: 40 ? 1507 ? 10687 = 5.7 %; Similarly, Al2O3 = 34.3 %; MgO (predetermined) = 60.0 % 160 Appendix C Materials and manufacturers Dead burnt magnesia ? MgO (purity 99% +) fired in excess of 1000 ?C, Nedmag Industries, Netherlands Light burnt magnesia ? MgO dehydrated at 700 to 1000 ?C, Fisher Scientific Sintered magnesia ? MgO fired with dopants at temperatures in excess of 1400 ?C, Baymag Industries, Exshaw AB Alphabond 300 ? Hydratable alumina from Alcoa, used for improving green strength of castables, Almatis, Inc., USA. A1000SG ? Al2O3 of >99% purity and d50 < 1 ?m, Almatis, Inc., USA. ADS3 ? Dispersant for high-alumina castables, containing organo-metallic compounds, Alcoa World Chemicals, Germany. ADW1 ? Dispersant for high-alumina castables, containing organo-metallic compounds, Alcoa World Chemicals, Germany. AEROXIDE Alu C ? nano alumina (Al2O3) of > 99.8 % purity, Evonik Degussa Corporation, Germany. AEROXIDE TiO2 P25 ? nano titania of > 99.8 % purity, Evonik Degussa Corporation, Germany. AERODISP W 740 X ? dispersed nano titania 40 wt. % in water soluble base, Evonik Degussa Corporation, Germany. 161 FS10 ? Polycarboxylate based organic dispersant, BASF - The Chemical Company, Germany. FW10 ? Polycarboxylate based organic dispersant, BASF - The Chemical Company, Germany. FS60 ? Polycarboxylate based organic dispersant, BASF - The Chemical Company, Germany. Secar 71 ? Cement containing 71 % Al2O3, not containing SiO2, used for castables, Kerneos Inc., USA Secar 80 ? Cement containing 80 % Al2O3, not containing SiO2, used for castables, Kerneos Inc., USA VP Disp W 640 ZX ? dispersed nano alumina 40 wt. % in water soluble base, Evonik Degussa Corporation, Germany. Ferric oxide ? d50 < 1 ?m, purity > 99 %, Fisher Scientific. Tin dioxide ? d50 < 1 ?m, purity > 99.9 %, Fisher Scientific. Titanium dioxide ? d50 < 1 ?m, purity > 99.5 %, Fisher Scientific. All other oxides and water soluble organics were purchased from Fisher Scientific and/or Alfa Aesar and were of > 99.5% purity. 162 Appendix D Instrumental and measurement precision Balance accuracy 0.1 g XRD instrumental error in 2? 0.04 ? Digital callipers 0.01 mm Hydraulic press 10 N Mass variation in mixes (max 50 g) 0.2 g XRD 2? values 0.08 ? Bulk density 0.1 g.cm-3 Apparent porosity 2 % Water absorption 0.5 % Apparent specific gravity 0.2 g.cm-3 Flow 1 % Working and setting time 1 min Lattice parameter 0.5 pm Mass error in bulk compositions 2 g 163 Appendix E Supplementary experimental data Table E.1. Conversion from size fraction in mm to standard sieve size in mesh for C1 Size fraction % Size fraction (mesh) % 3 ? 6 mm 12 -2 + 4 2.87 1 ? 3 mm 35 - 4 + 8 15.69 0 ? 1 mm 17 - 8 + 28 29.32 -180 # 16 - 28 + 100 10.95 - 100 + 180 0.69 - 180 16.29 Table E.2. Conversion from size fraction in mm to standard sieve size in mesh for C2 Size fraction % Size fraction (mesh) % 3 ? 6 mm 14 -2 + 4 3.35 1 ? 3 mm 36 - 4 + 8 17.37 0 ? 1 mm 20 - 8 + 28 30.43 -180 # 13 - 28 + 100 12.83 - 100 + 180 0.76 - 180 13.35 164 Table E.3. Castable composition and properties using C1 aggregates and in-house binder system (water added 5.5 wt. %) Material (%) M-C0 M-N0 M-F0 C-F0 Fused magnesia 7.0 0.0 0.0 0.0 Dead burnt magnesia 0.0 7.0 0.0 0.0 Light burnt magnesia 0.0 0.0 7.0 0.0 Chromia 0.0 0.0 0.0 5.5 Alumina 5.0 5.0 5.0 6.5 Alphabond 300 4.5 4.5 4.5 4.5 Titania 0.3 0.3 0.3 0.3 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Nano-titania powder 0.2 0.2 0.2 0.2 Nano-alumina powder 0.3 0.3 0.3 0.3 Secar 71 0.5 0.5 0.5 0.5 Results Flow (%) 15 - - 18 Working time (minutes) 5 - - 7 Setting time (minutes) 7 <4 <4 9 165 Table E.4. Castable composition and properties using C2 aggregates and in-house binder system (water added 5.5 wt. %) Material (%) M-C1 M-N1 M-F1 C-F1 Fused magnesia 7.0 0.0 0.0 0.0 Dead burnt magnesia 0.0 7.0 0.0 0.0 Light burnt magnesia 0.0 0.0 7.0 0.0 Chromia 0.0 0.0 0.0 5.5 Alumina 5.0 5.0 5.0 6.5 Alphabond 300 4.5 4.5 4.5 4.5 Titania 0.3 0.3 0.3 0.3 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Nano-titania powder 0.2 0.2 0.2 0.2 Nano-alumina powder 0.3 0.3 0.3 0.3 Secar 71 0.5 0.5 0.5 0.5 SPMA 0.05 0.05 0.05 0.05 Citric Acid 0.02 0.02 0.02 0.02 Results Flow (%) - - - 2 Working time (minutes) - - - 5 Setting time (minutes) <4 <4 <4 6 166 Table E.5. Castable composition and properties using C2 aggregates and in-house binder system (water added 5.5 wt. %) Material (%) M-C2 M-N2 M-F2 C-F2 Fused magnesia 7.0 0.0 0.0 0.0 Dead burnt magnesia 0.0 7.0 0.0 0.0 Light burnt magnesia 0.0 0.0 7.0 0.0 Chromia 0.0 0.0 0.0 5.5 Alumina 5.0 5.0 5.0 6.5 Alphabond 300 4.5 4.5 4.5 4.5 Titania 0.3 0.3 0.3 0.3 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Nano-titania powder 0.2 0.2 0.2 0.2 Nano-alumina powder 0.3 0.3 0.3 0.3 Secar 71 0.5 0.5 0.5 0.5 ADS3 0.25 0.25 0.25 0.25 ADW1 0.75 0.75 0.75 0.75 Results Flow (%) - - - 4 Working time (minutes) - - - 4 Setting time (minutes) <4 <4 <4 5 167 Table E.6. Castable composition and properties using C2 aggregates and in-house binder system (water added 5.75 wt. %) Material (%) M-C3 M-N3 M-F3 C-F3 Fused magnesia 7.0 0.0 0.0 0.0 Dead burnt magnesia 0.0 7.0 0.0 0.0 Light burnt magnesia 0.0 0.0 7.0 0.0 Chromia 0.0 0.0 0.0 5.5 Alumina 5.0 5.0 5.0 6.5 Alphabond 300 4.5 4.5 4.5 4.5 Titania 0.3 0.3 0.3 0.3 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Nano-titania powder 0.2 0.2 0.2 0.2 Nano-alumina powder 0.3 0.3 0.3 0.3 Secar 71 0.5 0.5 0.5 0.5 ADS3 0.0 0.25 0.25 0.25 ADW1 0.0 0.75 0.75 0.75 SPMA 0.05 0.05 0.05 0.05 Citric Acid 0.02 0.02 0.02 0.02 Results Flow (%) 3 2 - 7 Working time (minutes) 4 4 - 4 Setting time (minutes) 5 5 <4 5 168 Table E.7. Castable composition and properties using C2 aggregates and in-house binder system (water added 5.75 wt. %) Material (%) M-N4 M-N5 M-N6 M-N7 M-N8 Dead burnt magnesia 7.0 7.0 7.0 7.0 7.0 Alumina 5.0 5.0 5.0 5.0 5.0 Alphabond 300 4.5 4.5 4.5 4.5 4.5 Titania 0.3 0.3 0.3 0.3 0.3 Tin Oxide 0.6 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 0.8 Nano-titania powder 0.2 0.2 0.2 0.2 0.2 Nano-alumina powder 0.3 0.3 0.3 0.3 0.3 Secar 71 0.5 0.5 0.5 0.5 0.5 ADS3 0.0 0.25 0.00 0.25 0.25 ADW1 0.0 0.75 0.00 0.75 0.75 SPMA 0.05 0.05 0.05 0.05 0.05 Boric Acid 0.02 0.02 0.00 0.00 0.00 Oxalic Acid 0.00 0.00 0.02 0.02 0.02 Sucrose 0.00 0.00 0.00 0.00 0.05 Results Flow (%) - - - 5 6 Working time (minutes) - - - 4 4 Setting time (minutes) <4 <4 <4 5 5 169 In order to ascertain that hydration of -180# size fraction of C1 and C2 was creating flash-setting, the -180# size fraction is replaced by the available spinel aggregate. Table E.8. Castable composition and properties using modified aggregates and in-house binder system Material (%) M-N9 M-N10 M-N11 C-F4 Dead burnt magnesia 7.0 7.0 7.0 0.0 Chromia 0.0 0.0 0.0 5.5 Alumina 5.3 5.3 5.3 6.8 Alphabond 300 4.5 4.5 4.5 4.5 Titania 0.5 0.5 0.5 0.5 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Secar 71 0.5 0.5 0.5 0.5 ADS3 0.0 0.25 0.00 0.00 ADW1 0.0 0.75 0.00 0.00 SPMA 0.05 0.00 0.05 0.05 Citric acid 0.02 0.00 0.00 0.00 Boric acid 0.00 0.00 0.02 0.02 Water 5.75 5.75 5.75 5.5 Results Flow (%) - - - - Working time (minutes) - - - 4 Setting time (minutes) <4 <4 <4 5 170 In order to ascertain that hydration of mag-chrome fines is creating flash-setting; the -180# size fraction is replaced by the available spinel aggregate and the 0 - 1 mm size fraction is replaced by spinel aggregate of 0 - 1 mm size fraction. Table E.9. Castable composition and properties using modified aggregates and in-house binder system Material (%) M-N12 M-N13 C-F5 C-F6 Dead burnt magnesia 7.0 7.0 0.0 0.0 Chromia 0.0 0.0 5.5 5.5 Alumina 5.3 5.3 6.8 6.8 Alphabond 300 4.5 4.5 4.5 4.5 Titania 0.5 0.5 0.5 0.5 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Secar 71 0.5 0.5 0.5 0.5 ADS3 0.4 0.4 0.4 0.4 ADW1 0.6 0.6 0.6 0.6 SPMA 0.05 0.05 0.05 0.05 Citric acid 0.02 0.00 0.02 0.00 Boric acid 0.00 0.02 0.00 0.02 Water 5.75 5.75 5.5 5.5 Results Flow (%) - - - - Working time (minutes) - - - - Setting time (minutes) <4 <4 <4 <4 171 Table E.10. Castable composition and properties using modified aggregates and in-house binder system Material (%) M-N14 M-N15 C-F7 C-F8 Dead burnt magnesia 7.0 7.0 0.0 0.0 Chromia 0.0 0.0 5.5 5.5 Alumina 9.8 5.8 6.8 7.3 Alphabond 300 0.0 4.5 0.0 4.5 Titania 0.5 0.5 0.5 0.5 Tin Oxide 0.6 0.6 0.6 0.6 Iron Oxide 0.8 0.8 0.8 0.8 Secar 71 0.5 0.0 0.5 0.0 ADS3 0.4 0.4 0.4 0.4 ADW1 0.6 0.6 0.6 0.6 Water 5.75 5.75 5.5 .5 Results Flow (%) - 2 - 4 Working time (minutes) - 4 - 4 Setting time (minutes) <4 5 <4 5 172 Table E.11. Properties of chrome free aggregates of composition MTSF8 fired at 1450 ?C, 5 hours BD (g.cm-3) AP (%) WA (%) ASG (g.cm-3) CCS (MPa) 3.48 ? 0.05 2.2 ? 0.3 0.6 ? 0.1 3.56 ? 0.03 18.2 ? 1.5 Due to the inability to press and fire samples for static corrosion, these tests could not be carried out completely. Figures E.1 and E.2 below show the leakage of slag from the sides, hence these may be done at a later date when large batch of aggregates can be prepared. Figure E.3 shows a comparison with rebonded fused grain mag-chrome brick, which did not have slags leaking from the sides. Figure E.1. Pressed and fired chrome-free specimen for static corrosion 173 Figure E.2. Pressed and fired chrome-free specimen after static corrosion Figure E.3. Static corrosion test of industrial brick "@en . "Thesis/Dissertation"@en . "2013-11"@en . "10.14288/1.0103364"@en . "eng"@en . "Materials Engineering"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "Attribution-NonCommercial-NoDerivatives 4.0 International"@en . "http://creativecommons.org/licenses/by-nc-nd/4.0/"@en . "Graduate"@en . "Synthesis and sintering of chromium-free complex spinels in the MgO-Al2O3-FeOx-Me4+O2 systems"@en . "Text"@en . "http://hdl.handle.net/2429/45246"@en .