UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Reactive processing of ceramic binding systems for refractory castables Ye, Guotian 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2005-105917.pdf [ 24.76MB ]
Metadata
JSON: 831-1.0078777.json
JSON-LD: 831-1.0078777-ld.json
RDF/XML (Pretty): 831-1.0078777-rdf.xml
RDF/JSON: 831-1.0078777-rdf.json
Turtle: 831-1.0078777-turtle.txt
N-Triples: 831-1.0078777-rdf-ntriples.txt
Original Record: 831-1.0078777-source.json
Full Text
831-1.0078777-fulltext.txt
Citation
831-1.0078777.ris

Full Text

REACTIVE PROCESSING OF CERAMIC BINDING SYSTEMS FOR REFRACTORY CASTABLES by G U O T I A N Y E B . Sc., Wuhan University of Science and Technology, China 1982 M . Sc., Luoyang Institute of Refractories Research, China 1985 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y in T H E F A C U L T Y O F G R A D U A T E S T U D I E S ( M A T E R I A L S E N G I N E E R I N G ) T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A A U G U S T 2005 © G U O T I A N Y E , 2005 Abstract i i ABSTRACT The general applied objective of this project is to develop hydratable alumina-bonded castables of properties comparable to the high-temperature fired basic bricks. This work is focused on improving the sintering of the ceramic binding systems at relatively low temperatures (~1300°C) through incorporation of ultrafine powders. Ultrafine magnesium aluminate spinel powders were synthesized using already known and original methods. Combustible ingredients were incorporated to prevent the direct contacts of the precursor particles during drying and the combustibles leave a continuous pore network, when they burn off during calcination at 700-800°C. Mechanical activation of the spinel precursor prepared using a heterogeneous sol-gel process decreased the activation energy for spinel formation from 688 kJ/mol to 468 kJ/mol, and lowered the incipient temperature of spinel formation from 900°C to 800°C, and the temperature of complete spinellization from >1280°C to 900°C. A new method for preparing homogeneous spinel precursor was presented and completely crystallized spinel with specific surface area of 105 m /g, and crystallite size of 13 nm was formed from the precursor after calcination at 900°C. This work established the influence of ultrafine-sized powders on the sintering and strength development of multi-sized systems rich in large particles at relatively low temperatures, with emphasis on the temperatures around 1280°C. It was found that the relationship between the strength and linear shrinkage for the powder compacts consisting of uniform particle size distribution does not apply to the binding systems with wide particle size distributions (90 nm - 90 pm). The addition of the ultrafine powders (<0.5 pm median) up to 10% had no significant influence on strength in two-component binding Abstract m systems after firing at 1280°C and 1450°C. However, the ultrafine powders increased the strength of three-component binding systems presumably because the addition of ultrafine powders increased effective contacts for sintering. The particle size of the ultrafine power plays an important role in enhancing the strength of the castables after heat treatment at 816°C and 1280°C. Aggregate types have substantial influence on the strength of the castables after firing at 1280°C, eventually due to the cracks generated by in-situ spinel formation and mismatch in thermal expansion coefficients between the aggregate and the matrix. Spinel-based castables with high bending strength (>15 MPa) were obtained with addition of 2% ultrafine spinel (0.2 pm median) after firing at 1280°C. This research work also contributes to the reaction mechanisms of hydratable alumina with various forms of M g O . After hydration at 20°C for 48 h, a hydrotalcite hydrate was formed in the mixture of hydratable alumina and reactive magnesia, while such a hydrate was not observed in the mixtures of hydratable alumina and deadburnt or fused magnesite under the same hydration conditions. After hydration at room temperature for 48 h at 20°C and then for 12 h at 110°C, hydrotalcite compounds were formed in all three mixtures. The presence of deadburnt/fused magnesia in hydratable alumina-bonded castables improved the strength of the castables after drying at 110°C because of the hydrogen bonding of hydrotalcite and Mg(OH)2 on magnesia particles. The polycondensation accompanying dehydroxylation of hydrotalcite and M g ( O H ) 2 up to 816°C contributed to the higher strength of hydratable alumina-bonded castables containing magnesia. Table of Contents iv Table of Contents Abstract i i Table of Contents iv List of Tables x List of Figures xi List of Symbols xx List of Acronyms ... xxi Acknowledgements xx i i CHAPTER 1 INTRODUCTION 1 CHAPTER 2 LITERATURE REVIEW 7 2.1 Magnesium Aluminate Spinel 7 2.2 Spinel Synthesis Using Wet-Chemical Routes 8 2.2.1 Coprecipitation 9 2.2.2 Sol-Gel Processing 11 2.2.3 Other Methods 12 2.3 Spinel Synthesis via Mechanical Activation 13 2.3.1 Advantage of Mechanochemical Processing 13 2.3.2 Structural Evolution of Mechanically Activated Materials 14 2.3.3 Mechanical Activation Assisted Synthesis . . . . 15 2.4 Sintering in the Presence of Ultrafine Powders 16 2.4.1 The Driving Force for Sintering 16 2.4.2 Driving Force versus Particle Size 18 Table of Contents 2.4.3 Sintering Approaches 19 2.4.4 Sintering Behavior of Mul t i -Modal Systems 20 2.4.5 Sintering of Refractory Binding Systems 24 2.5 Hydratable Alumina-Bonded Castables :.. 26 2.5.1 Binders for Refractory Castables 26 2.5.2 Hydration of Hydratable Alumina 30 2.5.3 Ultrafine Powers for Castables 31 CHAPTER 3 SCOPE AND OBJECTIVES 32 3.1 Scope 32 3.2 Objectives 32 CHAPTER 4 METHODOLOGY 34 4.1 Spinel Preparation and Characterization Procedures 34 4.1.1 Preparation of Spinel Precursors by Coprecipitation 34 4.1.2 Preparation of Spinel Precursors by Heterogeneous Sol-Gel Process 35 4.1.3 Preparation of Spinel Precursors by Mechanical Activation 36 4.1.4 Mechanical Activation of Heterogeneous Sol-Gel Precursors 37 4.1.5 Preparation of Spinel Precursors by Sol-Gel and Precipitation 37 4.1.6 Morphology of Spinel Powders 38 4.1.7 Phases and Reaction Kinetics 41 4.2 Compact Preparation and Characterization Procedures 42 4.2.1 Raw Materials 42 4.2.2 Formulations of Binary Systems 43 Table of Contents vi 4.2.3 Formulations of Tr i -Modal Systems 45 4.2.4 Sample preparation 45 4.2.5 Particle Size Analysis 46 4.2.6 Sintering Kinetics 47 4.2.7 Physical Properties and Microstructure 47 4.3 Castable Preparation and Characterization Procedures 48 4.3.1 Raw Materials 48 4.3.2 Formulations of Castables 49 4.3.3 Sample Preparation 53 4.3.4 Flowability 53 4.3.5 Physical Properties 53 4.3.6 Hydration Experiments 54 C H A P T E R 5 S Y N T H E S I S OF M A G N E S I U M A L U M I N A T E S P I N E L 56 5.1 Spinel Synthesis by Coprecipitation 56 5.1.1 Morphology of the Calcined Powders 56 5.1.2 Phase Evolution during Calcination 58 5.2 Spinel Synthesis by Heterogeneous Sol-Gel 59 5.2.1 Morphology of the Calcined Powders 59 5.2.2 Phases of the Precursors 60 5.2.3 Phase Evolution during Calcination 62 5.3 Spinel Synthesis by Mechanical Activation 64 5.3.1 Phase Changes of the Precursors during M i l l i n g 64 Table of Contents vi 5.3.2 Phase Evolution during Calcination 65 5.3.3 Morphology of the Synthesized Powders 71 5.4 Mechanical Activation of Heterogeneous Sol-Gel Precursors 77 5.4.1 Phase and Morphology Changes of the Precursors during M i l l i n g 77 5.4.2 Dehydroxylation Behavior of the Precursors 80 5.4.3 Phase Evolution during Calcination 84 5.5 Spinel Synthesis by Sol-Gel Precipitation 89 5.5.1 Phase Evolution during Calcination 90 5.5.2 Morphology of the Calcined Powders 93 5.5.3 Particle Size and Agglomeration of the Calcined Powders 94 5.6 Summary 98 C H A P T E R 6 S I N T E R I N G O F B I N D I N G S Y S T E M S F O R C A S T A B L E S 101 6.1 B i -Moda l Systems 101 6.1.1 Influence of UF-Spinel-1 101 6.1.2 Influence of UF-Spinel-2 104 6.1.3 Influence of the In-House Prepared Spinel Powder 105 6.1.4 Comparison of Different Spinel Powders 108 6.2 Tri -Modal Systems I l l 6.2.1 Influence of Ultrafine Alumina 112 6.2.2 Influence of UF-Spinel-2 114 6.2.3 Comparison of Different-Sized Ultrafine Spinel 115 6.2.4 Contribution of Ultrafine Powders to Total Specific Surface Area 118 Table of Contents v i i i 6.3 Sintering Behavior of Two and Three-Component Compacts 121 6.3.1 Systems without Ultrafine Powders 121 6.3.2 Systems with Ultrafine Powders 123 6.3.3 Discussion on Sintering Evaluation 127 6.3.4 Sintering Kinetics 128 6.4 Summary 133 CHAPTER 7 SPINEL-BONDED BASIC CASTABLES 136 7.1 Influence of Ultrafine Alumina on Castable Properties 136 7.1.1 Castables with Magnesia Aggregate 136 7.1.2 Castables with Spinel Aggregate : 142 7.1.3 Influence of Aggregate Type 147 7.2 Influence of Ultrafine Spinel on Castable Properties 150 7.2.1 Influence of Ultrafine Spinel (d5 0=0.3 pm) 150 7.2.2 Influence of Ultrafine Spinel (d 5 0=0.2 pm) 155 7.3 Hydration of Hydratable Alumina in the Presence of Magnesia 160 7.3.1 Hydration of Hydratable Alumina and Different Forms of Magnesia 161 7.3.2 Hydration Products of Mixtures of Hydratable Alumina and Magnesia 164 7.3.3 Phase Compositions of Hydrotalcite Hydrates 166 7.4 Effect of Magnesia on the Bonding of Hydratable Alumina-Bonded Castables.... 167 7.4.1 Strength and Porosity 167 7.4.2 Morphology of Hydrates 172 7.4.3 Dehydroxylation of the Hydrates 173 Table of Contents ix 7.5 Summary 175 CHAPTER 8 CONCLUSIONS 178 CHAPTER 9 FUTURE WORK 183 REFERENCES 185 THESIS RELATED PUBLICATIONS. :.. 196 APPENDIX 198 Appendix A Sample Descriptions 198 Appendix B Physical Properties of the Two-Component Mixes 200 Appendix C Physical Properties of the Three-Component Mixes 203 Appendix D Glossary Related to Refractories 205 List of Tables x List of Tables Table 4.1-1 Process parameters for heterogeneous sol-gel preparation of spinel precursors 36 Table 4.2-1 Raw materials for compacts 43 Table 4.2-2 Formulations of bi-modal mixtures 44 Table 4.2-3 Formulations of tri-modal mixtures 45 Table 4.3-1 Raw materials for castables 48 Table 4.3-2 Formulations of castables with fused magnesia aggregate ...49 Table 4.3-3 Formulations of castables with fused spinel aggregate 50 Table 4.3-4 Formulations of castables with ultrafine spinel powder ( d 5 0 = 0.3 pm) 51 Table 4.3-5 Formulations of castables with ultrafine spinel powder (dsn = 0.2 pm) 52 Table 6.3-1 Summary of sintering kinetics 130 List of Figures x i List of Figures Figure 2.1-1 Spinel structure 7 Figure 2.1-2 MgO -Al2C»3 phase diagram 8 Figure 2.2-1 Schematic diagram for preventing agglomeration by combustible additives 10 Figure 2.3-1 Change in activation energy before and after mill ing 14 Figure 2.4-1 Schematic microstructural development during solid-state sintering 17 Figure 2.4-2 Scanning electron micrograph of sintering bonding between particles 17 Figure 2.4-3 Densification behavior around large particles 24 Figure 2.4-4 Schematic particle size composition of refractory castables 25 Figure 2.5-1 Effect of silica fume addition on flowability of cement-free castables 28 Figure 2.5-2 Effect of silica fume addition on H M O R of cement-free castables 28 Figure 4.1-1 Procedures for preparing M A - 9 spinel precursor by sol-gel and precipitation...38 Figure 5.1-1 S E M micrographs of (a) M A - 1 after calcination at 800°C, (b) M A - 2 calcined at 700°C, (c) M A - 3 calcined at 700°C and (d) M A - 5 calcined at 800°C 57 Figure 5.1-2 X R D patterns of (a) M A - 1 and (b) M A - 2 after calcination for 3h at different temperatures 58 Figure 5.2-1 X R D patterns of M A - 5 , M A - 5 A and M A - 6 precursors before calcination.. .60 Figure 5.2-2 X R D pattern of reactive magnesia hydrated for 22 h at 30°C 61 Figure 5.2-3 X R D patterns of M A - 5 (a) and M A - 6 (b) after thermal treatment at different List of Figures x i i temperatures 63 Figure 5.2-4 X R D patterns showing the effect of calcinating temperature on the formation of crystalline phases from a spinel precursor 64 Figure 5.3-1 X R D patterns of (a) MA-21 and (b) M A - 2 2 mixtures versus milling time 65 Figure 5.3-2 X R D patterns of M A - 2 1 after mill ing and calcination at (a) 400°C, (b) 600°C, (c) 800°C and (d) 1000°C for 5h 66 Figure 5.3-3 X R D patterns of M A - 2 2 versus mill ing time and heat treatment at (a) 400°C, (b) 600°C, (c) 800°C and (d) 1000°C for 5h 69 Figure 5.3-4 S E M micrographs of M A - 2 1 mixture after mil l ing for (a) Oh, (b) 5h, (c) lOh and(d)20h • , 72 Figure 5.3-5 S E M micrograph of M A - 2 1 mixture with mill ing for lOh (the particle circled in Figure 5.3-4) 72 Figure 5.3-6 S E M mapping of M A - 2 1 mixture; (a) and (b) without mill ing, (c) and (d) with 5h of mill ing 74 Figure 5.3-7 S E M micrographs of M A - 2 1 mixture after heat treatment at 1000°C for 5h without mil l ing 74 Figure 5.3-8 S E M mapping of (a) aluminum and (b) magnesium of the M A - 2 1 mixture after heat treatment at 1000°C for 5h without previous mill ing 75 Figure 5.3-9 S E M micrographs of M A - 2 1 mixture after heat treatment at 1000°C for 5h with previous mill ing for (a) Oh, (b) 5h, (c) lOh and (d)20h 75 Figure 5.4-1 X R D pattern of M A - 8 precursor milled for various duration of time 78 Figure 5.4-2 S E M micrograph of M A - 8 precursor after mill ing for (a) 0 h, (b) 1 h, (c) 3 h and (d) 5 h 79 List of Figures x i i i Figure 5.4-3 (a) D T A and (b) T G patterns (under a heating rate of 10°C/min) of the milled precursors 81 Figure 5.4-4 Activation energy of the dehydroxylation of the precursor under 300°C as a function of mill ing time 82 Figure 5.4-5 X R D pattern of the precursors milled for (a) 0 h, (b) 1 h, (c) 3 h and (d) 5 h and fired at different temperatures 85 Figure 5.4-6 Activation energy of spinel formation as a function of mil l ing time 88 Figure 5.4-7 S E M micrograph of the precursor milled for (a) 0 h, (b) 1 h, (c) 3 h and (d) 5 h and fired at 900°C for 5h 89 Figure 5.5-1 T G - D T A curves of the magnesium aluminate spinel precursor 90 Figure 5.5-2 X R D pattern of the magnesium aluminate spinel precursor 91 Figure 5.5-3 Spinel crystallization from the precursor after calcination for 5h at different temperatures 92 Figure 5.5-4 Temperature dependence of specific surface area and crystal size of the spinel powders 93 Figure 5.5-5 T E M micrograph of spinel prepared by sol-gel precipitation, after firing for 5h at 1000°C 94 Figure 5.5-6 Temperature dependence of particle size distribution of the spinel powders; (a) cumulative volume versus particle size, (b) differential volume versus particle size 96 Figure 5.5-7 Influence of mill ing on particle size distribution of the spinel powders; (a) cumulative volume versus particle size, (b) differential volume versus particle size 97 List of Figures xiv Figure 6.1-1 Particle size distribution of bi-modal mixture AO 101 Figure 6.1-2 Influence of UF-spinel-1 on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bi-modal mixtures after heat treatments 102 Figure 6.1-3 S E M micrograph of the spinel sample AO after firing at 1280°C 103 Figure 6.1-4 Influence of UF-spinel-2 on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bi-modal mixtures AO, A2-1 to A 2 -5 after heat treatment 104 Figure 6.1-5 Influence of MA-9-900 addition on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bi-modal mixtures AO, A3-1 to A3-5 after heat treatment 106 Figure 6.1-6 Influence of MA-9-900-mil led addition on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bi-modal mixtures AO, A4-1 to A4-5 after heat treatment 107 Figure 6.1-7 Influence of different ultrafine powders on green density of bi-modal mixtures after heat treatment for 24h at 110°C 108 Figure 6.1-8 Influence of different ultrafine powders on bulk density of bi-modal mixtures after heat treatment for 5h at (a) 1280°C and (b) 1450°C 109 Figure 6.1-9 Influence of different ultrafine powders on apparent porosity of bi-modal mixtures after heat treatment for 5h at (a) 1280°C and (b) 1450°C 110 Figure 6.1-10 Influence of different ultrafine powders on linear change of bi-modal mixtures after heat treatment for 5h at (a) 1280°C and (b) 1450°C 110 Figure 6.1-11 Influence of different ultrafine powders on compressive strength of bi-modal List of Figures xv mixtures after heat treatment for 5h at (a) 1280°C and (b) 1450°C I l l Figure 6.2-1 Particle size distribution of tri-modal mixture D l - 1 I l l Figure 6.2-2 Influence of UF-alumina addition on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-1, D2-2 and D2-3 with 5% UF-spinel-2 after heat treatment 112 Figure 6.2-3 Influence of UF-alumina addition on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D4-1, D4-2 and D4-3 with 5% UF-spinel-1 after heat treatments 113 Figure 6.2-4 Influence of UF-spinel-2 on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D l - 1 , D l - 2 and D l - 3 after firing at different temperatures 115 Figure 6.2-5 Influence of 5% different ultrafine spinel on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-1 and D4-1 after firing at different temperatures 116 Figure 6.2-6 Influence of 5% different ultrafine spinel on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-2 and D4-2 with 2% UF-alumina after firing at different temperatures 117 Figure 6.2-7 Influence of different ultrafine spinel on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-3 and D4-3 with 5% UF-alumina after firing at different temperatures 118 Figure 6.2-8 Contribution of UF-spinel-2 to the total specific surface area of the mixtures 119 Figure 6.2-9 Figure 6.3-1 S E M Micrograph of (a) UF-spinel-1 and (b) UF-spinel-2 120 Comparison between bimodal mix AO and tri-modal mix D l - 1 without List of Figures xvi ultrafine powder for (a) B D , (b) A P , (c) P L C and (d) C C S 122 Figure 6.3-2 Comparison between bimodal mix A1-4 and the tri-modal mix D4-1 with 5% UF-spinel-1 for (a) B D , (b) A P , (c) P L C and (d) C C S 124 Figure 6.3-3 Comparison between bimodal mix A2-4 and tri-modal mix D l - 2 with 5% UF-spinel-2 for (a) B D , (b) A P , (c) P L C and (d) C C S 125 Figure 6.3-4 Schematic particle packing in (a) bimodal and (b) tri-modal compacts with ultrafine powder 126 Figure 6.3-5 Linear shrinkage of bi-modal mixtures at heating rates of (a) 2°C/min and (b) 4°C/min; AO (without ultrafine powder), A l - 4 (5% UF-spinel-1), A2-4 (5% UF-spinel-2) and A4-4 (5% in-house spinel MA-9-900-milled) 129 Figure 6.3-6 Linear shrinkage of tri-modal mixtures at heating rates of (a) 2°C/min and (b) 4°C/min; D l - 1 (without ultrafine powder) , D2-1 (5% UF-spinel-2), D4-1 (5% UF-spinel-1) and D5-1 (5% in-house spinel MA-9-900-milled). . . .129 Figure 6.3-7 Constant-rate-of-heating plots of (a) bi-modal mixtures and (b) tri-modal systems at a heating rate of 2°C/min 130 Figure 6.3-8 Schematic representation of influence of surface diffusion on constant-rate-of-heating data plot 133 Figure 7.1-1 Influence of UF-alumina on (a) B D , (b) A P , (c) P L C and (d) M O R of magnesia-based castables M l - 0 , M l - 3 and M l - 6 after heat treatment at different temperatures 137 Figure 7.1-2 Micrograph of the matrix of the castable M l - 0 after heat treatment at (a) 110°Cand ( b ) 8 1 6 ° C 138 Figure 7.1-3 Micrograph of M l - 6 after firing at 1280°C; (a) cracks around magnesia aggregate and (b) cracks between magnesia aggregate and matrix 140 List of Figures xvn Figure 7.1-4 Influence of magnesia powder (0-4%) on (a) B D , (b) A P , (c) P L C and (d) M O R of magnesia-based castables M l - 6 and M2-6 after firing at different temperatures 142 Figure 7.1-5 Influence of UF-alumina on (a) B D , (b) A P , (c) P L C and (d) M O R of spinel-based castables S l - 0 , SI-3 and SI-6 after firing at different temperatures , ..143 Figure 7.1-6 S E M micrograph of (a) matrix and (b) the boundary between spinel aggregate and matrix in the castable SI-6 with 4% M g O powder after firing at 1280°C '...144 Figure 7.1-7 Influence of magnesia powder on (a) B D , (b) A P , (c) P L C and (d) M O R of the spinel-based castables (S1-6: with 4% magnesia powder; S2-6: without magnesia powder) after firing at different temperatures 146 Figure 7.1-8 S E M micrographs of (a) the matrix and (b) the boundary between spinel aggregate and matrix in the castable S2-6 without M g O powder after firing at 1280°C 147 Figure 7.1-9 Comparison between castables with spinel aggregate S l - 0 , SI-3 and SI-6 and magnesia aggregate M l - 0 , M l - 3 and M l - 6 fired at 1280°C for (a) P L C and (b) M O R 148 Figure 7.2-1 Influence of UF-spinel-1 on flowability of magnesia-based castables M3-0 , M3-3 and M3-5 with water addition of 5.1 % 150 Figure 7.2-2 Influence of UF-spinel-1 on (a) B D , (b) A P , (c) P L C and (d) M O R of magnesia-based castables M3-0 , M3-3 and M3-5 151 Figure 7.2-3 S E M micrograph of magnesia based castables (a) without ultrafine spinel (M3-0) and (b) with 5% UF-spinel-1 (M3-5) after firing at 1280°C 152 List of Figures xvm Figure 7.2-4 Influence of UF-spinel-1 on (a) B D , (b) A P , (c) P L C and (d) M O R of the spinel-based castables S3-0 and S3-5 153 Figure 7.2-5 S E M micrograph of spinel-based castables (a) without ultrafine spinel (S3-0) and (b) with 5% UF-spinel-1 (S3-5) after firing at 1280°C 154 Figure 7.2-6 Comparison between magnesia-based (M3-0 and M3-5) and spinel-based (S3-0 and S3-5) castables with different amounts of UF-spinel-1 after firing at 1280°C for (a) P L C and (b) M O R 155 Figure 7.2-7 Influence of UF-spinel-2 on (a) B D , (b) A P , (c) P L C and (d) M O R of spinel-based castables S4-0, S4-2 and S4-5 156 Figure 7.2-8 Micrographs of spinel based castables (a) without ultrafine spinel (S4-0) and (b) with 5% UF-spinel-2 (S4-5) after firing at 1280°C 157 Figure 7.2-9 Influence of ultrafine powders on the strength of spinel-based castables S2-6, S3-5, S4-5 after heat treatment at 816°C and 1280°C 159 Figure 7.3-1 (a) X R D patterns and (b) T G curves of hydratable alumina hydrated under (A) Hydration-A and (B) Hydration-B 161 Figure 7.3-2 X R D patterns of samples hydrated under (a) Hydration-1 and (b) Hydration-2; (A) reactive magnesia, (B) fused magnesia and (C) deadburnt magnesia 163 Figure 7.3-3 T G curves for the thermal decomposition of various forms of magnesia after (a) Hydration-A and (b) Hydration-B conditions 163 Figure 7.3-4 X R D patterns of samples (with A l 2 0 3 : M g O ratio = 16:84) hydrated under (a) Hydration-1 and (b) Hydration-2 164 Figure 7.3-5 T G curves for the thermal decomposition of the magnesia-alumina mixtures after (a) Hydration-A and (b) Hydration-B conditions 165 List of Figures xix Figure 7.4-1 Comparison of M O R of castables with different aggregates and different amounts of M g O powder after drying at 110°C and firing at 816°C 168 Figure 7.4-2 Comparison of A P of castables with different aggregates and different amounts of M g O powder after drying at 110°C and firing at 816°C 171 Figure 7.4-3 S E M micrographs of (a) hydratable alumina and (b) Mixture III after hydration at 20°C for 48 h and 110°C for 12 h 173 Figure 7.4-4 T G - D T A curves for the thermal decomposition of (a) hydratable alumina and (b) Mixture III after hydration at 20°C for 48 h and 110°C for 12 h . . . 174 List of Symbols LIST OF SYMBOLS A: Constant (in Equation 2.4-4) B: Parameter exponentially dependent on temperature (in Equation. 2. B0: Collection of material parameters (in Equation. 2.4-3) D: Diameter of particles L0: Initial length of samples Q: Apparent activation energy of reaction or sintering R: Gas constant (8.31451 J / m o l K ) T: Temperature t: Time r. Radius of spherical particles V/. Volume fraction of fine particles y: Surface energy Q. Volume of an atom AGex: Excessive energy per atom in spherical particles AL: Linear change of samples List of Acronyms xxi LIST OF ACRONYMS A P : Apparent Porosity B D : Bulk Density B E T : Brunauer-Emmett-Teller B S D : Broad particle Size Distribution C C S : Cold Crushing Strength C R H : Constant-Rate-Heating D T A : Differential Thermal Analysis H M O R : Hot Modulus of Rupture M A : Magnesium Aluminate M O R : Modulus of Rupture N S D : Narrow particle Size Distribution P L C : Permanent Linear Change P V A : Polyvinyl Alcohol S E M : Scanning Electron Microscopy SGP: Sol-Gel-Precipitation S P M A : Sodium Polymethacrylate T E M : Transmission Electron Microscopy T G A : Thermogravimetric Analysis U F : Ultrafine X R D : X-Ray Diffraction Acknowledgements xx i i ACKNOWLEDGEMENTS I would especially like to extend my gratitude to my supervisors, Dr. Tom Troczynski and Dr. George Oprea, for their guidance, support and tolerance throughout my study at U B C . I would like to thank my advisor, Dr. Kevin Smith, for his illuminating comments. Special thanks are extended to Dr. Mat i Rausepp for his expertise in X R D analysis. M y thanks go to Carmen Oprea and Shuxin Zhou for their help and discussions. I would like to thank Elizabetta Pani, Sing Y i c k , Randy Clark, Mary Mager, Sally Finora, Francisco Melo , and Franz Moraw for their assistance in using experimental facilities. I would like to thank many friends in China for their advice and help. I would like to thank all the industrial sponsors of the research consortium on refractories at UBCeram, Clayburn Industries Ltd. , Hatch Associates Ltd. , Inco Ltd., R H I Canada Inc. and Tek Cominco Metals Ltd., and the National Science and Engineering Research Council of Canada for their financial support. The U B C University Graduate Fellowship (UGF) and C . L . Wang Memorial Scholarship are gratefully acknowledged. I want to thank Almatis Inc. (Bauxite, U S A ) , C - E Minerals (King of Prussia, U S A ) , Baymag Inc. (Calgary, Canada), Clayburn Industries Ltd. (Abbotsford, Canada) and Martin Marietta Magnesia Specialties (Raleigh, U S A ) for providing experimental materials. I feel a deep sense of gratitude to my parents for their ever-lasting inspiration all the way. I want to convey my heartfelt appreciation to my wife and son for their encouragement and support along this journey. Chapter I Introduction 1 CHAPTER 1 INTRODUCTION Refractory materials are generally categorized into bricks and monolithics (the glossary of terms in science of refractories is in Appendix D). Refractory bricks are shaped and fired before use. In contrast, monolithics are normally supplied as dry or wet mixtures of particulate materials and are shaped in installation of industrial vessels and fired during service. So monolithics are also referred to as unshaped refractories. The last two decades did not see many changes in brick making, except for carbon-containing bricks, but significant progress occurred in monolithics [1]. In comparison with the bricks, the major advantages of monolithics include exemption from the processes of shaping and firing, less consumption of manpower and energy, easier and quicker installation, and no joints as existed in brick linings. As a result, monolithics have found wide applications in steel, cement, and petrochemical industries [2, 3]. However, magnesia-chrome bricks are still the only choice for the linings of non-ferrous furnaces, where the bricks are subjected to serious corrosion by molten slag and matte, and corrosive fumes and gases [4, 7]. There is growing concern about the use of chromia-containing bricks due to the tendency of trivalent chromium to form hexavalent chromium at elevated temperatures [8]. Hexavalent chromium compounds are considered carcinogenic [9]. Therefore, refractories containing chromia are environmentally hazardous and, as a result, chromia-free refractories are desired. Refractory castables, sometimes called refractory concretes, are usually supplied as dry mixtures of granular materials that need to be mixed with water or other specified liquid at the applications site. The materials are installed by pouring, usually followed by vibration, Chapter 1 Introduction 2 easily fitting into complex shapes with varied thicknesses. Refractory castables are the easiest to be installed and typically have the best quality among various monolithics. With regard to the advantages of monolithics described above, refractory castables without chromia would be preferred for the linings of the non-ferrous furnaces. Magnesium aluminate spinel ( M g A l 2 0 4 ) (referred to as "spinel" hereafter) possesses a combination of desirable properties for refractories: high melting point (2135°C), high resistance against chemical attack [10, 11], high mechanical strength both at room temperature ( M O R = 135-216 M P a , 98% true density) and elevated temperatures ( M O R = 120-205 M P a at 1300°C, 98% true density) [12]. These characteristics make spinel a widely used component of refractory materials [13] and the best candidate to replace chrome ore or chromia in refractories. In fact, the spinel-bonded castables have been well developed and widely used in the ladles and degassing furnaces of steel industries at temperatures over 1600°C [2, 3, 14-20]. However, castables with spinel bonding systems have not been successfully applied in industrial vessels with a service temperature of ~ 1300°C because this temperature is not high enough to activate solid-state sintering and develop sufficient strength. In order to increase the strength of the castables at this relatively low temperature, ultrafine spinel powders (<500nm) can be incorporated to improve sintering of the binding systems of the castables because these powders have high surface free energy and thus high driving force for sintering. The need for such ultrafine spinel-containing binding systems provides the rationale for the present work. Magnesium aluminate spinel is industrially produced from magnesia and alumina or Chapter 1 Introduction 3 from magnesite and bauxite [21] by fusion or sintering at high temperatures up to 1750°C. It is difficult to produce ultrafine and reactive spinel powders by milling from the aggregates produced by the fusion and sintering routes. For this reason, various nonconventional methods, such as coprecipitation [22, 23], sol-gel [10, 24], spray drying [25], spray pyrolysis [26], freeze drying [27, 28], flame spraying pyrolysis [29, 30] and mechanochemical techniques [31-33], have been used for preparing reactive spinel powders. In general, the spinel powders prepared by these methods showed high purity and small particle size, when compared with the spinel powders produced by the conventional solid-state reaction methods [34]. Spinel powders have been synthesized using coprecipitation and sol-gel routes. However, the main difficulty in preparing ultrafine spinel powders with the above-mentioned wet chemical methods lies in the fact that the precursor particles tend to agglomerate during drying and calcination. Severely agglomerated spinel powders are difficult to sinter, especially at the relatively low temperatures of 1300°C. In-situ spinel can be formed from ultrafine magnesia and alumina at the temperature of ~1300°C. However, the volume expansion (7.35%) accompanying the reaction makes the densification of the castables difficult. Therefore, the successful development of spinel-bonded castables lies, to a considerable extent, in the preparation of reactive spinel binding systems able to develop high-enough strength at the temperature of ~1300°C. There are no reports available on increasing the strength of the castables through incorporation of ultrafine powders when fired at the temperature. The binders for castables play a crucial role for the properties and performance of castables. Calcium aluminate cement has been commonly used as a binder in refractory Chapter 1 Introduction 4 castables. However, the presence of CaO is likely to produce relatively low-melting temperature phases, which unfavorably influence the resistance to slag attack. Consequently, ultralow cement and cementTfree castables have been developed to enhance the thermomechanical and corrosion resistance of the castables. In recent years, hydratable alumina has been used in castables [35-38] to replace calcium aluminate cement. However, the major concern with hydratable alumina-bonded castables is that they have relatively low strength after curing and drying (1.97 M P a modulus of rupture) [35] and after heat treatment at 800-1200°C ( M O R = 3.4 MPa) [36] when the hydraulic bonding is lost while the ceramic bonding is not yet significantly formed. It has been reported that the presence of reactive magnesia accelerates the hydration of hydratable alumina and a hydrotalcite-like compound forms when hydrated for 24 h at 20° and 30°C [39-40]. The formation of hydrotalcite-like hydrate may influence the strength of the castables after drying at 110°C and firing at 816°C. The hydration behavior of hydratable alumina in the presence of reactive magnesia has been investigated at room temperature [39]. However, in industrial practice, deadburnt magnesia and fused magnesia, rather than reactive magnesia, are normally used for production of refractory castables. The difference in reactivity and hydration between reactive magnesia and deadburnt or fused magnesia could be significant. Therefore, the mixtures of hydratable alumina and non-reactive magnesias may exhibit different hydration behavior from those of hydratable alumina and reactive magnesia. Moreover, in industrial practice, hydration of hydratable alumina in castables happens not only during curing at room temperatures, but also continues during drying/heating-up. Consequently, the hydration should be carried out at temperatures reflecting their curing and drying/heating-up. Chapter 1 Introduction 5 The bonding mechanisms and green strength of hydratable alumina-bonded castables could be altered by the presence of magnesia. In consequence, the strength development of the castables could be different from that of the hydratable alumina-bonded castables without magnesia. Such understanding is not yet available. Up to now, the sintering studies have been based on the models of uniform-sized spherical particles [41-42] and focused on narrow-sized powder systems [43-44]. However, the binding systems for refractory castables normally include powders with sizes from ~0.1 pm to -90 pm. The sintering behavior of such systems at temperatures of ~1300°C has not been tackled before. It is inferred from the above description that the development and application of hydratable alumina-bonded castables for use at ~1300°C depend on the availability of cost-efficient ultrafine spinel powders. The knowledge of the sintering behavior of the binding systems with wide particle size distributions, and the understanding of the bonding status in the castables after curing, drying and heat treatment from 800°C to about 1300°C is the key to developing the castables for use at ~1300°C. This work is an initial step to develop refractory castables for use at relatively low temperatures and has the following broad objectives: To develop novel methods for producing ultrafine spinel powders, with emphasis on avoiding hard agglomeration and decreasing the calcination temperature for phase-pure spinel formation. Chapter I Introduction 6 To gain understanding on sintering of powder compacts with a wide particle size distribution (0.1-90 pm) and investigate the effect of the ultrafine (<0.5 pm) powder addition on the sintering of such systems. The knowledge obtained is used to gain insight into strength development of the castables at relatively low temperatures of ~1300°C. To investigate the effect of ultrafine spinel powders on the properties of the castables when heat treated at ~1300°C and lower. Special attention is paid to the sintering and strength development of the castables with ultrafine powders, while optimizing the formulations of the castables to obtain high strength after firing at 800-1300°C. To gain understanding of the hydration behavior of the mixtures of hydratable alumina and various forms of magnesia and the strength development of the hydratable alumina-bonded castables at 110-800°C. The new knowledge is expected to throw light on the strength development of hydratable alumina-bonded castables after curing and heat treatment. Chapter 2 Literature Review 7 CHAPTER 2 LITERATURE REVIEW 2.1 Magnesium Aluminate Spinel The general formula of spinel is A B 2 O 4 . The unit cell formula of spinel is expressed as A8B16O32. The 32 oxygen ions form face-centered cubic lattice, providing 64 tetrahedral sites and 32 octahedral sites. In normal spinel, AB2O4, the divalent A cations are located in one-eighth of the tetrahedral sites and the trivalent B cations occupy one-half of the octahedral sites (Figure 2.1-1). In inverse spinel, B(AB)C>4, the divalent A cations and one-half of the trivalent B cations occupy the octahedral sites, with the remaining B cations occupying the tetrahedral sites. Various degrees of disorder intermediate between the normal and inverse spinel may also occur. For magnesium aluminate spinel, M g A l 2 0 4 , the M g 2 + cations fi l l tetrahedral sites and the A l 3 + cations reside in octahedral positions [45]. 0.4 0.6 Mole fraction A l 2 0 3 Cation in tetrahedral site Figure 2.1-1 Spinel structure Figure 2.1-2 M g O - A l 2 0 3 phase diagram Chapter 2 Literature Review 8 Magnesium aluminate spinel has a density of 3.56 g/cm 3 and a molar volume of 39.52 cm 3/mole. The formation of M g - A l spinel is accompanied by an increase in structural volume of 2.71 cm 3 /mol when formed directly from magnesia and alumina, which have molar volumes of 11.26 and 25.55 cm 3/mole respectively. This value represents a 7.35% volume increase, equivalent to a 2.45% linear expansion [46]. Magnesium aluminate spinel has high melting point (2135°C, see Figure 2.1-2), and high mechanical strength both at room temperature (135-216 M P a modulus of rupture, at 98% true density) and elevated temperatures (120-205 M P a modulus of rupture at 1300°C, 98% true density) [12]. Its thermal expansion coefficient (7 .6xl0" 6 o C"' , 20-1000°C) is lower than M g O (13.5xl0" 6 o C"' , 20-1000°C) and A 1 2 0 3 ( 8 . 8 x l 0 " 6 o C ' , 20-1000°C) [47]. As a result, spinel-based refractories have relatively high thermal shock resistance. Furthermore, magnesium aluminate spinel also has high resistance against chemical attack [10, 11]. Owing to the above properties, magnesium aluminate spinel has been widely used as a refractory material. 2.2 Spinel Synthesis Using Wet-Chemical Routes In the past three decades, various wet chemical techniques, such as coprecipitation [22] [23], sol-gel [10, 24], spray-drying [25], spray pyrolysis [26], freeze-drying [27, 28], and flame spray pyrolysis [29, 30] have been developed and used for producing oxide powders. The above wet chemical techniques are suitable for preparation of multi-component oxide systems such as M g A l 2 0 4 spinel. This derives from the ease of mixing solutions of salts in precise ratios and with homogeneity. After calcination, the precursors are converted to oxide powders, frequently preserving the homogeneity of the precursors. Chapter 2 Literature Review 9 2.2.1 Coprecipitation The most widely used chemical precipitation process is hydroxide precipitation, in which metal hydroxides are formed by changing the p H values. In order to increase the homogeneity of the mixed hydroxides, two ionic metals are converted to insoluble precipitants simultaneously through choosing the appropriate p H values. This process is referred to as coprecipitation. Among the solution methods, precipitation using inorganic salts is a relatively convenient and cost-effective way. But severe agglomeration commonly occurs in drying and calcination of the precursor. Agglomeration of the precursor during drying is mainly due to the hydrogen bonding between water and the hydroxyl groups on the surface of the precursor particles [48]. Free water molecules form bridges between the surface hydroxyl groups of neighboring particles through hydrogen bonding: The loss of the co-ordinated water molecules during drying brings the particles into close proximity and leads to soft agglomeration of the precursors. Hard agglomeration of the particles tends to occur during calcination of the precursors because a condensation reaction, as given by - M - O H + H O - M -—» - M - O - M - + H2O [48], happens between the surfaces of the precursor particles. The agglomeration can be reduced by substitution of water with alcohol in the precursor suspension because alcohol forms weaker hydrogen bonding with the surface hydroxyls of the precursor particles than water [49]. In order to reduce agglomeration of precursors synthesized by wet chemical routes, combustible ingredients can be introduced into the precursors. The combustible ingredients would act as physical barriers between the particles (Figure 1.1-1), so that "soft Chapter 2 Literature Review 10 agglomeration" of the particles can be weakened during drying and accordingly "hard agglomeration" can be reduced during calcination. O Q o o ° o Slurry After drying After calcination Figure 2.2-1 Schematic diagram for preventing agglomeration by combustible additives L i and co-workers [13] used ammonium carbonate as precipitant to synthesize spinel precursors from a mixed solution of magnesium and aluminum nitrates. The as-synthesized precursor was identified as a mixture of N H 4 A 1 ( 0 H ) 2 C 0 3 H 2 0 and M g 6 A l 2 C 0 3 ( O H ) i 6 - 4 H 2 0 . The presence of carbonate groups in the precursor reduced the hydrogen bonding between water molecules and the hydroxyls on the surface of the precursor and, as a result, reduced agglomeration of the synthesized powder. The precursor was loosely agglomerated after drying and ultrafine spinel powders (<100 nm) were obtained after the precursors were calcined in the temperature range of 900-1300°C. For the synthesis of magnesium aluminate spinel by coprecipitation, the p H value has to be carefully controlled. Normally aluminum is completely precipitated as hydroxide at pH = 6.5-7.5 and magnesium at p H = 12 [22]. In order for the precipitation of both ions to take place simultaneously, an intermediate pH range should be used. The stoichiometric spinel was achieved at p H = 9 [22]. (O: precursor particle; O combustible additive) Chapter 2 Literature Review 11 2.2.2 Sol-Gel Processing The sol-gel processing includes hydrolysis of metal alkoxides or inorganic salts and then gelation of the colloid dispersion (sol) by dehydration or p H control. This technique has been used to prepare stoichiometric and high-purity spinel. The characteristics of spinel precursors obtained by this process vary with the starting materials and processing parameters. Paulick et al [50] used A l ( O C 3 H 7 ) 3 and Mg(OC2H 5 ) 2 as starting materials and obtained an amorphous ultrafine powder after calcination at 500°C. The surface area of the calcined powder was 260m 2/g, and the particle size was in the order of 10 nm. After heat treatment at 1000°C for one hour, the average particle size of the spinel powder increased to about 30 nm. Shiono et al [10] synthesized spinel precursors using a heterogeneous sol-gel method, from aluminum isopropoxide and fine M g O powders. The average particle size of the M g O powder was 10 nm. The obtained precursor was identified as a mixture of boehmite (AlOOH)) and mixed hydroxide ( M g 4 A l 2 ( O H ) i 4 - 3 H 2 0 ) and M g O was not observed in the precursor composition. Phase-pure spinel was not obtained after calcination at 1200°C. After calcination at 1200°C, the powder had a specific surface area of 40m7g. Yamaguchi et al [51] synthesized spinel precursors from the hydrolysis of magnesium and aluminium alkoxides at 90°C. The precursors were mixtures of M g 4 A l 2 ( O H ) 1 4 - 3 H 2 0 and AIO(OH) , similar to the composition of the precursor synthesized by Shiono and co-workers using a heterogeneous sol-gel route [10]. The drying operation of the precursor suspensions plays a crucial role in the agglomeration of the final powder products. Drying by evaporation under normal conditions gives rise to capillary pressure, which causes the shrinkage of the gel network [52]. The Chapter 2 Literature Review 12 resulting gel, called a xerogel, is often reduced in volume by a factor of 5 to 10 compared to the original wet gel. If the wet gel is placed in an autoclave and dried under supercritical conditions, there is no interface between liquid and vapor, hence no capillary pressure, so the obtained gel (called an aerogel) incurs relatively little shrinkage. 2.2.3 Other Methods The spray-drying technique is a common approach to preparing ceramic powders. Montouillout et al [53] prepared MgAl204 precursor via spray-drying of a solution of magnesium and aluminum nitrates, which usually yields hollow spheres. After heat treatment at 800°C, the mean particle size ranges from 1 to 2 pm. The specific surface area of the synthesized M g A l 2 0 4 spinel powder varies with the thermal treatment, decreasing from ~200m 2/g at 400°-500°C to 62 m 2 /g after 1 h at 800°C. A significant development in the spray drying field is the movement toward the high temperature process (evaporative decomposition or spray pyrolysis), where drying and decomposition take place in a single step. This process has been developed to the point of large-scale industrial applications [34]. Suyama and Kato [26] prepared M g A l 2 0 4 spinel powder using this method. In contrast to the vaporization methods, freeze-drying utilizes sublimation rather than evaporation. The solution is dispersed into small droplets and freeze-dried to prevent segregation. The solvent is removed by sublimation at low pressure [54]. Freeze drying is quite successful in producing fine, homogeneous ceramic powders on a small scale, but has not been used in large commercial applications because of its high capital cost and energy inefficiency. Tang and his coworker [27] used alkoxides and Nakagawa [28] used sulfates as the starting materials. The primary particle size of spinel powder calcined at 1100°C for 12h was about 50nm. Chapter 2 Literature Review 13 Flame spray pyrolysis, which uses polymers as the precursors, begins by injecting the precursor into a combustion chamber via an aerosol generator, where the individual droplets are rapidly combusted, resulting in crystalline, homogeneous, ultrafine powders. Bickmore et al [30] prepared crystalline spinel powder using the alcoholic solution of a double alkoxide. The prepared powder had a specific surface area of 40-60 m /g and an average crystallite size of 25-45 nm (from specific surface data and X R D data). However, it is difficult to control the particle size into a narrow range. The powder prepared by Bickmore [30] had a wide particle size distribution, with the maximum size of about 3 pm. 2.3 Spinel Synthesis via Mechanical Activation 2.3.1 Advantages of Mechanochemical Processing Solid-phase reactions are activated by high temperature treatment. However, the efficiency of the solid-phase reactions is rather low because of the low diffusion rate through the product layer, no tight contacts between the particles of the components, and non-uniform particle-size distribution. It is proved that mechanical activation of the initial mixtures using high-energy mill ing can greatly promote solid-state reactions [55]. Mechanochemical synthesis is mainly based on the mechanical activation of the starting materials. During mill ing, the reaction area increases with the decrease of the particle size and the crystallite size of the reactants. Besides, the structure of the starting materials could be highly distorted. The structure distortions include generation of free radicals and unsaturated groups, and the formation of an amorphous phase. Consequently, the activation energy for atomic diffusion is decreased (Figure 2.3-1). Chapter 2 Literature Review 14 mmm •§--• mmm• mmm * * * (a) (c) (a) (b) After milling (c) Figure 2.3-1 Change in activation energy before and after mill ing [56] Furthermore, the high impact pressure at the collision points may also play an important role in facilitating the solid-state reactions [57]. It was estimated that the pressure generated by high-energy mil l ing could reach 6 GPa [58]. This high pressure may substantially increase the temperature at the collision point, cause the ingredients to mix intimately on an atomic scale and embed the ingredients into each other, increasing the homogeneity of the starting materials. Until now, the mechanochemical technique has been used only at the laboratory scale, generating only several grams of ultrafine powder per milling. However, recent studies have shown that the mechanochemical technique can readily be scaled up for mass production through the use of attritor mills [48]. 2.3.2 Structural Evolution of Mechanically Activated Materials Crystalline materials can be transformed into amorphous by high-energy milling. The Chapter 2 Literature Review - 15 ease of amorphization during mill ing is different for different starting materials: boehmite became completely amorphous, while diaspore remained crystalline, after both of them were milled for 24 h [59]. Materials can be changed from one crystalline form into another through mechanical activation. For example, gibbsite was transformed completely to boehmite after 40 h of mechanical mill ing in a nitrogen atmosphere [60]. Mechanical mill ing can change the dehydroxylation behavior of hydroxides. The dehydroxylation of kaolinite shifted to lower temperatures as the grinding time increased [59]. Gibbsite subjected to mechanical mill ing displayed a sharp weight loss of - 1 3 % under 100°C, which resulted from the removal of its structural water and its conversion into boehmite. The "free" water generated by the dehydroxylation of gibbsite during the high-energy mill ing was adsorbed onto the nanocrystalline boehmite formed [60]. New bonds can be formed during high-energy mill ing, For example, by grinding a mixture of kaolinite and boehmite, some of the Si -O-Si and A l - O - A l bonds were destroyed and new S i - O - A l bonds were created [61]. The formation of S i - O - A l bonds during milling facilitated the formation of mullite during the succeeding heat treatment. 2.3.3 Mechanical Activation Assisted Synthesis In some systems, in-situ formation of compounds can be realized during high-energy milling. Nanocrystalline perovskite Pb(Zn 1 / 3 ^ 2 / 3 ) 0 3 was formed by grinding a mixture of PbO, ZnO and Nb203 in a shaker mi l l for 20 h [62]. Nanosized lead titanate powders were prepared by mil l ing of PbO and T i 0 2 in a shaker mi l l for 15-20 h [63]. Hydroxyapatite was formed by high-energy mechanical activation of a dry powder mixture of calcium oxide (CaO) and anhydrous calcium hydrogen phosphate ( C a H P O ^ [64]. Chapter 2 Literature Review 16 In other systems, the in-situ formation of compounds does not take place just by milling, or is not economical, so subsequent calcination has to be carried out to complete the chemical reaction. The calcination temperatures are generally lower than those for the reaction without the mechanical activation of the same starting materials. For example, Mulli te was synthesized by grinding a mixture of gibbsite and silica gel followed by heat treatment at 1200°C [65]. 2.4 Sintering in the Presence of Ultrafine Powders 2.4.1 The Driving Force for Sintering Figure 2.4-1 shows the microstructural development of the solid-state sintering process of a powder compact with a uniform grain size. For most ceramic materials, the green powder compact consists of particles with point contacts (Figure 2.4-la). These contacts between particles grow in size during sintering (Figure 2.4-lb and Figure 2.4-lc), usually accompanied by grain growth and pore shrinkage. From the thermodynamic/macroscopic perspective, the driving force for sintering is the reduction of the surface energy inherent in powder compacts [67], through mass transport from the surface and grain boundary into the neck area of adjacent particles (Figure 2.4-2). Consequently, the neck growth is the basic and key phenomenon of the solid-state sintering process. In fact, the sintering kinetics has been established essentially based on the neck growth of adjacent particles. Chapter 2 Literature Review Figure 2.4-1 Schematic microstructural development during solid-state sintering [66]. Figure 2.4-2 Scanning electron micrograph of sintering bonding between particles [68]. Chapter 2 Literature Review 18 For solid-state sintering, material transport usually takes place through diffusion. The paths for solid-state sintering include surface diffusion, boundary diffusion, lattice diffusion, and vapor evaporation-deposition. Mass always transfers from either the surface or grain boundary to the neck area for al l the four diffusion paths [41, 42] and vapor-phase mass transport. For the vapor-phase mass transport, the material must be heated to a temperature sufficiently high for the vapor pressure to be appreciable [41]. The vacancy concentration on a neck concave surface is greater than that on a flat surface, which in turn is greater than that on a convex surface. Therefore the difference in vacancy concentration between the particle surface and the neck area is the driving force to transport the materials from surface or grain boundary to the neck [41]. 2.4.2 Driving Force versus Particle Size As described before, the solid-state sintering driving force is essentially the tendency to reduce surface energy in the powder compacts. The energy difference of an atom between a sphere surface and a flat surface is represented by the following relation [67, 69]: r Where AGex is the excessive energy per atom in the spherical particle, y the surface energy, £2. the volume of the atom and r the radius of the spherical particle. Equation 2.4-1 indicates that the excessive energy per atom on the surface of spherical particles is inversely proportional to the particle size and the difference in vacancy concentration between the neck area and particle surface is also inversely proportional to the particle size [41], demonstrating that the driving force for mass transport is increased when particle size is decreased. That is why ultrafine powders have been used in order to promote faster sintering. Chapter 2 Literature Review 19 2.4.3 Approaches to Modeling Sintering The kinetics of initial stage sintering is studied under isothermal conditions using the following relationship [70]: V L o J B t (2.4-2) 2nD Where L0 is the initial length of the sample, AL is the dimensional change of the sample, D is the diameter of the particles, n and m are parameters related to mass transfer mechanisms, t is time, and B is a parameter exponentially dependent on temperature: B = B0cxp(--j^j (2.4-3) where B0 is a collection of material parameters (such as surface energy, atomic size, and the geometry of the powder compacts), Q activation energy, R gas constant, and T temperature. Under isothermal conditions of sintering, B can be obtained for a plot of l n - ^ versus lnf, according to Equation 2.4-2. Using the B values of different temperatures, the sintering activation energy can be derived from Equation 2.4-3, by plotting In B aga ins t^ , . Sintering studies under isothermal conditions have provided most of the available sintering data [71]. However, isothermal experiments present several problems because considerable development and densification can occur during the heating-up period [72], Firstly, under isothermal conditions, it is difficult to monitor the initial portion of the sintering [73]. Secondly, large samples cannot be subjected to very high heating rate in order to minimize the heating-up period. Therefore, in order to circumvent the problems in isothermal sintering, constant-rate-of-heating (CRH) technique has been used to study sintering. Chapter 2 Literature Review 20 Apparent activation energy in the initial and intermediate stages of sintering can be estimated using the following relationship [72-74]: (2.4-4) RT, Where AL = the change in length of specimen (mm) L0 = the initial length of the specimen (mm) T = Temperature ( K ) A = a constant depending on mass transfer mechanisms and the material parameter R = the gas constant = 8.3144 J/(mole • K ) Q = the apparent activation energy (J/mole) The apparent activation energy can be determined by plotting of the left-hand side of Equation 2.4-4 versus 1/T for each heating rate. 2.4.4 Sintering Behavior of Multi-Modal Systems The effect of particle size on sintering of powder compacts is well substantiated both quantitatively and empirically [75]. As seen from Equation 2.4-1, the driving force for sintering and grain growth increases inversely with the size of the powders. Also, as indicated by Equation 2.4-2, the sintering rate would increase when the size of the powders decreases. Accordingly, mixes composed of finer powders can achieve higher density than those of coarser powders at a certain sintering temperature; or mixes composed of finer powders can acquire the same density at a lower sintering temperature as those composed of coarser powders at a higher sintering temperature. Chapter 2 Literature Review 21 However, Equations 2.4-1, and 2.4-2 are based on the models of uniform-sized spherical particles. In reality, powder compacts are normally composed of particles with a certain size distribution. Then, the sintering and densification of the multi-sized compacts may not completely follow the sintering kinetics for the compacts consisting of single-sized spherical particles. The variation in sintering behavior between uniform-sized and multi-sized systems arises mainly from the difference in their packing geometry [75]. Theoretically, shrinkage kinetics of powder compacts consisting of uniform particles is independent of the particle arrangement or packing [75]. However, in multi-sized systems rich in large particles, the shrinkage during sintering is governed primarily by the large particles which may form a skeleton of the compact. Packing of mono-sized spherical powders yields a relatively low density with a fractional density of 0.64 in a random arrangement [76]. Mixtures of powders with different sizes give improved packing densities because the small particles can fit into the interstices between larger particles. High bulk density achieved in fabrication reduces shrinkage to achieve a desired density, so that dimensional specifications can be more easily met [77]. For effective filling of small particles in the interstices formed by the large particles in binary [78] or multi-sized [79] powder mixtures, the particle size ratio should be at least 7. O'Hara and Cutler [75] used alumina powders of 3 to 6 pm in diameter as the fine portion and tens of microns in diameter (with a maximum average particle size of 66 pm) as the coarse portion of the mixtures. The isothermal shrinkage was performed at 1500°C and 1550°C. The results showed that the shrinkage rate decreased with increase in volume fraction of the coarse particles in the bimodal compacts. Chapter 2 Literature Review 22 It was also found that, at 1500°C, the alumina with a particle size of 1-3 pm sintered by grain boundary diffusion and the alumina with a particle size of 30.5-33.5 pm, sintered by volume diffusion [75]. The results indicate that mass transport mechanisms may change at the same temperature, depending on the size of the powders. Smith and Messing [71] prepared mixtures of fine-grained alumina powders having a median particle size of <0.5pm and a coarse-grained alumina powder having a median particle size of 4 to 5 pm. When the fine volume fraction (V f ) ranged from 0 to 0.3, the sintering of the coarse component controlled the shrinkage of the bimodal mixture, since the fine particles occupied the void spaces between the coarse particles. When sintered, the fines densified first, then the coarse particles began to sinter and densify. In the range between 0.0 and 0.2 of the fine volume fraction, the shrinkage was independent of composition for each sintering temperature. This confirmed that the coarse fraction controlled densification for these compositions. For 0.3 < Vf < 1.0, the large particles are dispersed in a matrix of fine particles. When fired, the fine particles began to sinter and the compact densified until completely dense or until the coarse particles come into contact, at which time further densification was controlled or inhibited by the coarse component. Chaim and Gedanken [43] used nanocrystalline alumina (75 nm) and conventional alumina (0.86 pm median) to prepare bimodal mixtures. The maximum density was found to be at 70% coarse particles composition for the green samples and at firing temperatures between 1300°C and 1600°C. The authors calculated the sintering activation energy using the constant-rate-of-heating shrinkage data and, based on the data, concluded that addition of the nanocrystalline powder component enhanced the surface and grain boundary diffusion mechanisms during the sintering of the bimodal powders. Chapter 2 Literature Review 23 Yeh and Sacks [44] used two powders with the same median size of -0.4 um; one had a broad particle size distribution (BSD) (0.08-1.5 pm) and the other had a narrow particle size distribution (NSD) (0.2-0.7 pm). The B S D sample had a higher green density (-73% of the theoretical density) and the N S D had a lower green density (-65% of the theoretical density); besides, the former sample had smaller-sized pores (-29 nm), while the latter had larger-sized pores (-50 nm). After firing at 1260°C for 12 h, Both B S D and N S D samples had the same density (-96%). The relative densities for the both samples remained the same with longer sintering time at 1260°C or at higher sintering temperature (1340°C). Because B S D samples had higher green density than the N S D samples, less shrinkage occurred in the B S D compacts than in the N S D compacts during sintering. It is accepted that regions with the highest packing density in a powder compact undergo the most rapid shrinkage during sintering. Besides, grain growth occurs more readily in these regions. Moreover, the B S D green compacts had a smaller median pore channel radius (-29 nm) compared to the green compacts with a narrow particle size distribution (-50 nm median pore radius). Therefore, the mean curvature in the neck region between particles and, consequently, the average driving force for sintering in the B S D samples were probably higher than in the N S D samples during the early stage of sintering. The uniform sized powder compacts densify uniformly (see Figure 2.4-3a) and, accordingly, no sintering stress occurs between particles. However, sintering stress occurs in the bimodal powder (two-sized) compacts [76, 80] in two modes. On one hand, the addition of small particles to the large particle network (large particles form a skeletal structure and small grains f i l l the interstices formed by adjoining large grains) can provide a compressive stress to the large particles that increases sintering, though the small particles bonded to the Chapter 2 Literature Review 24 neighboring large particles exert relatively little influence on the overall densification rate set by the large particle skeleton. On the other hand, in binary mixtures where large particles are separated by small grains, the densification of small particle network is impeded by the addition of large particles. The small particle shrinkage is constrained by the full-density large particles. This may result in a localized stress that reduces shrinkage and opens cracks in the matrix (see Figure 2.4-3b). Figure 2.4-3. Densification behavior around large particles [81] 2.4.5 Sintering of Refractory Binding Systems For most sintering studies on ceramics, the ultimate objective is generally to obtain products with highest-possible density and uniform-grained microstructure at relatively low temperature [44]. As described before, the use of mixtures of different particle sizes increases the green bulk density of powder compacts and consequently reduces shrinkage during firing [77]. Chapter 2 Literature Review 25 In order to acquire high density and decrease shrinkage during firing service, refractory castables are composed of particles with a wide particle size distribution (0.1pm - 5000pm or even larger) and so are the binding systems of the castable (0.1-90pm). Mixtures of powders with differing sizes give improved packing densities [76]. The large particles form the skeleton of the compacts and the smaller particles f i l l the interstices of the adjoining large particles (see Figure 2.4-4). The particle size composition of refractory materials falls in the region of 60-70% large-sized aggregates and 30-40% small-sized powders. Therefore, the sintering of bimodal (or two-sized or two-component) and multimodal mixtures is of special interest to refractory materials. Sintering of refractories mainly depends on the matrixes consisting of different-sized powders. Thus, particular attention of this work is paid to the sintering of the matrixes (also referred as to binding systems for castable). Figure 2.4-4 Schematic representation of particle packing in refractory material. Chapter 2 Literature Review 26 In recent years, technical advances in refractories have been in a stagnation period [82]. W i l l application of nanotechnology bring a breakthrough development in refractories? The answer is not available yet. However, it has been reported that introduction of nanoparticles is beneficial to refractories in two aspects. Firstly, the amounts of certain additives for refractories can be greatly reduced i f the additives are nano-sized. For example, the total content of carbon in M g O - C brick was reduced from 15-20% to 3-5%, including the carbon from the binder [83]. The incorporation of less carbon in M g O - C bricks wi l l potentially solve problems such as energy loss and the heat load to the vessel. Secondly, nano-powders facilitate sintering of the matrix of refractory ceramics because the powders have high surface free energy and driving force for sintering is the surface energy. Therefore, ultrafine powders potentially allow for sintering of castable matrixes at relatively low temperatures. It is reported [84] that addition of 1-2% nano-alumina (with average particle size of 100 nm) decreased the sintering temperature of corundum refractories by 100°C, in the temperature ranged between 1450 and 1650°C. The addition of 1% nano-silica (with average particle size of 12nm) also decreased sintering temperature of corundum refractories by 100°C, when sintering at 1450-1650°C. The addition of 2% nano-Fe2C>3 decreased the sintering temperature of magnesia-chrome refractory by 150°C, for the same range of sintering temperatures [85]. 2.5 Hydratable Alumina-Bonded Castables 2.5.1. Binder Systems for Castables Calcium aluminate cement has been widely used as a binder for castable refractories. The phases in calcium aluminate cement consist of mainly C A and C A 2 with minor amounts Chapter 2 Literature Review 27 of C 1 2 A 7 (C=CaO, A = A 1 2 0 3 , H = H 2 0 ) . These phases react with water to form hydrates C A H -10, C 2 A H 8 , C 3 A H 6 and A H 3 [86]. The hydrated phases impart the room-temperature strength to the castables containing this cement. However, the calcium oxide contained in the calcium aluminate cement deteriorates the properties of the castable refractories because CaO tends to form low-melting compounds, such as gehlenite ( 2 C a O A l 2 0 3 S i 0 2 , melting point (mp) = 1590°C), anorthite ( C a O A l 2 0 3 - 2 S i 0 2 , mp = 1550°C) [35], monticellite ( C a O M g O S i 0 2 , mp=1498°C), and merwinite ( 3 C a O M g O - 2 S i 0 2 , mp = 1577°C) [87]. These phases wi l l decrease the resistance of the refractory linings to melt and slag attack. It is proved that, by decreasing the contents of calcium aluminate cement, the hot strength and resistance to slag attack of the castables was increased [88]. When hydratable alumina is used as a binder for castables, the incorporation of CaO would be avoided. Hydratable alumina reacts with water to form a hydraulic bonding [35]. However, a major challenge in using hydratable alumina in castables is the fact that the binder wi l l lose its strength at intermediate temperatures (800-1200°) , similar to calcium aluminate cement. The hydratable alumina is generally used in cement-free castables together with > 3% silica fume (also referred to as microsilica) [35]. The addition of silica fume not only increases the flowability of the alumina-based castables (see Figure 2.5-1), but also enhances their hot modulus of rupture ( H M O R ) starting with 1200°C because of mullite formation in the matrix (see Figure 2.5-2). However, as described in Chapter 1, hydratable alumina-bonded castables without silica fume have relatively low strength after curing and Chapter 2 Literature Review 28 drying (MOR=1.97 MPa) [35] and after heat treatment at dehydration temperatures (800-1200°C) (MOR=3.4 MPa) [36]. Flow (%) 120 , 80 • 0% Mierosilica — 1% Mierosil ica — 5% Mierosil ica 0 10 20 30 40 50 60 70 Time (minutes) Water addition - 5.6% Figure 2.5-1 Effect of silica fume addition on flowability of cement-free castables [35]. H M O R (MPa) 1,200 1,500 Firing temperature (°C) Tested by British Steel Brilsh Standard t S I M O , Section 4.5 Figure 2.5-2 Effect of silica fume addition on H M O R of cement-free castables [35] The castables bonded by hydratable alumina are more vulnerable to explosion spalling during drying and initial heat-up than the castables bonded by calcium aluminate Chapter 2 Literature Review 29 cements because the alumina gel formed from hydratable alumina lowers the permeability and porosity of the castables [37, 38, 60]. It has been proposed that the formation of hydrotalcite hydrate would possibly facilitate drying and heat-up of the castables because the dehydration of the hydrotalcite-like compound would generate a micro-porous network in the bonding systems [37] and the vapor can escape through the pore channels during heating-up. Sil ica fume and magnesia powder can act as binders for refractory castables. Magnesium silicate hydrates can form from the silica fume-magnesia-water system and the formation of - M g - O - S i - O - M g - chain provides the bonding strength [89]. Si l ica fume can also reduce the hydration of magnesia powder during mixing and installation of the castables [90]. It is assumed that the adsorption of silica fume on the surface of M g O particles and the formation of - M g - O - S i - chain on the surface of the M g O particles decreases the degree of M g O hydration [89]. Sil ica fume is an amorphous Si02 consisting of spherical particles with an average diameter of -0.1 pm. In alumina and magnesia-based refractories, silica fume is highly reactive and can form a ceramic bonding (forming mullite 3Ai203-2Si02 and forsterite 2MgO-Si02) at reduced firing temperatures of 1200°C, resulting in high hot strength at the temperature (Figure 2.5-2) [91]. Si l ica fume can f i l l the voids between the coarse particles, releasing the entrapped water and leading to high flowability (Figure 2.5-1). However, silica fume is not always desired for refractories because it may form low melting compounds with calcium aluminate cement during service, as mentioned above, thus lowering the high-temperature strength and the resistance to slag attack. A s a result, silica Chapter 2 Literature Review 30 fume in the binding systems of the castables for non-ferrous furnaces should be avoided. 2.5.2 Hydration of Hydratable Alumina The hydration behavior of transition alumina ( p - A l 2 0 3 , commercially called hydratable alumina) has been investigated [92-94]. Transition alumina first reacts with water to yield orderly pseudoboehmite gel and then transforms to large amounts of bayerite and traces of boehmite [93, 94]. However, crystalline bayerite formation strongly depends on the hydration temperature. Bayerite was not observed when hydration was performed at 5°C for 8 days; and bayerite accounted for 48 and 57% after transition aluminas were hydrated for 8 days at 25°C and 50°C respectively [93]. It was also observed that transition alumina was not completely hydrated in the temperature range of 15° and 55°C for 24 h [92]. Hydratable alumina after hydration in the castables has the same hydration behavior and yields the same hydration products as the hydratable alumina hydrated alone, when the castables are composed of alumina [36-38]. However, the presence of reactive magnesia accelerates the hydration of hydratable alumina and the hydration products include a hydrotalcite-like compound [39]. Hydratable alumina is, with the presence of reactive magnesia, almost completely consumed to form the hydrotalcite hydrate when hydrated for 24 h at 20° and 30°C [39]. Formation of a hydrotalcite hydrate from hydratable alumina during hydration with the presence of magnesia is beneficial to castables in two aspects. In the first place, formation of hydrotalcite-like hydrate in castables bonded by hydratable alumina increases the strength of the castables after drying at 110°C and firing at 816°C [95]. Secondly, formation of the hydrotalcite hydrate could make the heat-up of castables safer. It has been reported [36, 38] that the castables bonded by hydratable aluminas are more liable to explosion spalling during Chapter 2 Literature Review 31 drying and initial heat-up than the castables bonded by calcium aluminate cements because alumina gel formed from hydratable alumina lowers the permeability and porosity of the castables. Formation of hydrotalcite hydrate would facilitate drying and heat-up of the castables because the dehydration of the hydrotalcite-like compound would generate a micro-porous network in the bonding systems [39], and this porous network would provide channels for water vapor escape during heating-up. 2.5.3 Ultrafine Powers for Castables As described above, it is imperative to prepare reactive binding systems for castables without cement and silica fume for applications at relatively low temperatures. Only highly reactive binding systems have the potential capability of providing the castables with high-enough strength after heat treatment at temperatures ranging from 800°C to 1300°C. If ultrafine magnesia and alumina powders are added to basic castables to form in-situ spinel in service, there are three potential difficulties to overcome. Firstly, it is very difficult to prevent hydration of magnesia during mixing, casting and drying. The hydration is accompanied by volume expansion (120%), which wi l l damage the strength of the castable. Secondly, the formation of spinel from magnesia and alumina wi l l bring about volume expansion (7%), which wi l l generate cracks and damage strength. Lastly, it is very difficult for the spinel matrix to develop in these castables, because firing at 1200-1300°C wil l not allow for the complete spinel formation from magnesia and alumina. In order to promote the sintering reactivity of binding systems, ultrafine spinel powders are desired to promote sintering of the matrix of the castables at relatively low temperatures. Chapter 3 Scope and Objectives 32 CHAPTER 3 SCOPE AND OBJECTIVES 3.1 Scope Hydratable alumina-bonded castables with spinel matrix are in increasing demand to replace chromia-containing bricks for non-ferrous furnaces. However, it is difficult for castables to develop high-enough strength (MOR>12 MPa) at the service temperature of ~1300°C. Other major challenges include the low strength of hydratable alumina-bonded castables at room temperature (MOR=1.97 M P a [35]) and at temperatures up to 1000°C (MOR=3.4 M P a [36]). This work is an attempt to develop refractory castables comparable with the high-temperature fired basic bricks and is focused on improving the strength of hydratable alumina-bonded castables, through incorporation of ultrafine powers, after firing at temperatures up to 1300°C. 3.2 Objectives The goal of the first part of this work is to develop novel approaches for preparing ultrafine and reactive magnesium aluminate spinel powders. The specific objectives include: 1. Investigate the influence of combustible additives on the deagglomeration of spinel precursors prepared by coprecipitation. 2. Investigate and develop a novel method combining sol-gel process with precipitation process to prepare ultrafine spinel powder. 3. Investigate the effect of mechanical activation on synthesis of the spinel powders to achieve complete spinellization at relatively low calcination temperature. Chapter 3 Scope and Objectives 33 The goal of the second part of this work is to study the sintering of binding systems with a wide particle size distribution and the effect of ultrafine powders on sintering. The following are the specific objectives: 1. Investigate the sintering behavior of binding systems containing the powder mixtures with bi-modal and tri-modal particle compositions. 2. Investigate the influence of ultrafine powders on the sintering behavior and strength development of binding systems with a particle wide size distribution (from ~0.1 p:m to -90 u.m) at relatively low temperatures, with emphasis on the temperatures below 1300°C. The goal of third part of this work is to develop basic castables with hydratable alumina as the hydraulic binder and ultrafine spinel powders as sintering enhancing additives. The specific objectives include: 1. Investigate the effect of ultrafine powders of various sizes on the castable properties fired below 1300°C. 2. Investigate the hydration behavior and hydration products of hydratable alumina in the presence of various forms of magnesia. 3. Investigate the relationship between the hydration products and strength development of hydratable alumina-bonded castables heat treated between up to 816°C. Chapter 4 Methodology 34 CHAPTER 4 METHODOLOGY 4.1 Spinel Preparation and Characterization Procedures 4.1.1 Preparation of Spinel by Coprecipitation A magnesium aluminate spinel precursor was prepared via co-precipitation, using magnesium nitrate hexahydrate [ M g ( N 0 3 ) 2 - 6 H 2 0 ] and aluminum nitrate nonahydrate [A1(N0 3 ) 3 -9H 2 0] as reactants. Ammonium hydroxide [ N H 4 O H ] (28%, analytical grade) was used as a precipitant. The reactants and the precipitant were of chemical purity and supplied by Fisher Scientific. The magnesium and aluminum nitrates were dissolved in 500 ml distilled water to obtain 0.05M and 0 .1M concentrations respectively, at pH=3. Nitric acid was used to decrease the pH to 1.5 and the as-obtained solution was heated and maintained at 80°C under magnetic stirring during the drop-wise addition of ammonium hydroxide, until a 9.5 p H was achieved. The resultant suspension was kept at 80°C and p H = 9.5 for half an hour, for the reaction to complete, then it was cooled to room temperature, filtered using suction filtration, and washed three times with distilled water. Three samples were prepared to investigate the influence of combustible ingredients on deagglomeration of the spinel powders: The filtered paste was dried in air at 105°C for 24 hours. The obtained sample is referred to as M A - 1 (all samples are listed in Appendix A ) . The filtered paste was dispersed in ethyl alcohol (20 wt% solid load in proportion to the final M g A l 2 0 4 spinel product) under stirring and was divided into two portions. Carbon black (AL-2155, Anachemia) with a median particle size of -50 nm was added into one portion of the suspension until a molar ratio metalxarbon of 1:3 was reached and the Chapter 4 Methodology 35 suspension was stirred for one more hour, then filtrated and dried at 105°C for 24 h. The obtained sample is referred to as M A - 2 . A solution of 10% polyvinyl alcohol ( P V A ) (w/v = weight of P V A in gram/volume of water in cubic centimeter) was prepared using P V A (average molecule weight = 13000-23000, Aldrich) and distilled water, was added into the other portion of the suspension under stirring, as described above. The metal ions :PVA monomer molar ratio was 1:3. The mixed solution was heated at about 70°C to evaporate water and alcohol under stirring and then dried at 105°C for 24 h. The obtained sample is referred to as M A - 3 . 4.1.2 Preparation of Spinel Precursors by Heterogeneous Sol-Gel Process Using a heterogeneous sol-gel process, magnesium aluminate spinel (MA-5) was synthesized from aluminum isopropoxide (Sigma, Cat. N o A4399) and reactive magnesia powder (dso = 3.09 pm, Fisher Scientific), and spinel-coated magnesia ( M A - 6 ) was prepared from aluminum isopropoxide and magnesia powder (<45 pm, Dynamag K , Washington Mi l l s ) . Aluminum isopropoxide was dissolved in isopropyl alcohol and refluxed for 1 h. The magnesium oxide powder was then added and the solution was refluxed for 5 h to complete the reaction, followed by drop-wise addition of distilled water under continuous magnetic stirring. After the water addition, the refluxing was continued for 30 minutes and then the suspension was dried for 12 h at 50°C to remove water and alcohol, followed by 12 h in the drying oven at 80°C. The fluxing time after the addition of light M g O was increased from 2 h (MA-5) to 5 h ( M A - 5 A ) to determine the influence of the refluxing time on the phases of the precursors. The parameters for preparing the precursors M A - 5 and M A - 6 are listed in Table 4.1-1. Chapter 4 Methodology 36 Table 4.1-1 Process parameters for heterogeneous sol-gel and preparation of spinel precursors Sample code MA-5 MA-5A MA-6 A l ( 0 - i - C 3 H 7 ) 3 , mol 0.1 0.1 0.1 i - C 3 H 7 O H , mol 1.5 1.5 1.5 M g O , mol 0.05 (Reactive) 0.05 (Reactive) 0.10 (Magnesia) H 2 0 , mol 2.0 2.0 2.0 Refluxing time, h Dissolve aluminum salt 1 1 1 After M g O addition 2 5 2 After water addition 0.5 0.5 0.5 Molar ratio of M g O to A 1 2 0 3 1:1 1:1 2:1 4.1.3 Preparation of Spinel Precursors by Mechanical Activation For the synthesis of M g A l 2 0 4 spinel with the assistance of mechanical activation, magnesium hydroxide (Mg(OH) 2 , brucite, Fisher Scientific), boehmite ( A l O O H , S O L - P 2 K , Condea, Germany) and aluminum tri-hydroxide (amorphous A l ( O H ) 3 , Cat. No. 23918-6, Aldrich) were used as the starting materials. The mixture of M g ( O H ) 2 and A l O O H is denoted as M A - 2 1 and the mixture of M g ( O H ) 2 and A l ( O H ) 3 is referred to as M A - 2 2 . Both mixtures had an M g O / A l 2 0 3 molar ratio of 1:1. The experimental mixtures were mixed in ethanol for 48 h in a ball mi l l , using zirconia balls of 5 mm in diameter as the mill ing media, followed by drying for 24 h at 80°C. The dried powders were ground using an agate mortar and pestle and then passed through a 40-mesh sieve. The mechanochemical activation was carried out in 5 gram batches using a Spex 8000 mixer/mill in a cylindrical vial (40 mm in diameter and 40 mm in length), with one stainless steel ball (12.7 mm in diameter). Chapter 4 Methodology 37 4.1.4 Mechanical Activation of Heterogeneous Sol-Gel Precursors The starting materials were aluminum isopropoxide (Sigma), reactive M g O powder (d 5 0 = 1.82pm, MagChem 5 0 M , Martin Marietta Magnesia Specialties) and isopropyl alcohol (Aldrich). A l l these materials were used in the as-received state. The preparation procedures of the precursor were described in Section 4.1.2. The obtained precursor is denoted M A - 8 . The mechanical mil l ing was as described in Section 4.1.3, but without the mixing process. The grinding was carried out continuously in Section 4.1.3, while it was suspended for 0.5 h after every 0.5 h of grinding for this set of experiments. The ground precursors were calcined at temperatures between 800° and 1280°C for 2 h. 4.1.5 Preparation of Spinel Precursors by Sol-Gel and Precipitation The sol-gel-precipitation (SGP) preparation process of the spinel precursor is schematically shown in Figure 4.1-1. Aluminum isopropoxide [Al( -0- i -C3H 7 ) 3 ] and magnesium acetate tetrahydrate [Mg(CH3C00) 2 -4H20] were used as the starting materials. Aluminum isopropoxide (0.10 mol, 20.42g) was dissolved in isopropyl alcohol (3.0 mol, 180.30g) and refluxed for 2 h. Magnesium acetate tetrahydrate (0.05 mol , 10.72 g) was dissolved in distilled water (1.05 mol, 19.0 g) under stirring at room temperature to produce near-saturated solution to minimize the amount of water used. Magnesium acetate aqueous solution was then added drop-wise to the aluminum isopropoxide solution and refluxed for 3h. At this stage, aluminum isopropoxide hydrolyzed to the sol phase, while magnesium still remained in the solution. In order to initiate the second stage of the process (i.e. the magnesium hydroxide precipitation, simultaneously with gelation of the aluminum hydroxide phase), ammonium hydroxide (28%, analytical grade) was added slowly, while stirring, to the solution and the p H value was maintained at 9.5-10 for 3 h. The obtained Chapter 4 Methodology 38 suspension was then dried at 60°C for 12 h, milled for 15 h in isopropyl alcohol with zirconia balls, dried again at 80°C for 12 h and further dried at 110°C for 24 h. The precursor samples were then calcined for 5 h at 600°C and 900°C. Part of the 900°C calcined powder was milled for 0.5 h in isopropyl alcohol with a planetary mi l l (Fritsch, Pulverisette 7) with 100 grams of zirconia balls. The milled precursor was dried at 110°C for 24 h. 0.1 mol A l ( 0 - i - C 3 H 7 ) 3 (20.42g) 0.05mol Mg-acetate (aqueous solution) -7ml NH 4OH 3 mol i - C 3 H 7 O H (180.30g) in 1L flask Reflux for 2 h Reflux for 3 h Stir for 3 h, pH = 9.5 Drying at 60°C for 12 h Milling in isopropanol for 15 h Drying at 80°C for 12 h Spinel precursor MA-9 Figure 4.1-1 Procedures for preparing M A - 9 spinel precursor by sol-gel and precipitation 4.1.6 Morphology of Spinel Powders The crystallite size, particle size and morphology of the spinel precursors before and Chapter 4 Methodology 39 after calcination, prepared through coprecipitation, heterogeneous sol-gel and mechanical activation methods, were determined using a Hitachi scanning electron microscopy ( S E M , S3000N) and a Hitachi transmission electron microscopy ( T E M , H-800). The apparent crystallite size of the calcined powders was calculated using a Siemens X -ray diffractometer ( X R D , model 5000,) patterns and the Scherrer equation [96]: Lcos# 0 Where X: the wave length of X-ray using C u K a l (0.154056 nm) L: the mean dimension of the crystallites (nm) $o- the Bragg angle of the corresponding plane in a crystal lattice (°) /32e'- the full width at half maximum ( F W H M ) in radians The full width at half maximum was corrected to minimize the diffraction line broadening due to grain size or lattice strain using the following equation: 0=(/?eXp-/?instr)1/2 (4.1-2) Where /J: the corrected width of the diffraction line (in radians) /3exp: the experimental half-width (in radians) Pinstr- the instrumental broadening (in radians) Pinstr is determined experimentally with an M g A l 2 0 4 sample composed of large crystallites (>500 nm). The strongest diffraction line (311) (corresponding to 26 = 36.87° using C u K a l ) was chosen for the measurement of the apparent crystallite size. The Chapter 4 Methodology 40 diffraction patterns were recorded using a 20 step of 0.02° and a step time of 1.5 seconds. Specific surface area of the calcined powders was measured with Brunauer-Emmett-Teller (BET) nitrogen-gas adsorption method (Autosorb-1, Quantachrome, U S A ) . The "equivalent average spherical particle" size of the calcined powders was also calculated from the specific surface area. Such calculated "particle size" is therefore somewhat speculative measure of the connectivity between the crystallites. The relationship between the specific surface area and the equivalent spherical diameter is governed by the following equation: Where n: number of particles in weight-unit powder A: specific surface area, m 2 /g D: equivalent spherical diameter, nm p: true density of magnesium aluminate spinel, 3.5782 g/cm3 Then the equivalent spherical particle size of the calcined powders is calculated from specific surface areas [30]: The agglomerate size distribution of the powders was measured with a laser particle size analyzer (MasterSizer 2000, Malvern Instruments). Such determined "agglomerates" describe connection between many crystallites or particles, and thus are many order of A = (4.1-3) D 1000x6 (4.1-4) pxA Chapter 4 Methodology 41 magnitude larger than the "crystallites" or "particles". Typically such agglomerates are quite porous and easy to crush to reduce their size through grinding. Agglomeration of the powders was indirectly determined by comparing the crystallite size calculated from X R D data, the particle size obtained from B E T test and agglomeration size distribution. 4.1.7 Phases and Reaction Kinetics The phases and crystallinity of the precursors and the calcined samples were determined using an X-ray diffractometer ( X R D , Siemens, model 5000) with a scanning speed of 27min. The decomposition behavior and phase change of the precursors with temperature were analyzed using thermogravimetric ( T G A ) and differential thermal analysis (DTA, Setaram, TG-96); the measurement was performed at 5, 10 and 15°C/minute under a flow of helium. The normal approach to study reaction kinetics is to carry out quantitative X-ray analyses of the samples isothermally treated at a series of temperatures. In this study, an alternative method developed by Kissinger [97] was adopted to obtain the activation energy for the reactions of the precursor during heat treatments. The Kissinger analysis requires only the differential thermal peak temperature of the reaction at various heating rates. The Kissinger equation is expressed as: In {dTldt)\_-Ea ( 1 + B (4.1-5) B = In AR (4.1-6) = constant Where T is temperature in K f is time in second Chapter 4 Methodology 42 Tm is the peak maximum temperature of the reaction Ea is the activation energy A is the Arrhenius constant R is the gas constant (8.31451 J / m o l K ) vs. -^J- for various heating rates, activation energy is obtained from the 4.2 Compact Preparation and Characterization Procedures 4.2.1 Raw Materials The raw materials used in the experimental work are shown in Table 4.2-1. AR-78 is an alumina-rich magnesium aluminate (22.0-23.0% M g O , 76.4-77.6% A 1 2 0 3 ) . TSP-15 is an ultrafine magnesium aluminate spinel power with a purity of over 99.9%, a median size of 0.3 pm and a specific surface area of 12m /g (referred to as UF-spinel-1). S30CR is an ultrafine magnesium aluminate spinel with a median size of 0.2 pm and a specific surface area of 30 ± 5m 2 /g (referred to as UF-spinel-2). Alphabond 300 is a hydratable alumina with a specific surface area of 200m /g and a median size between 2.1 and 2.9pm [92]. This hydratable alumina is suitable for castables either with or without silica fume. T-64 is a tabular a-alumina with A 1 2 0 3 > 99.4%. The tabular name comes from its hexagonal tablet-shaped crystal. RG100 is an ultrafine a-alumina with 99.8% A 1 2 0 3 , a median size of 0.5 pm and a specific surface area of 7.7m /g (referred to as UF-alumina). The above values without specified sources are quoted from the specifications of the products. B y plotting - I n [dTldt) Chapter 4 Methodology 43 Table 4.2-1 Raw materials for compacts Materials Commercial name Particle size (pm) Producer M g A l 2 0 4 Spinel AR-78 0-90 (median 24)* Almatis AR-78 0-45 (median 14)* Almatis AR-78 0-20 (median 3)* Almatis T S P - 1 5 A 0.3 (median)* Taimei Chemicals S 3 0 C R A 0.2 (median)* Baikowski Hydratable alumina Alphabond 300 2.1-2.9 (median) Almatis Alumina T-64 150-850 Almatis R G 1 0 0 A 0.5 (median)* Almatis * From the specifications provided by the producers A TSP-15 = UF-spinel-1; S30CR = UF-spinel-2; RG100 = UF-alumina 4.2.2 Formulations of Binary Systems In this work, the bi-modal (or two-component) and tri-modal (or three-component) powder systems refer to those with two and three peaks in the particle size distributions respectively. The formulations of the mixtures shown in Table 4.2-2 were designed to form bi-modal mixtures in terms of particle sizes. A l l the mixtures in Table 4.2-2 were in the region rich in large particles (70% >5 pm). The amount of ultrafine spinel powders (UF-spinel-1 and UF-spinel-2) varied from 0% to 10 wt%. UF-spinel-1 and UF-spinel-2 were used to investigate the influence of particle size of ultrafine spinel powders on the sintering and strength development of the mixtures. The influence of the agglomeration of ultrafine spinel powders on sintering and strength development was investigated by comparing the properties of the mixtures containing the in-house synthesized spinel powders with mill ing (A4 series) and without mill ing (A3 series) respectively. Moreover, a comparison was made Chapter 4 Methodology 44 between the mixtures containing commercial ultrafine spinel powders and those including the in-house ultrafine spinel powders. Table 4.2-2 Formulations of bi-modal mixtures M i x code AR-78 AR-78 S30CR* TSP-15* M A - 9 -900 M A - 9 -900-milled 0-90 pm 0-20 pm 300 nm 200 nm AO 70 30 - - - -A l - 1 70 29 1 - - -A l - 2 70 28 2 - - -A l - 3 70 27 3 - - -A l - 4 70 25 5 - - -A l - 5 70 20 10 - - -A2-1 70 29 - 1 - -A2-2 70 28 - 2 - -A2-3 70 27 - 3 - -A2-4 70 25 - 5 - -A2-5 70 20 - 10 - -A3-1 70 29 - - 1 -A3-2 70 28 - - 2 -A3-3 70 27 - - 3 -A3-4 70 25 - - 5 -A3-5 70 20 - - 10 -A4-1 70 28 - - - 1 A4-2 70 27 - - - 2 A4-3 70 25 - - - 3 A4-4 70 20 - - - 5 A4-5 70 28 - - - 10 *TSP-15 = UF-spinel-1; S30CR = UF-spine Chapter 4 Methodology 45 4.2.3 Formulations of Tri-Modal Systems The formulations of the mixtures shown in Table 4.2-3 were designed to form tri-modal mixtures in terms of particle size. The fraction of > 90 pm is not a component of the binding systems; this fraction was introduced as a substitution of the coarse fraction in real castable mixes. The effect of the ultrafine alumina and spinel powders on the sintering behavior and properties of the mixtures after heat treatment at 816-1450°C were investigated. A hydratable alumina (oc-bond 300) was used in all the mixes as a binder. Table 4.2-3 Formulations of tri-modal mixtures Mix code T-64 cc-bond 300 AR-78 RG100 A S30CR A TSP-15 A MA-9-900-milled (150-850 nm) 2.1-2.9 |im 0-45 um 500 nm 200 nm 300 nm D l - 1 66 4 30 - 0 - -D l - 2 66 4 25 - 5 - -D l - 3 66 4 20 - 10 - -D2-1* 66 4 25 0 5 - -D2-2 66 4 23 2 5 - -D2-3 66 4 20 5 5 - -D4-1 66 4 25 0 - 5 -D4-2 66 4 23 2 - 5 -D4-3 66 4 20 5 - 5 -D 5 - l & 66 4 25 - - - 5 *D2-1 = D l - 2 ; for constant-heating-rate sintering A TSP-15 = UF-spinel-1; S30CR = UF-spinel-2; RG100 = UF-alumina 4.2.4 Sample Preparation The mixtures were wet mixed in ethyl alcohol for 2 h with addition of three drops of Triton X-100, which is a dispersant used in aqueous and organic solvents. The mixing was Chapter 4 Methodology 46 performed in a traditional ball mi l l for 2 h with alumina balls of about 5 mm in diameter; the ball-to-material weight ratio was 3:1. The bi-modal mixtures were mixed manually with 5.5% aqueous polyvinyl alcohol ( P V A ) solution, which contained 10 % P V A by weight, while the tri-modal mixtures were manually mixed with 5% water. P V A was not used in the tri-modal mixtures because they contained 4% hydratable alumina, which acts as a binder. Immediately after mixing with water, the mixtures were compacted. The specimens for measuring bulk density, apparent porosity, linear change and compressive strength were compacted uniaxially into cylinders of 14 mm in diameter and 14 mm in height at a pressure of 125MPa. The maximum pressure was held for 1 minute. However, the specimens for dilatometry test were compacted uniaxially into cylinders of 8 mm in diameter and 10 mm in height at a pressure of 300 M P a . In order to achieve uniform density, the maximum pressure was held for 3.5 minutes. The compacts were dried at 110°C for 24 h, and then fired in air for 5 h at 816° -1450°C, at a heating rate of about 5°C/min. Constant heating rate sintering was performed in air at two heating rates of 2°C/min and 4°C/min. The linear shrinkage during sintering was monitored using a dilatometer (RPZ-2A, Hengda, China) with correction of the instrument length variation and of the thermal expansion of the samples. 4.2.5 Particle Size Analysis Particle size analysis of the experimental mixes was performed using a laser particle size analyzer (Mastersizer 2000, Malvern Instruments Inc.). The samples of the mixtures were made into pastes with additions of appropriate amounts of distilled water and a drop of Triton X-100 (composed of octylphenol ethoxylate). Then appropriate amounts of the pastes were moved into the analyzer tank with distilled water and dispersed under mechanical and ultrasonic stirring. Chapter 4 Methodology 47 4.2.6 Sintering Kinetics The shrinkage rate of the compacted samples was determined by a dilatometer ( R P Z - 2 A , Hengda, China). A t constant rate of heating, the relationship of linear shrinkage and temperature is governed by Equation 2.4-4 for a dominating diffusion mechanism (grain boundary diffusion or volume diffusion). The apparent activation Td(AL/ L0) energy of sintering is calculated from the slope of a plot of In versus 1/T dT from the linear change of the sample under heating at a constant rate (refer to Section 2.4.3) [72-74]. Surface diffusion occurring in the sintering is recognized in the constant-rate-of-heating ( C R H ) plot. 4.2.7 Physical Properties and Microstructure The green density was determined from the sample weight and dimensions. The permanent linear change (PLC) was obtained by measuring the heights of the samples before and after firing. The bulk density (BD) and apparent porosity (AP) of fired samples were measured by the immersion method in kerosene under vacuum using the Archimedes' principle. Kerosene, instead of water, was used because the samples would possibly hydrate when subjected to water. The measurement of cold crushing strength was performed using a Tinius Olson hydraulic press with a head speed,of 0.0127 mm/min. Each value was the average of the test results of three samples of 14 mm in diameter and 14 mm in height. Microstructural characteristics of the samples after firing at different temperatures were examined using S E M (S3000N, Hitachi). Chapter 4 Methodology 48 4.3 Castable Preparation and Characterization Procedures 4.3.1 Raw Materials The raw materials used in the work are shown in Table 4.3-1. The Spinel-25 aggregate was crushed in the laboratory conditions and all the other raw materials were used as received. Table 4.3-1 Raw materials for castables Materials Commercial name Particle size (pm) Producer/supplier Magnesia Baymag 96 (MgO >96.0%, fused) 3350-4750 Baymag 1180-3350 <1180 P-98 (MgO 98.0%, deadburnt) <75 Martin-Marietta M g A l 2 0 4 Spinel SP-25 (MgO 25.0%, A 1 2 0 3 74.3%, fused) 3350-4750 C E Minerals 1180-3350 <1180 <75 AR-78 (MgO 22.0-23.0%, A 1 2 0 3 76.4-77.6%, fused) <45 pm Almatis <20pm TSP-15* 0.3 pm (median) Taimei Chemicals S30CR* 0.2 pm (median) Baikowski Hydratable alumina Alphabond 300 2.1-2.9 (median) Almatis Alumina CTC53 1.1pm (d50) Almatis A3000FL 2.8pm (median) Almatis RG100* 0.5pm (median) Almatis *TSP-15 = UF-spinel-1; S3 OCR = UF-spinel-2; RG100 = UF-alumina Chapter 4 Methodology 49 4.3.2 Formulations of Castables The formulations of the experimental castables are shown in Tables 4.3-2 to 4.3-5. The mixes M l - 0 to M l - 6 in Table 4.3-2 were designed to investigate the influence of the ultrafine alumina powder (varying from 0% to 6%) on the castable properties. Also investigated was the effect of the in-situ spinel formation on the castable properties by comparing M l - 6 with 4% M g O powder and M2-6 without M g O power. A l l the castables contained magnesia aggregate, 4% hydratable alumina as the hydraulic binder, and 0.10% sodium polymethacrylate ( S P M A ) and 0.02% citric acid as the dispersants. Table 4.3-2 Formulations of castables with fused magnesia aggregate Particle size (pm) Composition (%) M l - 0 M l - 3 M l - 6 M2-6 Baymag 96 3350-4750 10 10 10 10 1180-3350 25 25 25 25 600-1180 20 20 20 20 100-600 10 10 10 10 SP-25 <75 16 16 16 16 AR-78 <45 - - - 4 <20 5 5 5 5 Alphabond 300 2.1-2.9 (median) 4 4 4 4 P-98 <75 4 4 4 -A3000FL 2.8 (median) 6 3 - -RG100* 0.5 (median) 0 3 6.0 6 * RG100 = UF-alumina Chapter 4 Methodology 50 Table 4.3-3 Formulations of castables with fused spinel aggregate Particle size (pm) Composition (%) S l - 0 . S l - 3 S l - 6 S2-6 SP-25 3350-4750 10 10 10 10 1180-3350 25 25 25 25 600-1180 20 20 20 20 100-600 10 10 10 10 <75 16 16 16 16 AR-78 <45 - - - 4 <20 5 5 5 5 Alphabond 300 2.1-2.9 (median) 4 4 4 4 P-98 <75 4 4 4 -A 3 0 0 0 F L 2.8 (median) 6 3 - -RG100* 0.5 (median) 0 3 6.0 6 *RG100 = UF-alumina The experimental mixes in Table 4.3-3 were designed to study the influence of the ultrafine alumina powder and the in-situ spinel formation on the properties of the castables in the same way the mixes in Table. 4.3-2 were formulated. The only difference between the mixes in Table 4.3-3 and Table 4.3-2 is that the aggregate in the mixes in Table 4.3-3 is spinel, replacing the magnesia aggregate in Table 4.3-2. The experimental mixes in Table 4.3-4 were designed to investigate the influence of U F -spinel-1 (varying from 0% to 5%) on the castable properties. The effects of the two aggregates and the in-situ spinel formation were also investigated using the mixes. M3-0 , Chapter 4 Methodology 51 M3-3 and M3-5 contained 4% M g O , indicating that in-situ spinel would form during firing; in contrast, mixes S3-0 and S3-5 had no M g O and, as a result, the in-situ spinel was not expected to form during firing. The aggregate in M3-0 , M3-3 and M3-5 was fused magnesia and the in-situ spinel could form on the surface of the magnesia aggregate because the alumina in the binding system would react with aggregate to form spinel. On the other hand, the aggregate in the mixes S3-0 and S3-5 was fused spinel and no in-situ spinel could form on the surface of the aggregate. Table 4.3-4 Formulations of castables with ultrafine spinel powder ( d 5 0 = 0.3 pm) Particle size Composition {%) (pm) M3-0 M3-3 M3-5 S3-0 S3-5 Baymag-96 3350-4750 20 -1180-3350 25 -<1180 20 -SP-25 3350-4750 - 20 1180-3350 - 25 <1180 - 20 SP-25 <75 23 20 18 31 26 TSP-15 0.3 (median) 0 3.0 5.0 0 5 P-98 <75 4 4 . 4 - -Alphabond 300 2.1-2.9 (median) 4 4 4 4 4 C T C 5 3 1.1 urn (median) 4 4 4 - -*TSP-15 = UF-spinel-1 Chapter 4 Methodology 52 The mixes in Table 4.3-5 were designed to study the influence of UF-spinel-2 (varying from 0% to 5%) on the castable properties. Furthermore, these mixes and S3-0 and S3-5 in Table 4.3-4 were used to compare the influence of the two ultrafine spinel powders on the castable properties. The particle size distributions of the mixes were analyzed and in agreement with Andreassen equation with a particle-size distribution coefficient of 0.27. Table 4.3-5 Formulations of castables with ultrafine spinel powder (dso = 0.2 pm) Particle size (pm) Composition (%) S4-0 S4-2 S4-5 SP-25 3350-4750 10 10 10 1180-3350 25 25 25 600-1180 20 20 20 100-600 10 10 10 <75 7 7 7 AR-78 <45 13 13 13 Alphabond 300 2.1-2.9 (median) 4 4 4 A3000FL 2.8 (median) 6 4 1 RG100* 0.5 (median) 5 5 5 S30CR* 0.2 (median) 0 2 5 *RG100 = UF-alumina; S3 OCR = UF-spinel-2 The binding systems had multi-modal particle size distribution of 45-75, 1-3 and 0.2-0.3 pm. As found by Furnas [43, 98] and discussed in Chapter 6, the multi-modal powders are beneficial to increase the density and decrease the shrinkage of the castables during heat treatment. Chapter 4 Methodology 53 4.3.3. Sample Preparation The mixes were dry mixed for 3 minutes and wet mixed for 5 minutes after adding tap water (5.1 wt%), using a Hobart mixer (model N50). A l l the samples were cast using a vibrating table with dimensions of 25.4x25.4x177.8 mm (1x1x7 in). The castables were cured in the mold in air at ambient temperature for 24h. After de-molding, they were cured in sealed plastic bags for another 24h. The samples were dried at 110°C for 24h and fired for 5 h at 816°C and 1280°C respectively. 4.3.4 Flowability Flowability of the castables was measured using a flow cone according to A T S M C860-91. The castable mixes were wet mixed for 5 minutes after adding tap water, using a Hobart N50 mixer. The flow cone [(|>(70-100)x50 mm] was placed on a vibrating table, with the larger end up. Excess amount of castable was filled into the flow cone and vibrated until all the large air bubbles cease evolving from the castable. The excess castable was scraped off. The sample was vibrated again until its upper surface was smooth and even with the upper rim of the cone. Then the sample with the cone was moved to the center of a flow table, with the smaller end up. After the cone was vertically lifted away, the flow table was dropped 15 times in 9 s. The flow value is the increase in average base diameter of the castable mass, expressed as a percentage of the original base diameter. 4.3.5 Physical Properties Three bars of 25.4x25.4x177.8 mm (1x1x7 in) after drying and firing were used to do three-point-bending modulus of rupture ( M O R ) ; then halves of the samples were used to measure C C S and the other halves were used to measure B D and A P . The M O R tests were Chapter 4 Methodology 54 performed using an Instron machine with a span of 127 mm and a head speed of 0.127 mm/min. P L C , C C S , B D , AP and microstructure were determined as described in Section 4.2.6. A l l physical properties of the castables were determined on the average value of at least three specimens. The increase in apparent porosity of the castables after firing was calculated as follows: AP — AP APD Where APJ : Increase in apparent porosity of castables after firing at temperature T (%) APT = Average apparent porosity of castables fired at temperature T (%) APD = Average apparent porosity of castables dried at 110°C (%) 4.3.6 Hydration Procedures The raw materials were hydratable alumina (Alphabond 300, dso = 2.1-2.9 pm) from Almatis (Bauxite, U S A ) , reactive magnesia (dso = 3.09 pm) from Fisher Scientific, 98.0% M g O deadburnt magnesia (<75 pm) from Martin Marietta (P-98, Baltimore, U S A ) and 96-98% M g O fused magnesia (<75 pm) from Washington M i l l s (Dynamag K , Niagara Falls, U S A ) . Twenty five grams of the as-received individual oxides as well as three mixtures of hydratable alumina and magnesia with a MgO:Al i03 weight ratio of 84:16 were mixed with 20 grams of distilled water. The three mixtures were composed of hydratable alumina and reactive magnesia (Mixture I), hydratable alumina and fused magnesia (Mixture II) and hydratable alumina and deadburnt magnesia (Mixture III). Two hydration conditions were employed: one hydration was performed for 48 h at 20°C in sealed polyethylene bags (in air atmosphere) (Hydration-A), and the second Chapter 4 Methodology 55 hydration included 48 h at 20°C in sealed polyethylene bags and then 12 h at 110°C in steam at a pressure of 34.5kPa (5 psi) in an autoclave (Hydration-B). A l l the samples were dried at 110°C for 24 h following hydration. The phase composition of the hydrated samples was determined by X R D (model 5000, Siemens). The thermal decomposition behavior and hydration extent of the hydrated samples (ground to powder) was investigated using T G A (TG-96, Setaram) at a heating rate of 10°C/minute under a flow of helium. The sample size for T G A analysis was 20-50 mg. The weight loss due to evaporation of adsorbed water (up to 110°C) was subtracted from the overall weight loss measured. The morphological features of the hydrated Alphabond 300 and Mixture III were examined using S E M (S3000N, Hitachi). Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 56 CHAPTER 5 SYNTHESIS OF MAGNESIUM ALUMINATE SPINEL POWDERS 5.1 Spinel Synthesis by Coprecipitation 5.1.1 Morphology of the Calcined Powders The preparation of the spinel precursors by coprecipitation was described in Section 4.1.1. As observed, the precursor M A - 1 showed a very large shrinkage and became very hard after drying at 110°C, while the precursors M A - 2 and M A - 3 exhibited much less shrinkage during drying and were soft and quite easy to break. The textures of the precursors calcined at 700°C or 800°C are shown in Figure 5.1-1. Because M A - 1 after calcination was in the form of a hard lump, the material was ground by hand using a mortar and pestle. As seen in Figure 5.1-la, the large particles of M A - 1 after firing at 800°C have a solid texture. As described above, the precursor of M A - 1 was hard after drying, indicating that the agglomeration of the precursor occurred during drying. For the precursor suspension dispersed in water, the free water molecules form bridges between the surface hydroxyl groups of neighboring precursor particles through hydrogen bonding [48]. During drying, the evaporation of the coordinated water molecules brings the precursor particles into close proximity. Hard agglomerates would form during calcination of the precursors because a condensation reaction, as given by - M - O H + H O - M - —> - M - O - M -+H2O [48], occurred between the surfaces of the precursor particles. That is why severely agglomerated large particles were observed in M A - 1 after calcination (Figure 5.1-la). In contrast to M A - 1 (Figure 5.1-la), M A - 2 after calcination at 700°C displays a highly porous/loose texture (Figure 5.1-lb). It was visually observed that the precursor M A - 2 incurred much less shrinkage during drying than the precursor M A - 1 and was soft and quite easy to break. Because carbon black (-50 nm median particle size) was well mixed with the Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 57 precursor suspension in ethyl alcohol, it served as a barrier against the direct contact between the precursor particles in suspension and consequently, against their direct contact during drying (refer to Figure 2.2-1). After burning of the carbon black during calcination, the space occupied by carbon black becomes a continuous pore network, making the material structure very porous and loose. Figure 5.1-1. S E M micrographs of (a) M A - 1 after calcination at 800°C, (b) M A - 2 calcined at 700°C, (c) M A - 3 calcined at 700°C and (d) M A - 5 calcined at 800°C. Through introduction of P V A in the M A - 3 precursor suspension, the precursor particles were dispersed in the polymeric network of the P V A and, as a result, the direct contact Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 58 between the precursor particles was prevented. The polymeric structure of P V A with precipitate particles distributed within the network was maintained during drying of the M A -3 suspension. As described before, the agglomeration of the precursor particles during drying is mainly due to the hydrogen bonding between water molecules and the hydroxyl groups on the surface of the precursor [48]. The P V A network where the precursor particles were dispersed prevented the hydrogen bonding from bringing the precursor into close proximity during drying. After the decomposition and combustion of P V A , the spaces between the solid particles, previously occupied by P V A , were preserved as pores (Figure 5.1-lc). Consequently, a continuous network of pores occurred in the resulting material. 5.1.2 Phase Evolution during Calcination As shown in Figure 5.1-2a, the material from M A - 1 after calcination at 500°C is almost completely amorphous. The intensity of the peaks belonging to spinel increased with calcination temperatures and phase-pure spinel was obtained after calcination at 1100°C. 15 25 35 45 55 65 15 25 35 45 55 65 20 (°) 29 (°) Figure 5.1-2 X R D patterns of (a) M A - 1 and (b) M A - 2 after calcination for 3h at different temperatures. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 59 Figure 5.1-2b shows that the phase evolution of M A - 2 after calcination at temperatures between 700°C and 1100°C was the same as of M A - 1 . The phase change of M A - 3 after calcination in the same temperature range was also the same as that of M A - 1 (the X R D patterns are not shown here). These results indicate that the combustible ingredient addition did not impede the spinel formation. The combustible additives were added after the precursor precipitates were formed, so their incorporation does not influence the direct contact between the magnesium hydroxide and aluminate hydroxide in the precursors. Consequently, the addition of the combustible ingredients into the already precipitated precursors prepared by co-precipitation did not impede the spinel formation. 5.2 Spinel Synthesis by Heterogeneous Sol-Gel 5.2.1 Morphology of Calcined Powders The preparation of the spinel precursors by heterogeneous sol-gel process was described in Section 4.1.2. The precursor of M A - 5 felt soft and loose after drying. Figure 5.1-ld shows that most of the M A - 5 particles after calcination at 800°C are disintegrated and a continuous pore network is present within the material. There are large particles of ~3-5pm, but not as solid as seen in Figure 5.1-la. Instead, they have a hollow structure and as a result, they are easy to break into very fine particles. For the preparation of the 0.05 mol M g A l 2 0 4 mol precursor, 1.5 moles of isopropyl alcohol (114.78ml) and 2 moles of water (36.04ml) were used in the precursor suspensions. The volume ratio of alcohol/water was 3.18. The concentration of water molecules on the neighboring precursor particles would be decreased due to the presence of alcohol. As a result, the hydrogen bonding between the water molecules and the surface hydroxyl groups Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 60 of neighboring precursor particles would be decreased [48], and the agglomeration of the precursor particles during drying would be decreased. Consequently, the calcined product (Figure 5.1-ld) was not as severely agglomerated as the powder (Figure 5.1-la) prepared using coprecipitation in a completely aqueous suspension. 5.2.2 Phases of the Precursors As seen in Figure 5.2-1, the crystalline phases in the M A - 5 precursor are boehmite (A lOOH) and magnesia (MgO). In comparison, M g 4 A l 2 ( O H ) i 4 - 3 H 2 0 was the main phase and the M g O phase was not observed in the spinel precursor [10] prepared using the same process when M g O powder with a median particle size of 10 nm was used. In contrast, the median particle size of the M g O powder used in this work was 3.09 pm. Obviously, because the M g O powder used in this work was relatively coarse, the M g O phase was found in the M A - 5 precursor. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 61 Figure 5.2-1 shows that the M A - 5 precursor had the same phases as M A - 5 A , indicating that extending the fluxing time after the addition of light M g O from 2 h to 5 h had no influence on the phases of the precursors. From an economical and efficiency perspective, short refluxing time is preferred. Based on these results, the fluxing time of 2 h after addition of the M g O powder was chosen for the preparation of spinel-coated magnesia. The M g ( O H ) 2 peaks are not noticeable in the M A - 5 , M A - 5 A and M A - 6 precursors (Figure 5.2-1). However, reactive magnesia was completely hydrated to form brucite [Mg(OH) 2 ] after curing for 22 h at 30°C (see Figure 5.2-2) [39]. Schulle et al [99] also observed that sintered M g O powder (< 45 \xm) stored for 24h in water at room temperature was considerably hydrated. With regards to the weak peaks of M g ( O H ) 2 in the M A - 5 and M A - 5 A precursors (Figure 5.2-1), it is not clear i f the presence of A l O O H sol in aqueous M g O / M g ( O H ) 2 suspensions decreased the hydration of M g O or impeded the crystallization of M g ( O H ) 2 during refluxing. 29 (°) Figure 5.2-2 X R D pattern of reactive magnesia hydrated for 22 h at 30°C [39]. (B=brucite M g ( O H ) 2 , A=aluminum sample holder). Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 62 5.2.3 Phase Evolution during Calcination The only crystallite phase after calcination of M A - 5 at 600-800°C was M g O , indicating that A 1 2 0 3 was in an amorphous state (Figure 5.2-3a). Spinel started to form after calcination at 1000°C and phase-pure spinel was obtained after calcination at 1200°C. In the report [10] where spinel was synthesized using the same method and nano-sized M g O (with median particle size of 10 nm), completely crystallized spinel was not achieved when calcination temperature was increased to 1200°C (Figure 5.2-4). The precursor consisted of a mixture of boehmite and M g 4 A l 2 ( O H ) i 4 - 3 H 2 0 [10] and the spinel formation from the precursor is assumed to take place in two steps. The "primary spinel" phase formed from the decomposition of Mg4Al 2 (0H) i 4 -3H 2 0 during heating: M g 4 A l 2 ( O H ) 1 4 - 3 H 2 0 -> M g A l 2 0 4 + 3MgO + 1 0 H 2 O l (5.2-1) The excess M g O reacts with A 1 2 0 3 in the precursor to form the "secondary spinel". To our knowledge, this kind of "secondary spinel" or "secondary spinellization" has not been discussed before. However, the secondary mullite formation is very similar to this and has been investigated in detail [100-104]. In a mixture of kaolinite ( 2 S i 0 2 A l 2 0 3 - 2 H 2 0 ) and alumina, the primary mullite is formed from the decomposition of kaolinite at 1300°C or below: 3 [ 2 S i 0 2 A l 2 0 3 - 2 H 2 0 ] - » 3 A l 2 0 3 - 2 S i 0 2 + 4 S i 0 2 + 6H 2 01 (5.2-2) Then the secondary mullite formation takes places from the excess silica and the added alumina at 1400°C or higher [100-102]. In the same way, the secondary spinel formation takes place at higher temperatures than the primary spinel formation because the extra M g O has to diffuse out of the primary spinel agglomerates before reacting with the added A 1 2 0 3 . Therefore, phase-pure spinel was not obtained at 1200°C in the report [10] because the Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 63 secondary spinellization did not take place completely at the temperature. In contrast, phase-pure spinel was obtained at 1200°C in this work because the spinellization took place in only one step. It is inferred from the above discussion that the reactive M g O is more beneficial to phase-pure spinel formation than nano-sized spinel formation. 26 O (b) • 1200°C . I ° I .fi r I P i | 1000'C . 800°C 15 25 35 45 26 (°) Figure 5.2-3 X R D patterns of M A - 5 (a) and M A - 6 (b) after thermal treatment at different temperatures ( • M g O ; • M g A l 2 0 4 ) . Figure 5.2-3b shows .that M g O was the only crystalline phase in M A - 6 after calcination at 600-800°C, indicating that AI2O3 was in an amorphous state. Spinel peaks are not noticeable after calcination at 1000°C. In comparison, spinel peaks are appreciable in M A - 5 after calcination at the same temperature (Figure 5.2-3a). This is because the finer reactive magnesia in M A - 5 is easier to react with alumina to form spinel than the coarser magnesia in M A - 6 . After calcination at 1200°C, both spinel and M g O sharp peaks are present. From the above results, it is concluded that spinel-coated magnesite powder was prepared by the heterogeneous sol-gel processing, followed by calcination at 1200°C. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 64 c 3 flj c - i 1 1 1 1 1 r — O: MgAfeCu * : MgO O A : a -AI 2 0 3 • : y -AI 2 0 3 O o O O 1200°C A O O 0 0 800°C 700°G 500°C 400°C Precursor 10 20 30 40 50 60 70 80 90 2 8 / degree ( C u K J Figure 5.2-4 X R D patterns showing the effect of calcinating temperature on the formation of crystalline phases from a spinel precursor [10]. 5.3 Spinel Synthesis by Mechanical Activation 5.3.1 Phase Changes of the Precursors during Milling The starting materials and the grinding and heating processes were described in Section 4.1.3. Figure 5.3-1 shows the X R D patterns of the M A - 2 1 and M A - 2 2 mixtures that were subjected to grinding for various time periods ranging from 5h to 40h. A s in Figure 5.3-la, the peaks of boehmite and brucite in M A - 2 1 are broadened after mil l ing for 5h and become lower with increasing the mil l ing time, until they nearly disappear after 40h of milling, indicating that the starting materials are converted to an amorphous state after mill ing for 40 h, as reported in reference [59]. As shown in Figure 5.3-lb, the mixture without mil l ing displays only peaks characteristic of brucite, because the as-received aluminum hydroxide in M A - 2 2 is Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 65 amorphous. After grinding for lOh, the peaks corresponding to brucite are significantly broader and lower. A l l the peaks belonging to brucite almost completely vanish after the milling for 20 h. After 40h of mill ing, the starting materials become completely amorphous. The broadening of the peaks in Figure 5.3-1 also indicates that refinement of crystallite size took place during mechanical grinding. "cn c 10h 20h Figure 5.3-1 X R D patterns of (a) M A - 2 1 and (b) M A - 2 2 mixtures versus mill ing time ((O) brucite, ( • ) boehmite). 5.3.2 Phase Evolution during Calcination The MA-21 Mixture Figure 5.3-2 shows the X R D patterns of the M A - 2 1 mixtures milled for different periods of time, followed by heat treatment for 5h at different temperatures. As seen in Figure 5.3-2a, after calcination at 400°C, the M g O peaks decrease with mill ing time and completely disappear after 20h of mill ing. It is the same case for the mixtures calcined at 600°C and 800°C. After calcination at 1000°C, M g O peaks appear only in the mixture without milling. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 66 CO c 0) (a) Oh 1 , . rn in r'l T-"—ffrfr*•,>! i f f ^ t , , „ tHiuf1** 40h 15 25 35 45 55 65 75 29(°) 03 CO c 3 03 CO c o 3 03 CO c 0) (d) (mi A • • (311) 0 (220) A ( 4 0 0 1 A • . 1 A ,422)(511) («0) A » A 0 A Oh J - A . ,A 5h A A ...J LJ - A t 10h 1 20h - J l — —p-1—i—I—|—r I k i—I—I r-n—i—r ^ J L 40h 15 25 35 45 55 65 29(°) 75 Figure 5.3-2 X R D patterns of M A - 2 1 after milling and calcination at (a) 400°C, (b) 600°C, (c) 800°C and (d) 1000°C for 5h ((A) M g A l 2 0 4 spinel, (O) M g O , ( • ) A 1 2 0 3 ) . Spinel begins to form during calcination at 400°C and 600°C (Figure 5.3-2a and Figure 5.3-2H respectively). However, the spinel peaks are quite broad and low. In the mixtures subjected to mechanical activation for 5h and calcination at 800 and 1000°C (Figure 5.3-2c and Figure 5.3-2d respectively), the spinel peaks are much stronger than those of the mixtures without mill ing, indicating that mechanical activation of the starting mixtures enhanced spinel formation during heat treatment at the temperatures. Only spinel peaks Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 67 appear in the mixture subjected to 20h of milling and calcination for 5h at 1000°C (Figure 5.3-2d), indicating that phase-pure spinel was obtained. At calcination 1000°C, the spinel peaks are signicantly higher after mil l ing for 5 h, compared to the mixtures without mill ing (Figure 5.3-2). However, the spinel peaks do not increase significantly when the mill ing time increases from 5h to 40h, indicating that 5h is sufficient for mechanical activation of the starting materials of brucite and boehmite. As described in Section 5.1, phase-pure spinel was not formed from the precursor prepared by a coprecipitation method until the calcination temperature was raised to 1100°C. Bratton [105] prepared spinel powder using the same method and the spinel was not well crystallized after calcination at 1000°C. The results in Section 5.2 indicate that the calcination temperature had to reach 1200°C for the phase-pure spinel formation from the precursor synthesized via a heterogeneous sol-gel process. Shino et al [10] prepared spinel using the heterogeneous method and did not obtain completely crystallized spinel when calcination temperature was 1200°C (Figure 5.2-4). Tang et al [27] prepared spinel precursor using a freezing drying technique and phase-pure spinel was obtained until the calcination temperature was raised to 1200°C. In this work, phase-pure spinel was achieved at a relatively low temperature (1000°C) with mechanical activation. It is seen from Figure 5.3-2a and Figure 5.3-2b that spinel peaks are clearly noticeable when the mixtures were milled for 10 h and followed by calcination at 400°C and 600°C, respectively. Spinel peaks are even noticeable in the mixture subjected to calcination at 400°C, without grinding in Figure 5.3-2a. As described in 5.3.1, the experimental mixtures were homogenized in ethanol for 48 h with a traditional mill using zirconia balls as the milling media. This milling with the zirconia balls would also mechanically activate the starting powders. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 68 In contrast, when the coprecipitation process was used, the spinel began to form at 900°C, as described in Section 5.1 and at 800°C, as in reference [13]. Spinel was not observed when the precursor prepared by the heterogeneous sol-gel technique was calcined at 800°C (Figure 2.1-1 and Figure 5.2-3a and). In comparison with these results, the spinel began to form in the mixtures subjected to mechanical activation at 400°C in our experimental work, which is a relatively low temperature. As described in Section 2.2, mechanical activation can generate fracture, lattice deformation [59], dehydroxylation [59] de-bonding and re-bonding [61] in the staring materials. Mechanical activation can also induce intimate mixing in the starting materials and can lead to in-situ synthesis of oxide compounds [62, 63]. In fact, phase-pure M g A l 2 C » 3 spinel was prepared by a mechanochemical route at room temperature with grinding up to 160 h using y-AI2O3 and M g O as the starting materials [106]. Therefore, mechanical mill ing decreased the commencing temperature of the spinel formation. The MA-22 Mixture Figure 5.3-3 shows the X R D patterns of the M A - 2 2 mixtures milled for different times and heat treated at different temperatures. It is seen from Figure 5.3-3a and b that after heat treatment at 400°C and 600°C, there are no visible spinel peaks, whether the mixtures were subjected to mechanical mill ing or not. The only visible peaks belong to periclase when the mixtures were subjected to 5 h and 10 h of grinding. However, when the mixtures were ground for 20 h, the periclase peaks can barely be seen. These results indicates that M g O in the mixtures was in an amorphous state, just as its predecessor M g ( O H ) 2 was in an amorphous state after grinding for 20 h (Figure 5.3-lb). After calcination at 800°C (Figure 5.3-3c) and 1000°C (Figure 5.3-3d), the spinel peaks Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 69 are visible in all mixtures, regardless of grinding. The spinel peaks increase apparently with the mill ing time up to 20 h when calcined at 800°C and up to only 10 h when calcined at 1000°C. Wel l crystallized spinel was formed when the mixture was milled for 10 h and calcined at 1000°C (Figure 5.3-3d). The presence of the minor M g O phase in Figure 5.3-3d is attributed to the excess of M g O in the starting materials as alumina phase is not visible. 15 25 35 45 55 65 75 20 (°) (b) ) o O 1 Oh V it 5h ^ ^ ^ ^ 15 25 35 45 55 65 75 20 (°) (d) « ' ' ) o (111) ( 2 2 0 ) f 1 1 0 0 ) ( 5 1 1 ) ( « 0 ) t A A Ii i l k 0 h . J JU UL^JLI .IOH. A J_J 15 25 35 45 55 65 75 20 (°) Figure 5.3-3 X R D patterns of M A - 2 2 versus mill ing time and heat treatment at (a) 400°C, (b) 600°C, (c) 800°C and (d) 1000°C for 5h ((A) M g A l 2 0 4 spinel, (O) MgO) . Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 70 Phase-pure spinel can be achieved after calcination at 1000°C through mechanical mill ing using M g ( O H ) 2 and amorphous A l ( O H ) 3 as the starting materials. As described above, this calcination temperature is relatively low compared to other synthesis approaches, such as coprecipitation (1100°C) [105], heterogeneous process (>1200°C) [10] and freeze drying (1200°C) [27]. Influence of Starting Materials As shown in Figure 5.3-3a and Figure 5.3-3b, peaks characteristic of spinel in M A - 2 2 are not visible when the mixtures were calcined at 400°C and 600°C respectively, after the mixtures of brucite [Mg(OH) 2 ] and amorphous aluminum hydroxide [Al(OH) 3 ] were ground for 20h. In comparison, the spinel peaks are noticeable in M A - 2 1 after calcination at 400°C (Figure 5.3-2a) and 600°C (Figure 5.3-2a) when the mixtures of brucite [Mg(OH) 2 ] and boehmite ( A l O O H ) were previously ground for 20h. These results demonstrate that spinel formed in M A - 2 1 more readily than in M A - 2 2 , with the same mechanical mill ing and heat treatment at the relatively low temperatures (400-600°C). This difference should be attributed to the difference in raw materials: boehmite was used in M A - 2 1 and amorphous aluminum hydroxide in M A - 2 2 . The results indicate that using boehmite favors the spinel formation over using amorphous aluminum hydroxide. Because the boehmite and amorphous aluminum hydroxide have different compositions and structures, the structural changes of these materials during mill ing could also be different. It has been reported that, among y - A l 2 0 3 - M g O , A l O O H - M g O and a- A l 2 0 3 - M g O mixtures, the formation rate of spinel was the highest in the y - A l 2 0 3 - M g O mixture during mechanical mil l ing [106]. The structure of y - A l 2 0 3 is cubic and the formula can be expressed as A l 0 . 6 7 A l 2 O 4 (0.67 aluminum ions occupy tetrahedral sites and two aluminum ions take the octahedral sites). Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 11 Y-AI2O3 can be considered as a defective spinel in which cation vacancies are considerably ordered. Therefore, it was concluded the structure of the starting materials similar to that of spinel favors spinel formation [106]. K i m and Saito [32] observed that gibbsite [Al(OH) 3] was transformed to boehmite [AlOOH] during mechanical milling, and the mixture milled for 2 h formed spinel much faster than the mixture without milling. Accordingly, the amorphous Al (OH )3 in M A - 2 2 could be transformed to amorphous A l O O H during milling and this transformation would consume mechanical energy to some extent. As a result, the same time of milling would impart more effective mechanical activation to MA-21 than to M A - 2 2 and spinel formed in M A -21 more readily than in M A - 2 2 with the same mechanical milling and heat treatment conditions. However, detailed studies need to be carried out to clarify why the MA-21 mixture is more favorable for spinel formation than the M A - 2 2 mixture. 5.3.3 Morphology of the Synthesized Powders The evolution of the morphology and particle size of the M A - 2 1 mixture during milling is shown in Figure 5.3-4. In the starting mixture (Figure 5.3-4a), the boehmite has the round shape with smooth surface and its large particles are of 8-12 pm in size. The brucite is irregular in shape and its large particles are of 12-15 pm in size. The major changes occurred in the first 5 h of mechanical mill ing (Figure 5.3-4a). The boehmite and brucite particles cannot be separately distinguished in the agglomerates (<15pm) formed during mill ing. As seen in Figure 5.3-5, the agglomerates of the mixture after 10 h of mil l ing are composed of small particle of -2-3 pm, indicative of the considerable decrease in the particle size and the significantly increase homogeneity of the starting materials (<15 pm). The particle refinement is due to the mechanical impact which fractures the particles and the particle agglomeration results from the so-called welding of the activated particles [107]. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 72 Figure 5.3-4 S E M micrographs of M A - 2 1 mixture after mil l ing for (a) Oh, (b) 5h, (c) lOh and (d) 20h. Figure 5.3-5 S E M micrograph of M A - 2 1 mixture after mill ing for lOh (the particle circled in Figure 5.3-4). Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 73 Figure 5.3-6a and b shows the separate distribution of aluminum and magnesium of the starting materials. After mill ing for 5 h, a uniform aluminum and magnesium distribution in the agglomerates was formed (Figure 6.3-6c and d), indicating that the two components were mixed on a micron scale. The microstructural development of the staring materials during mill ing indicates that the particle size was decreased and the ingredients were well mixed and in tight contacts. Referring to Figure 5.3-1, the starting materials underwent deformation and amorphization. A s a result, the starting materials were brought into a higher energy state and the diffusion distance of M g 2 + and A l 3 + ions was shortened. Consequently, the formation of spinel was enhanced and the temperature for the pure-phase spinel formation was lowered. Additional mil l ing up to 20 h produced no significant changes in the morphology of the materials; only a slight decrease of the average size of the agglomerates is noticeable (Figure 5.3-4c and Figure 5.3-4d). This is in agreement with the observations [106, 107] that prolonged mechanical mil l ing of oxide mixtures brings about no significant changes in the morphology and size of agglomerates because the fracture and co-welding reaches a balance during prolonged mill ing. Figure 5.3-7 shows the S E M micrograph of the M A - 2 1 mixture after calcination at 1000°C. The large particles (-15-20 pm) are mainly composed of alumina (refer to Figure 5.3-8a) and the small particles (-6-10) are mainly composed of magnesia (refer to Figure 5.3-8b). This is in agreement with the X R D results shown in Figure 5.3-2d that magnesia, alumina and spinel co-exist after calcination at 1000°C. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 74 Figure 5.3-7 S E M micrographs of M A - 2 1 mixture after heat treatment at 1000°C for 5h without mill ing. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 75 Figure 5.3-9 S E M micrographs of M A - 2 1 mixture after heat treatment at 1000°C for 5h with previous mill ing for (a) Oh, (b) 5h, (c) lOh and (d) 20h. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 76 Figure 5.3-9 shows that the mechanical mill ing time on the particle size and morphology of the M A - 2 1 mixtures after heat treatment at 1000°C for 5h. The particle size in the mixture without mil l ing (Figure 5.3-9a) ranged from -12 pm to 2 pm. After 5h of grinding, the particles became agglomerates and the agglomerate size range between 2-5 pm (Figure 5.3-9a). The large agglomerates after mill ing for 5-20 h and calcination at 1000°C (Figures 5.3-4a-c) were composed of submicron-sized particles (200-300nm), similar to the sizes before calcination (Figure 5.3-5). It is worth noting that the agglomerates in the mixtures after calcination (Figure 5.3-9c to Figure 5.3-9d) (2-5 pm) are smaller than the agglomerates before the heat treatment (<15 pm), due to the fragmentation of the agglomerates during dehydroxylation [32]. As mentioned in Section 1.2, higher calcination temperatures tend to lead to harder agglomeration because sintering (neck growth between contacting particles) is proportionally related to the temperature. Compared to the calcination temperatures required for other processes (coprecipitation: 1100°C [105]; heterogeneous process: >1200°C [10]; and freeze drying: 1200°C [27]), the 1000°C for the formation of completely crystallized spinel following mechanical activation is a relatively low temperature. Consequently, it is expected that the agglomerates in the powders prepared through mechanical activation could be softer than for other synthesis methods. The specific surface area of the M A - 1 1 powder after lOh of mechanical mill ing and followed by calcination at 1000°C is 29m 2/g, and the corresponding equivalent particle size is 57 nm. The fact the equivalent particle size (57 nm) is considerably smaller than the primary particle size (200-300 nm) indicates that the observed primary particles are also porous and the agglomerates have a porous texture. But more work is needed to determine i f it is easy to break agglomerates. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 77 5.4 Mechanical Activation of Heterogeneous Sol-Gel Precursors The results in Section 5.2 showed that completely crystallized spinel was not obtained until the calcination temperature was raised to 1200°C when the precursor (MA-5) was prepared by a heterogeneous method. A s presented in Section 5.3, completely crystallized spinel was achieved when magnesium and aluminum hydroxides were milled for 20 h and calcined at 1000°C. In this section, a spinel precursor was prepared using the same process as in Section 5.2 and then grinding of the precursor was carried out, aiming at decreasing the calcination temperature for completely crystallized spinel formation. 5.4.1 Phase and Morphology Changes of the Precursors during Milling The preparation of the spinel precursor by heterogeneous sol-gel process and grinding of the precursor were described in Sections 4.1.2 and 4.1.4. Figure 5.4-1 shows that boehmite was formed and magnesia remained in the precursor (MA-8) prepared using the heterogeneous sol-gel process. The precursor phases are almost the same as the M A - 5 precursor prepared using the same method (Figure 5.2-1). There is one difference in preparing M A - 5 and M A - 8 precursors: the median size of the M g O powder was 3.09 pm for M A - 5 and 1.82 pm for M A - 8 . The finer M g O power was used here to examine the influence of the particle size on the. hydration degree. A s shown in Figure 5.4-1, crystalline magnesium hydroxide was not observed in the precursor, indicating that M g O was not converted to crystalline brucite during fluxing in isopropyl alcohol with water for 0.5 h. This is the same as shown in Figure 5.2-1 in which no brucite is noticeable in the M A - 5 precursor. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 78 O 3 CO (0 c 0) A O A - O A 5h 1 2 . - i 1 r 10 20 30 40 50 60 70 2 Theta (9) Figure 5.4-1. X R D pattern of M A - 8 precursor milled for various duration of time ' (O : boehmite; A : periclase). Comparing Figure 5.2-1 and Figure 5.4-1 reveals that the decrease in the size of M g O powders from 3.09 pm to 1.82 pm has no obvious effect on the composition of the precursors and the M g O powder with finer size (1.82pm) was not completely hydrated during fluxing in the alcohol and water solution with the presence of alumina sol. As mentioned in Section 5.2, the precursor phases are different from those using the same synthesis, but with nano-sized M g O as the starting material [10]. Figure 5.4-1 also depicts the influence of mechanical mil l ing on the crystallinity of the precursor. The peaks characteristic of periclase and boehmite are not much changed after 1 -3 h of mill ing. However, boehmite peaks are obviously weakened and broadened after 5h of milling, indicating that the lattice structure of boehmite was disordered. The peaks belonging to periclase are also broadened after 5 h of mill ing, implying that the crystal size of M g O was reduced and its lattice structure was deformed to some extent. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 79 Figure 5.4-2 shows the evolution of the morphology and particle size of the precursor mixture during mill ing. A s seen in Figure 5.4-2a, the particles in the as-prepared precursor are of 10-18 pm in size. After mill ing for 1 h, the particles were finely ground to 200-300 nm, but the aggregates were of ~ 10 pm in size. Additional mill ing did not show significant influence on the agglomerate size. A uniform aluminum and magnesium distribution in the precursor particles was observed with the E D X analysis, indicating that the M g O particles were uniformly covered by the boehmite formed during precursor preparation. Figure 5.4-2 S E M micrograph of the M A - 8 precursor after mill ing for (a) 0 h, (b) 1 h, (c) 3 h and (d) 5 h. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 80 As discussed in Section 5.3.4, the particle size of the staring materials during mill ing was decreased and the ingredients were well mixed and brought into tight contacts; the starting materials underwent partly deformation and amorphization. Thus, the starting materials were brought in higher energy state and the diffusion distance of M g 2 + and A l 3 + ions was decreased. Then, it can be expected that the formation of spinel would be enhanced and the temperature for the pure-phase spinel formation lowered. 5.4.2. Dehydroxylation Behavior of the Precursors Figure 5.4-3 shows the influence of mechanical activation on dehydroxylation and spinel formation in the precursor. Figure 5.4-3a indicates that the endothermic peak of the as-prepared precursor at 80°C corresponded to the removal of physically adsorbed water molecules from the micro-pores of boehmite [108]. The endothermic peak at 190°C in Figure 5.4-3 corresponds to the decomposition temperature of aluminum tri-hydrate [Al (OH)3] , indicating that [Al (OH)3] was also formed in the precursor in an amorphous state. The endothermic peak at 440°C was due to the dehydroxylation of boehmite. The two endothermic peaks of the as-prepared precursor in the temperature range of 50-200°C in Figure 5.4-3a became one endothermic large valley centered at about 130°C, when the precursor was milled for 1-5 h, indicating that chemically bonded water in aluminum hydroxide and the physically adsorbed water were "mixed up" during mill ing. The valley temperatures were 122°C, 127°C and 133°C when the precursor was milled for 1, 3, and 5 h respectively, presumably due to the increasing bonding between M g O and hydroxyls from aluminum hydroxide and free water. The reason why the valley temperature increased wi l l be further discussed later. The weight loss of the as-prepared precursor under 100°C was 5.8% (Figure 5.4-3b); in contrast, the precursor milled for 1, 3 and 5 h exhibited Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 81 the weight losses of 3.7, 3.0 and 2.2%, in the temperature range up to 100°C, confirming that physically adsorbed water was partially transformed into chemically bonded to the precursors. o X CO X o C LU [ 5 b—— \ A! 3H A / l h -1/ / A Oh / (a) 0 300 600 900 1200 1500 Temperature (fiC) 300 600 900 1200 Temperature (2C) Figure 5.4-3 (a) D T A and (b) T G patterns (under a heating rate of 10°C/min) of the milled precursors. To avoid evaporation of adsorbed water during and after grinding, the vial was air tight during grinding and the vial was not opened after the grinding until the vial was cooled to room temperature. A s seen from Figure 5.4-3H, the total weight loss of the precursor at 1200°C is 27.1%, while the total weight losses of the precursor milled for 1, 3 and 5 h are 26.9, 26.7 and 26.5%, respectively, indicating that there was minor water evaporation after mill ing. The decrease in total weight loss of the precursor with the mil l ing time could be Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 82 ascribed to the dehydroxylation of amorphous aluminum tri-hydrate [Al (OH)3] and the formation of boehmite [A lOOH] during mill ing [32, 60] . Figure 5.4-4 shows the mill ing influence on the activation energy of water evaporation and dehydroxylation of the precursors below 200°C. For the as-prepared precursor, the activation energy for the evaporation of the physically adsorbed water is 49 kJ/mol and the activation energy of the reaction A l ( O H ) 3 - » A l O O H + H 2 0 is 58 kJ/mol. When the precursors were milled for 1 h, the above two dehydration processes merged into one (Figure 5.4-3a), and the activation energy was 43 kJ/mol. This value is obviously lower than 58 kJ/mol, the activation energy for the reaction A l ( O H ) 3 - » A l O O H + H 2 0 . 15.4 15.0 ZT 14.6 5 14.2 I-i f 13.8 13.4 13.0 h r 2 E a (kJ/mol) o 0 0.998 49 o 0 0.994 58 • 1 0.984 43 A 3 0.998 48 X 5 0.994 84 2.0E-03 2.3E-03 2.5E-03 2.8E-03 3.0E-03 1ATm Figure 5.4-4 Activation energy of the dehydroxylation of the precursor below 300°C as a function of milling time (O: physically adsorbed water; Ochemically bonded water in Al(OH) 3 ) . Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 83 It was reported that free water was generated from dehydroxylation of gibbsite during mechanical mil l ing [51], and the dehydroxylation of kaolinite shifted to lower temperatures with mill ing time [59], indicating that hydroxyls in the starting materials were "loosened" during mill ing. In this work, the dehydroxylation temperature of A l ( O H ) 3 —» A l O O H + H 2 0 was decreased from 190°C to ~ 130°C with mechanical activation of the precursor, (Figure 5.4-3a), implying the similar impact of grinding on the hydroxyls in the precursor. As a result, the reaction A l ( O H ) 3 —> A l O O H + H 2 0 in the milled sample exhibits lower activation energy than in the as-prepared precursor. As shown in Figure 5.4-4, the activation energy for dehydration was increased when the mill ing time was prolonged from l h to 3 h and 5 h. It was estimated [109] that the pressure generated by the high-energy mill ing ranges between 3.30 and 6.18GPa [58]. This high-pressure impact not only decreases the particle size and increases reaction surface (generating de-bonding) of the reactants, but also leads to tight contacts of the reactants and creates re-bonding between the reactants [61]. It is assumed that aluminum hydroxide and the free water were closely bonded to M g O during prolonged mill ing, resulting in the higher activation energy of dehydration of the precursors. The endothermic peak at 439°C of the as-prepared precursor (Figure 5.4-3a) was due to the dehydroxylation of boehmite. After the precursor was ground for 1, 3 and 5 h, the dehydroxylation temperatures were 389, 378 and 369°C, indicating that the hydroxyls of boehmite in the precursor became loosely bonded when the precursor was exposed to grinding. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 84 5.4.3. Phase Evolution during Calcination The exothermic peak at 1100°C of the as-prepared precursor (Figure 5.4-3a) is due to the phase transformation of y alumina to a alumina [109]; the peak at 1256°C corresponds to the massive formation of spinel in Figure 5.4-3a. After the precursor was milled for l-5h, the alumina transformation peak disappeared, and the temperature corresponding to the spinel massive formation was lowered with mill ing time. For example, the spinel formation temperature was 846°C when the precursor was subjected to 5 h of mill ing, which is 410°C lower than the temperature of massive spinel formation for the as-prepared precursor. As shown in Figure 5.4-1, the intensity of boehmite peaks decreased with grinding time, indicating that the aluminum hydroxide was structurally disordered, its crystal size decreased, and as a result, the reactivity of the precursor increased through mechanical milling. Moreover, during high-energy mill ing, the aluminum hydroxide and periclase were homogeneously mixed and the number of close particle contacts increased. Accordingly, the mechanical activation of the precursor favored the solid-state reaction of alumina and magnesia and the temperature of massive spinel formation was considerably lowered Figure 5.4-5 exhibits spinel formation versus mill ing times and calcination temperatures. Figure 5.4-5a shows that M g O was the only crystalline phase when the as-prepared precursor was subjected to calcination at 800°C. Aluminum hydroxide decomposed after heat treatment at this temperature (refer to Figure 5.4-3a) and was present in an amorphous state, while M g A l 2 0 4 did not form yet. After calcination at 900°C, spinel started to form, along with crystallization of 8-alumina and y-alumina. The spinel peaks increased with calcination temperature in the range of 900°C to 1100 and the peaks belonging to M g O , 8-alumina and y-alumina did not decrease significantly. After calcination at 1280°C, the spinel Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 85 peaks were obviously higher than those after calcination at 1100°C, while the peaks corresponding to 8-alumina and y-alumina disappeared and the peaks of a-alumina occurred. (a) * * 1280°C o 1100°CA5 1 iooo^JcJU^ 800°C . A 15 25 35 45 2 Theta (°) 55 65 15 25 35 45 2 Theta (°) 55 65 (C) * «J j 1 1 0 0 ° C J I U JU j A . i o o o ° c _ 15 25 35 45 55 2 Theta (°) 65 a C (d) 11100°C L l 1000°C JU Li j ^ 900' . 800°C 15 25 35 45 55 2 Theta (") 65 Figure 5.4-5 X R D pattern of the precursor milled for (a) 0 h, (b) 1 h, (c) 3 h, and (d) 5 h and fired at different temperatures (y: y-alumina, 8: 8-alumina, O: a-alumina, A : magnesia, >k: spinel). Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 86 These results were in agreement with the D T A results in Figure 5.4-3a that the exothermic peak at 1110° was ascribed to the transformation of 8-alumina and y-alumina to a-alumina. However, both magnesia and alumina were present after calcination at 1280°C, indicating that monolithic M g A l 2 0 4 spinel was not obtained after calcination at this temperature. This result is in agreement with the report [10], stating that monolithic M g A l 2 0 4 spinel was not obtained after the precursor synthesized using M g O of 10 nm in the heterogeneous process was thermally treated at 1200°C. In contrast to the as-prepared precursor (Figure 5.4-5a), 8-alumina and y-alumina did not appear in the precursors milled for 1-5 h (Figure 5.4-3b, 3c and 3d) after calcination in the temperature range of 900-1100°C, which is in agreement with T G results discussed above. One more remarkable difference between the ground precursors and the as-prepared precursor is that the spinel formation occurred at 800°C in the milled precursors, while the peaks corresponding to spinel did not appear in the as-prepared precursor until fired at 900°C, indicating that mechanical activation of the precursor lowered the incipient temperature of spinel formation. Figure 5.4-5a and Figure 5.4-5b shows that after calcination at 1280°C, the peaks belonging to a-alumina were present in the as-prepared precursor and the precursor milled for 1 h, indicating that monolithic M g A l 2 0 4 phase was not formed in the as-prepared precursor and the precursor milled for 1 h after calcination at the temperature. For the precursor milled for 3 h (Figure 5.4-4c), monolithic M g A l 2 0 4 phase was obtained at 1000°C, which is in accordance with the massive spinel formation temperature of 967°C of the precursor milled for 3h in Figure 5.4-3a. Phase-pure M g A l 2 0 4 phase was formed at 900°C in the precursor milled for 5 h (Figure 5.4-4c), which is in agreement with the massive Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 87 spinel formation temperature of 846°C, as shown in Figure 5.4-3a. These results indicate that mill ing of the precursor for 3 and 5 h decreased the calcination temperature from over 1280°C to 1000°C and 900°C for completely crystallized spinel formation. As described before, the calcination temperature of 900°C for the formation of completely crystallized spinel is relatively low compared to reported synthesis approaches, such as coprecipitation (1100°C) [105], heterogeneous process (>1200°C) [10] and freeze drying (1200°C) [27]. This temperature is also lower than the results in this study using coprecipitation (1100°C), heterogeneous process (1200°C) and mechanochemical technique (1000°C) with magnesium and aluminum hydroxides as the starting materials. It is worth noting that the peaks of magnesia appear in the as-prepared precursor and the precursors milled for 1 and 3 h and calcined at 800°C (Figure 5.4-4a, 4b and 4c), but are not noticeable in the precursor milled for 5 h and calcined at 800°C (Figure 5.4-5d). For the precursor milled for 5 h and calcined at 800°C, MgAl 2C>4 spinel peaks were very weak (Figure 5.4-5d) and the temperature of massive spinel formation is above 846°C (Figure 5.4-3a), indicating that magnesia in the sample should be present in an amorphous-like state after calcination at 800°C. These results implied that the lattice structure of magnesia introduced into the precursor was severely distorted due to the mechanical mill ing for 5 h. Figure 5.4-6 shows that the activation energy of spinel formation from the as-prepared precursor is 688 kJ/mol; in contrast, the activation energy for the precursor after 5 h of grinding is 468 kJ/mol, indicating that the mechanical mil l ing decreased the activation energy of spinel formation. A s discussed in Sections 5.3.4 and 5.4.2, high-energy grinding distorted the lattice structure of the precursor and brings the reactants into intimate contacts on an atomic scale, resulting shortened inter-diffusion distance for M g 2 + and A l 3 + ions. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 88 Therefore, the energy barrier for ion diffusion in the precursors is lowered and the activation energy for spinel formation is decreased. This hypothesis also provides an explanation for the decreased temperature of spinel formation in the milled precursors CM E 18.0 17.6 17.2 16.8 5 2 . 16.4 ' 16.0 15.6 15.2 h r 2 O 0 0.995 • 1 0.999 A 3 0.994-X 5 0.992 E a (kJ/mol) 688 570 505 468 6.0E-04 7.0E-04 8.0E-04 9.0E-04 1.0E-03 1/TM Figure 5.4-6. Activation energy of spinel formation as a function of mil l ing time Figure 5.4-7 shows the influence of the mechanical mill ing on the morphology and particle size of the powders calcined at 900°C. A s seen in Figure 5.4-2a, the large particles in the powder without mill ing were of -20 pm in size and are agglomerated. After mill ing for 1-5 h, the large aggregates were of 5-10 pm in size and the small ones were of l -3pm (Figure 5.4-2c - Figure 5.4-2d). It is noted that the particle size of the powder with 5h of mill ing is obviously not smaller than that with l h of mill ing. The observation confirms that Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 89 prolonged mil l ing does not result in powders with a finer particle size because fracturing and welding of the particles would reach a balance during milling for a certain time [107]. Figure 5.4-7. S E M micrograph of the precursor milled for (a) 0 h, (b) 1 h, (c) 3 h and (d) 5 h and fired at 900°C for 5h. 5.5 Spinel Synthesis by Sol-Gel Precipitation In this section, a novel chemical route which combines sol-gel and precipitation processes is presented. In this method, aluminum isopropoxide [Al(-0-i-C3H7)3] is first hydrolyzed to form a sol; at the same time magnesium acetate aqueous solution is homogeneously mixed in the alumina sol. Then the pH value is raised to form magnesium Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 90 hydroxide precipitate within the alumina sol. It is expected that homogeneous mixing of the magnesium and aluminum-bearing components in the sol-gel-precipitation (SGP) process benefits formation of magnesium aluminate spinel at a relatively low temperature. This hypothesis is confirmed experimentally. 5.5.1 Phase Evolution during Calcination The preparation of the spinel precursor by sol-gel-precipitation process was described in Section 4.1.5. The reactions of the precursor during heat treatment can be linked to the respective sections on the T G - D T A curves in Figure 5.5-1. The endothermic peak at about 100°C was due to the evaporation of water. The two endothermic peaks at 230°C and 400°C were due to the dehydration of aluminum and magnesium hydroxides, respectively [40, 60] and the two endothermic peaks were accompanied by weight loss (13 wt% and 19 wt% respectively), confirming dehydration at these temperatures. The low exothermic peak at about 830°C was probably due to the massive formation of magnesium aluminate spinel. 0 200 400 600 800 1000 1200 Temperature (°C) Figure 5.5-1 T G - D T A curves of the magnesium aluminate spinel precursor. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 91 5 15 25 35 45 55 2 Tneta (a) Figure 5.5-2 X R D pattern of the magnesium aluminate spinel precursor. Crystallite magnesium-bearing hydroxides are typically detected in the precursors prepared using the precipitation technique [13, 105, 110]. It is however seen from Figure 5.5-2 that the precursor was completely amorphous, indicating that the magnesium hydroxide precipitates were also in an amorphous state. It may be hypothesized that the magnesium hydroxide precipitates were too small to be detected by X R D , thus magnesium-bearing hydroxide homogeneously mixed with alumina sol. The aluminum and magnesium hydroxides in the precursor were completely dehydrated at 600°C (Figure 5.5-1) and, the material left was also amorphous (Figure 5.5-3). Spinel began to form at 700°C and phase-pure spinel was obtained after firing at 900°C, lower than by coprecipitation (1100°C, Section 5.1), heterogeneous sol-gel (1200°C, Section 5.2) and mechanochemical process (1000°C, Section 5.3). The temperature for phase-pure spinel formation obtained in this work (900°C) is also significantly lower than those achieved by other Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 92 researchers using coprecipitation (1100°C) [105], freeze drying (1200°C) [27] and heterogeneous sol-gel (above 1200°C) [10]. 3 cd a Figure 5.5-3 Spinel crystallization from the precursor after calcination for 5h at different temperatures. It is accepted that the calcination temperature is a key parameter controlling agglomeration of the powder [111]. Consequently, the relatively low calcination temperature for phase-pure spinel formation obtained in this work through the novel Sol-Gel-Precipitation (SGP) route is beneficial to avoid hard agglomeration of the calcined powders. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 93 5.5.2 Morphology of the Calcined Powders Figure 5.5-4 shows the change of the surface area, crystallite size, particle size of the precursor and the calcined powders as a function of the calcination temperature. -The precursor had a specific surface area of 486m 2/g. The specific surface area of the calcined powders decreased to 105m 2/g at 900°C, and 42m 2 /g at 1000°C, corresponding to equivalent spherical particle sizes of 28.0 nm, and 39.9 nm respectively. The crystallite size of the powder calcined at 900°C and 1000°C, calculated from X R D data, is 13 nm and 27 nm respectively. 500 250 100 200 400 600 800 1000 1200 1400 Temperature (9C) Figure 5.5-4 Temperature dependence of specific surface area and crystal size of spinel powders (—O— : particle size from specific surface area; — A — : crystal size from X R D data). Figure 5.5-5 shows that the spinel powder after calcination at 1000°C had a crystallite size varying between 20nm to 80nm, roughly in agreement with the X R D data. The specific Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 94 surface area of the calcined powder decreases sharply from 700°C to 1000°C and slowly from 1000°C to 1300°C. However, the crystallite size calculated from both the specific surface area and X R D data increases slowly with the temperature in the range 700°C-1000°C and relatively quickly from 1000°C to 1300°C. These results indicate that the optimum calcination temperature should be in the range 900°C to 1000°C, in order to obtain a powder with relatively high specific surface area and small crystallite size, i.e. with high sintering reactivity. lOOnm Figure 5.5-5 T E M micrograph of spinel prepared by sol-gel precipitation, after firing for 5h at 1000°C. 5.5.3 Particle Size and Agglomeration of the Calcined Powders In Section 5.1, magnesium aluminate spinel precursor was prepared using coprecipitation with a completely aqueous suspension and the obtained precursor was severely agglomerated Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 95 after drying. In Section 5.2, magnesium aluminate spinel precursor was prepared using a heterogeneous sol-gel method, in which the volume ratio of alcohol to water of the precursor suspension was 3.18:1. The precursor was agglomerated after drying, but to an obviously less degree than the precursor synthesized using the coprecipitation approach. For comparison, the volume ratio of alcohol to water was 12:1 in this sol-gel precipitation process. In fact, magnesium acetate tetrahydrate was dissolved into the least possible amount of water to obtain a high volume ratio of alcohol to water in the precursor suspensions. In this case, the concentration of water molecules on the neighboring precursor particles would be decreased due to the dominant presence of alcohol [48]. As a result, the powder had high specific surface areas and small equivalent particle sizes after calcination at 900-1000°C (Figure 5.5-4), indicating that hard agglomeration did not occur after the dried precursor and calcined powder. Figure 5.5-6 shows the change of the agglomerate size distribution with the calcination temperature. The agglomerate size ranged from 0.5 pm to 50 p m after calcination at 900°C; the median agglomerate size was -360 times the average particle size calculated from B E T specific surface data or X R D data. However, the B E T results show that the calcined powder has a large specific surface area and the particle size is equal to about that of two crystals, demonstrating that the calcined powder possesses a porous network with crystals having limited contact areas. It may be expected therefore that the agglomeration of the calcined powder is relatively weak and the agglomerated particles can be easily broken. This is further verified through the mill ing experiments (see below). It is interesting to note from Figure 5.5-6 that the minimum agglomerate size was about 400 nm after calcination at 1200°C and about 500 nm after calcination at 900°C. Also , about Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 96 10% of the powder calcined at 900°C is below 2 pm, while about 15% of the powder calcined at 1200°C is under the same size. The higher fraction of the particles below 2 pm after firing at 1200°C should be attributed to the larger shrinkage of the small agglomerates in the precursors. 0.1 1 10 100 1000 0.1 1 10 100 1000 Agglomerate size (um) Agglomerate size (um) Figure 5.5-6 Temperature dependence of particle size distribution of the spinel powders; (a) cumulative volume versus particle size, (b) differential volume versus particle size. Figure 5.5-6b shows that the largest agglomerates in the powder calcined at 900°C are about 40-50 pm in size. However, large agglomerates ranging from 50 pm to 400pm were generated when the calcination temperature was increased to 1100 or 1200°C, indicating that the size of agglomerates increased with the temperature. A s mentioned above, the strength of the interparticle bonds strongly depends on the calcination temperature [111]. As a result, relatively weak agglomerates in the powder calcined at 900°C, are relatively easy to break; in contrast, harder agglomerate would form after calcination at higher temperatures (up to 1100°C or 1200°C) and it would be more difficult to break the harder agglomerates. Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 97 Consequently, as concluded before, the optimum calcination temperature of the precursor ranges from 900°C or 1000°C. Agglomerate size (um) Agglomerate size (um) Figure 5.5-7 Influence of milling on particle size distribution of the spinel powders calcined at 900°C; (a) cumulative volume versus particle size, (b) differential volume versus particle size. Figure 5.5-7 shows the influence of mil l ing on the agglomerate size distribution of the spinel powder. The minimum agglomerate size was decreased from about 500 nm to about 100 nm, and the median size from 10 pm to 600 nm, indicating that large agglomerates of the spinel powder were broken during the mill ing process. This result also confirms that large agglomerates in the spinel powder calcined at 900°C were soft agglomerates of fine particles. However, compared to the crystallite size determined through X R D and observed through T E M and the equivalent particle size from specific surface areas, the milled powder still contained agglomerates of the primary crystals. 5.6 Summary Magnesium aluminate spinel precursor prepared by coprecipitation tends to agglomerate severely during drying and calcination. Incorporation of combustible ingredients, such as Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 98 carbon black and polyvinyl alcohol ( P V A ) , into the precipitate precursors prevented the direct contacts of the precursor particles and reduced agglomeration of the precursors during drying. These combustible ingredients leave behind a continuous pore network when they burned off during calcination at 700-800°C. It is inferred that the incorporation of the combustible ingredients into the precursor suspensions is an effective approach to prevent agglomeration of the precursors during drying and thus ultrafine spinel powders can be obtained after burning of the combustible ingredients. When the combustible ingredients were added after the formation of the precursor precipitates, the combustible ingredients did not obstruct the direct contact between the magnesium and aluminum-bearing components and consequently did not hamper the spinel formation. Phase-pure spinel was achieved after the precursor prepared using a heterogeneous sol-gel process with aluminum isopropoxide and reactive magnesia (3.09 pm median) as starting materials was calcined at 1200°C. Spinel formation took place in one-step reaction from the precursor synthesized using the reactive magnesia. In contrast, spinel formed in two-step process from the precursor prepared using nano-sized magnesia and complete spinellization was not achieved at 1200°C. As a result, for phase-pure spinel formation at relatively low temperatures, it is better to use the reactive magnesia than the nano-sized magnesia. Spinel-coated magnesite was obtained from aluminum isopropoxide and magnesite powder (<45 pm) using the heterogeneous sol-gel approach when the precursor was calcined at 1200°C. The starting mixtures of magnesium and aluminum hydroxides were transformed into an amorphous state when the mil l ing time was extended up to 40 h. The mechanical mill ing decreased the particles from 6-15 pm to 200-300 nm and produced a uniform aluminum and Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 99 magnesium distribution in the agglomerates. Mechanical activation reduced the crystal size and disordered the lattice structure in the precursors. Completely crystallized spinel was obtained when the mixtures were milled for 20 h and calcined at 1000°C, while phase-pure spinel was not obtained when the mixtures without mill ing were calcined at the same temperature. Spinel was noticeably formed in the mixtures of brucite [Mg(OH) 2 ] and boehmite ( A l O O H ) subjected to 20 h of grinding and calcination at 400°C to 600°C. However, spinel was not formed when the mixtures of brucite [Mg(OH) 2 ] and amorphous magnesium hydroxide [A1(0H) 3] were subjected to the same mill ing and calcination processes. These results indicate that boehmite is preferable to amorphous aluminum tri-hydroxide for the spinel formation when the mixtures were subjected to the same mechanical and thermal treatments. Grinding of the precursor prepared by the heterogeneous sol-gel process decreased the A l ( O H ) 3 dehydroxylation temperature from 190°C to about 130°C. The grinding also reduced M g O crystal size, deformed its lattice structure and increased homogeneity in the precursor. Consequently, the activation energy for spinel formation was decreased from 688 kJ/mol for the as-prepared precursor to 468 kJ/mol for the precursor subjected to 5 h of milling. Grinding of the precursor lowered the incipient temperature of spinel formation from 900°C to 800°C, and the temperature of monolithic M g A l 2 0 4 spinel formation from >1280°C to 900°C. Spinel precursor was prepared using a novel Sol-Gel-Precipitation (SGP) method. The key feature of the technology is separation of the steps of alumina sol formation, from the step of magnesium hydroxide precipitation (simultaneous with alumina sol gelation). We Chapter 5 Synthesis of Magnesium Aluminate Spinel Powders 100 believe that such a process enhances the precursor homogeneity, thus (as supported by the experiments) allowing synthesis of phase-pure and truly nano-particulate spinel at temperatures as low as 800-900C. Isopropyl alcohol was used as the dominant liquid medium in the precursor suspension, to reduce the tendency for particle agglomeration. Soft agglomeration of the precursor after drying and the powder after calcination at 900°C was confirmed by the apparent crystallite size, specific surface area and equivalent particle size of the calcined powder. As a result, ultrafine spinel powder ( d 5 0 = 600 nm, specific surface area = 105 m 2 /g, crystallite size 13 nm) was obtained after calcination at 900°C. Phase-pure spinel was formed after the precursor was calcined at a relatively low temperature (900°C), demonstrating that a homogenous precursor was acquired using this novel Sol-Gel Precipitation (SGP) method. The relatively low calcination temperature would not induce strong interparticle bonding and hard agglomeration of the powder after calcination. Based on the relationship of specific surface areas, crystal size and particle size with calcination temperature, the optimum temperature for acquiring sinterable spinel powder is in the range of 900-1000°C. Higher temperatures of 1200°C or above generate larger agglomerates, which hinders the sintering process. Chapter 6 Sintering of Binding Systems for Refractory Castables 101 CHAPTER 6 SINTERING OF BINDING SYSTEMS FOR REFRACTORY CASTABLES 6.1 Bi-Modal Systems The formulations of the bimodal mixtures are shown in Table 4.2-2. Figure 6.1-1 shows that the mixture without any ultrafine powder is a typical bimodal particle system with a peak at -20 pm and a shoulder at -3 pm. The physical properties of the compacts after drying and firing are presented in Appendix B . The properties of the compacts are discussed in the following three sections. Particle size (urn) Figure 6.1-1 Particle size distribution of bimodal mixture AO. 6.1.1 Influence of UF-Spinel-1 Figure 6.1-2 shows the influence UF-spinel-1 on the green density, bulk density, apparent porosity, linear shrinkage and compressive strength of the samples after heat treatment at 816°C, 1280°C and 1450°C. When the amounts of the ultrafine powder increased from 1% to 10%, the green density of the compacts increased almost linearly (Figure 6.1-2a), indicating that the ultrafine spinel particles filled the voids between the Chapter 6 Sintering of Binding Systems for Refractory Castables 102 larger particles [71]. Accordingly, increasing the amounts of the ultrafine powder between 1% and 10% raised the bulk density and lowered the apparent porosity of the mixtures after firing at all three temperatures (816°C, 1280°C and 1450°C). Figure 6.1-3 confirms that noticeable voids exist between large particles in the sample without the addition of the ultrafine powder. 0 2 4 6 8 10 0 2 4 6 8 10 uf spinel-1 (%) uf spinel-1 (%) Figure 6.1-2 Influence of UF-spinel-1 on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bimodal mixtures after heat treatment. Chapter 6 Sintering of Binding Systems for Refractory Castables 103 Figure 6.1-2 also shows the influence of the firing temperature on the properties of the mixtures. After firing at 816°C, there is little linear shrinkage (Figure 6.1-2c) and the compressive strength (Figure 6.1-2d) is relatively low. These results indicate little sintering happening at this temperature. In comparison, higher bulk density, linear shrinkage and compressive strength and lower apparent porosity are obtained after firing at 1250°C and 1450°C, demonstrating that the sintering was improved at the temperatures. The increase in density, linear shrinkage and strength and decrease in apparent porosity from 816°C to 1280°C, are considerably lower than those from 1280°C to 1450°C, indicating that the firing temperature should be above 1280°C to accomplish substantial sintering of the mixtures. It is seen from Figures 6.1-2c and 6.1-2d that increasing addition of UF-spinel-1 up to 10% had no significant influence on linear change and cold crushing strength after firing at 816°C, 1280°C and 1450°C. This can be attributed to the fact that the large particles (peaked at 20 pm in Figure 6.1.1) formed the skeleton and controlled the sintering of the whole mixtures, while the ultrafine spinel powder (300 nm) contributed little to the sintering [71]. Figure 6.1-3 S E M micrograph of the spinel sample AO after firing at 1280°C. Chapter 6 Sintering of Binding Systems for Refractory Castables 6.1.2 Influence of UF-SpineI-2 1 0 4 4 6 8 uf spinel-2 (%) 0.0 -0.5 I -1.5 o (0 « -2.0 -2.5 -3.0 -•—•—•-10 -o— 816°Cx5h - • — 1280°Cx5h -&—1450°Cx5h 4 6 8 uf spinel-2 (%) 10 28 70 j r s <0 0. 60 5 A .c • 50 C tre 40 CO u> c 30 !E (0 3 i— 20 i U TJ O 10 O 816°Cx5h 1280°Cx5h 1450°Cx5h 4 6 8 uf spinel-2 (%) 10 (d) -816Cx5h -1280Cx5h -1450Cx5h -• *-0 * 4 6 8 uf spinel-2 (%) 10 Figure 6.1-4 Influence of UF-spinel-2 on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bimodal mixtures AO, A 2 4 to A2-5 after heat treatment. Figure 6.1-4 shows that the variation of the physical properties of the bimodal mixtures with the ultrafine powder amounts and the firing temperature. Comparing Figure 6.1-4 with Figure 6.1-2 reveals that the influence of UF-spinel-2 is almost the same as that of U F -Chapter 6 Sintering of Binding Systems for Refractory Castables 105 spinel-1. That is, the ultrafine spinel particles filled the voids between the larger particles and, as a result, raised the bulk density and lowered the apparent porosity of the mixtures after firing at all three temperatures (816°C, 1280°C and 1450°C). However, just as with UF-spinel-1, addition of UF-spinel-2 up to 10% contributed little to sintering of the compacts because the larger particles controlled the sintering of the whole systems. The firing temperature exerted the same influence on the sintering of the systems containing UF-spinel-2 as those containing UF-spinel-1. The linear shrinkage (Figure 6.1-4c) and the compressive strength (Figure 6.1-4d) indicate that little sintering happened at 816°C, and the bulk density, linear shrinkage and compressive strength and lower apparent porosity obtained after firing at 1250°C and 1450°C demonstrate that the sintering was enhanced at the temperature. It is inferred from Figure 6.1-4 that, just as in the systems containing UF-spinel-1 (Figure 6.1-2), the firing temperature should be above 1280°C to accomplish considerable sintering of the mixtures. 6.1.3 Influence of the In-House Prepared Spinel Powder It is seen from Figure 6.1-5a that the green density of the mixtures decreases with the amount of the in-house spinel powder MA-9-900 . A s shown in Figure 5.5-6, the particle size of the powder ranges from 0.5 pm to 50 pm after calcination at 900°C, though the crystallite size of the powder is -30 nm (Figure 5.5-5), indicating that the powder consists of agglomerates. The agglomerated particles contain voids and are unfavorable for the packing of the compacts. Chapter 6 Sintering of Binding Systems for Refractory Castables 106 2.40 2.35 E o ^ 2.30 | 2.25 CD 2.20 2.15 0.0 -0.5 (a) — • — 1280oCx5h^~~~~~ —ir— 1450°Cx5h ^ —*—Green 40 ~ -1.0 v D) £ - i . o o to «! -2.0 -2.5 -3.0 0 2 1 [ 1 1 • 4 6 8 10 MA-9-900 (%) (c) i ^ ^ # _ —• — 1 2 8 0 ° C x 5 h — A — 1450°Cx5h A. A — — A A A __________ —A 0 2 \ I I 4 6 8 i 10 MA-9-900 (%) (b) 30 60.0 .1280°Cx5ri 1450°Cx5h 4 6 8 MA-9-900 (%) 10 4 6 8 MA-9-900 (%) 10 Figure 6.1-5 Influence of MA-9-900 addition on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bimodal mixtures AO, A3-1 to A3-5 after heat treatment. Because of the significant decrease in green density with the addition of the synthesized spinel powder, the bulk density and compressive strength of the mixtures after firing at 1280°C and 1450°C show a decrease and the apparent porosity exhibits an increase with the addition of the powder (Figure 6.1-5). It is noted in Figure 6.1-5c that linear shrinkage does not increase with the addition of the synthesized powder after firing at 1280°C and 1450°C, Chapter 6 Sintering of Binding Systems for Refractory Castables 107 suggesting that the sintering within agglomerates [71] [44] at these firing temperatures (Figure 5.5-6) did not significantly influence the shrinkage of the compacted mixtures. 1280°Cx5h -•—1450°Cx5h -A— Green 38 2 4 6 8 10 MA-g-900-milled (%) 0.0 -0.5 Ji -1.5 o &-(0 « -2.0 -2.5 -3.0 (c) —•—•-•1280°Cx5h •1450°Cx5h 2 4 6 8 MA-g-900-milled (%) 10 (b) 28 70.0 (0 | 60.0 {? 50.0 2 Z 40.0 c w 30.0 3 k. U 2 20.0 o O 10.0 • — 1 2 8 0 ° C x 5 h 1450°Cx5h 2 4 6 8 MA-9-900-milled (%) (d) •1280°Cx5h •1450°Cx5h 2 4 6 8 MA-9-900-milled (%) 10 10 Figure 6.1-6 Influence of MA-9-900-mil led addition on (a) bulk density, (b) apparent porosity, (c) linear change and (d) cold crushing strength of bimodal mixtures AO, A4-1 to A4-5 after heat treatment. On the contrary, the addition of the milled spinel powder increased the green density, the bulk density, linear shrinkage and compressive strength and decreased the apparent porosity of the mixtures after firing at 1280°C and 1450°C (see Figure 6.1-6). A s shown in Figure Chapter 6 Sintering of Binding Systems for Refractory Castables 108 5.5-7, the median particle size of the milled powder is much smaller than that of the synthesized powder without milling, indicating that the large agglomerated particles were broken into smaller particles during mill ing. As a result, the addition of the milled powder improved the packing and sintering of the mixtures at 1280°C and 1450°C. 6.1.4 Comparison of Different Spinel Powders As presented in the above three sections, sintering did not significantly happen in the mixtures after firing at 816°C. This section does not include the properties of the mixtures after firing at this temperature. 2.44 n : , 2.34 -I 1 1 , , , — 1 0 2 4 6 8 10 Ultrafine spinel (°/o) Figure 6.1-7. Influence of different ultrafine spinel powders on green density of bimodal mixtures (see Table 4.2-2) after heat treatment for 24h at 110°C. Figure 6.1-7 shows the influence of the three ultrafine spinel powders on the green density of the mixtures. The compacts with the three ultrafine spinel powders up to 3% exhibit almost the same the green density, indicating that the ultrafine powders impart the same influence to the green density. The addition of 5% of the in-house prepared spinel power and UF-spinel-2 Chapter 6 Sintering of Binding Systems for Refractory Castables 109 improved the green density of the compacts to the same degree, while the same amounts of UF-spinel-1 enhanced the green density to a less extent. When the addition of the ultrafine spinel was 10%, the order of promoting the bulk density of the compacts is UF-spinel-2 > in-house prepared spinel > UF-spinel-1. The difference arises from the fact that the former two ultrafine spinel powers have smaller particle sizes than the latter. ••— uf spinel-1 « — uf spinel-2 MA-9-900- milled — uf spinel-1 a — uf spinel-2 MA-9-900-milled 2 4 6 8 10 Ultrafine spinel (%) 2 4 6 8 Ultrafine spinel (%) 10 Figure 6.1-8 Influence of different ultrafine spinel powders on bulk density of bimodal mixtures (see Table 4.2-2) after heat treatment for 5h at (a) 1280°C and (b) 1450°C. After firing at 1280°C and 1450°C , the addition of MA-9-900-mil led increased the bulk density (Figure 6.1-8), linear shrinkage (Figure 6.1-10) and compressive strength (Figure 6.1-11) and decreased apparent porosity (Figure 6.1-9) to the highest degree, the addition of UF-spinel-1 to the lowest, and the addition of UF-spinel-2 in between. This should be ascribed to the difference in particle size, particle size distribution and specific surface area among the three ultrafine spinel powders. The specific surface areas of UF-spinel-1, U F -spinel-2 and MA-9-900-mil led are 14, 23 and 105m 2/g respectively. Obviously, the higher specific surface area of the ultrafine powders provides the compacts with higher driving sintering force. Chapter 6 Sintering of Binding Systems for Refractory Castables 110 36 32 (a) ••— uf spinel-1 m— uf spinel-2 * — MA-9-900- frilled 32 2 4 6 8 Ultrafine spinel (%) 10 (b) •—uf spinel-1 . uf spinel-2 « — MA-9-90O-rrilled 2 4 6 8 Ultrafine spinel (%) 10 Figure 6.1-9 Influence of different ultrafine spinel powders on apparent porosity of bimodal mixtures (see Table 4.2-2) after heat treatment for 5h at (a) 1280°C and (b) 1450°C. Figure 6.1-10 Influence of different ultrafine spinel powders on linear change of bimodal mixtures (see Table 4.2-2) after heat treatment for 5h at (a) 1280°C and (b) 1450°C. Chapter 6 Sintering of Binding Systems for Refractory Castables 111 Ultrafine spinel (%) Ultrafine spinel (%) Figure 6.1-11 Influence of different ultrafine spinel powders on compressive strength of bimodal mixtures (see Table 4.2-2) after heat treatment for 5h at (a) 1280°C and (b) 1450°C. 6.2 Tri-Modal Systems Figure 6.2-1 illustrates that the mixture D l - 1 had a wide particle size distribution (0.4-900 pm) with peaks at ~340pm, ~45pm and ~5pm. The physical properties of the compacts after drying and firing are listed in Appendix III. The properties of the compacts are discussed in the following four sections. Particle size (Mm) Figure 6.2-1 Particle size distribution of tri-modal mixture (Dl -1 in Table 4.2-3) Chapter 6 Sintering of Binding Systems for Refractory Castables 112 6.2.1 Influence of Ultrafine Alumina Figure 6.2-2a shows that the addition of UF-alumina (0.5 um median particle size) up to 5% increased the green density of the compacts, indicating that the ultrafine particles filled the voids between the large particles. As a result, the bulk density (Figure 6.2-2a) increased and the apparent porosity (Figure 6.2-2b) decreased with the addition after firing at 816°C, 1280°C and 1450°C. 2.75 '55 c o TJ m 2.70 2.65 2.60 x - 1 1 0 ° C o--816 sC 12809C ^-1450 9 C 2 4 uf alumina (%) 30 -o--816 2C - • — 1280SC • 14509C 2 4 uf alumina (%) (b) 2 4 uf alumina (%) 2 4 uf alumina (%) Figure 6.2-2 Influence of UF-alumina addition on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-1, D2-2 and D2-3 with 5% UF-spinel-2 after heat treatment. Chapter 6 Sintering of Binding Systems for Refractory Castables 113 After firing at 816°C, the linear shrinkage (Figure 6.2-2c) and the compressive strength (Figure 6.2-2d) were low, indicating the sintering happened to a limited extent. However, addition of UF-alumina seemed to promote the sintering at this temperature because compressive strength of the mixes with 2% and 5% the ultrafine powder was 23% and 46% higher than that of the mix without the ultrafine alumina, respectively. The linear shrinkage (Figure 6.2-2c) and compressive strength (Figure 6.2-2d) increased with the ultrafine alumina after firing at 1280°C and 1450°C, demonstrating its contribution to the sintering. Figure 6.2-3 Influence of UF-alumina addition on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D4-1, D4-2 and D4-3 with 5% UF-spinel-1 after heat treatment. Chapter 6 Sintering of Binding Systems for Refractory Castables 114 The influence of UF-alumina with 5% UF-spinel-1 (Figure 6.2-3) on the properties is similar to that of UF-alumina with 5% UF-spinel-2 (Figure 6.2-2). The comparison of the U F -spinel-1 and UF-spinel-2 with respect to their influence on sintering wil l be discussed later in this chapter. 6.2.2 Influence of UF-Spinel-2 The influence of UF-spinel-2 (Figure 6.2-4) on the properties is similar to that of U F -alumina with 5% UF-spinel-2 (Figure 6.2-2) or with 5% UF-spinel-1 (Figure 6.2-2). However, the major difference is that the addition of UF-spinel-2 increased the compressive strength of the mixtures after firing at 816°C. The mixes with 2% and 5% the ultrafine spinel had compressive strength of 7.29 and 11.80 M P a respectively, which are 2.5 and 4.1 times that of the mix without the ultrafine spinel (2.87 MPa), respectively. The results indicate that addition of the ultrafine spinel promoted the sintering at the temperature. As described the report [75], at 1500°C, volume diffusion dominates the sintering if the alumina bimodal systems is relatively coarse (30.5-33.5 pm), while grain boundary diffusion contribute to the sintering i f the systems are relatively fine (1-3 pm). In the tri-modal systems, the ultrafine particles not only fill in the voids between the larger particles, but also exist between the large particles. As a result, the ultrafine spinel particles (300 nm) between the large particles contributed to the sintering through grain boundary and surface diffusion at 816°C. As seen in Figure 6.2-4d, addition of 5% UF-spinel-2 increases the strength of the compacts after firing at 1280°C and 1450°C. However, the influence of 10% addition of UF-spinel-2 on the strength was alleviated at these temperatures. As a result, addition of UF-spinel-2 should not exceed 5% with regard to the cost effectiveness and sintering promotion at the temperature above 1280°C. This statement is supported by the changes in the bulk density and apparent results: the Chapter 6 Sintering of Binding Systems for Refractory Castables 115 bulk density increases and the apparent porosity decreases to a higher extent when the ultrafine spinel increases from 0% to 5%, but to a much lower extent when it increases from 5% to 10%. 32 T -0.2 DI C to co cu c -0.4 -0.6 -0.8 4 6 8 uf spinel-2 (%) 10 - o - 8 1 6 A C - • — 1 2 8 0 E C ^ — 1 4 5 0 9 C 4 6 8 uf spinel-2 (%) 10 27 a 8 1 6 9 C - » - 1 2 8 0 S C 1 4 5 0 S C (b) 4 6 8 uf spinel-2 (%) 10 8 1 6 9 C « — 1280 9 C A - 1 4 5 0 S C 4 6 8 uf spinel-2 (%) 10 Figure 6.2-4 Influence of UF-spinel-2 on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D l - 1 , D l - 2 and D l - 3 after firing at different temperatures. 6.2.3 Comparison of Different-Sized Ultrafine Spinels Figure 6.2-5 shows that, after heat treatment at 816°C, 1280°C and 1450°C, the compacts with 5% UF-spinel-2 have higher bulk density, linear shrinkage and compressive Chapter 6 Sintering of Binding Systems for Refractory Castables 116 strength, and lower apparent porosity than the compacts with 5% UF-spinel-1. This is attributed to the difference in the particle size of the two ultrafine spinel powders. UF-spinel-2 has finer particle size (0.2 pm median) and higher specific surface area (30+5 m7g) than UF-spinel-1 (0.3 pm median, 12 m 2/g). As described in Sections 2.4.1 and 2.4.2, the finer particles have free surface energy and higher driving force for sintering than the larger particles. Consequently, UF-spinel-2 promoted sintering to a higher extent than UF-spinel-1. 816 1280 1450 816 1280 1450 Firing temperature, °C Firing temperature, °C Figure 6.2-5 Influence of 5% of different ultrafine spinel on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-1 and D4-1 after firing at different temperatures. Chapter 6 Sintering of Binding Systems for Refractory Castables 117 Figure 6.2-6 compares the influence of 5% UF-spinel-2 and 5% UF-spinel-1 on bulk density, apparent porosity, linear shrinkage and compressive strength of the compacts with 2% U F -alumina. The difference in the influence of UF-spinel-2 and UF-spinel-1 with 2% UF-alumina (Figure 6.2-6) is similar to that without UF-alumina (Figure 6.2-5). However, when UF-alumina is increased to 5%, the difference in the influence of UF-spinel-2 and UF-spinel-1 (Figure 6.2-7) is diminished, because of the presence of 5% UF-alumina. 816 1280 1450 816 1280 1450 Firing temperature, °C Firing temperature, °C Figure 6.2-6 Influence of 5% of different ultrafine spinel on (a) B D , (b) A P , (c) P L C and (d) CCS of tri-modal mixes D2-2 and D4-2 with 2% UF-alumina after firing at different temperatures. Chapter 6 Sintering of Binding Systems for Refractory Castables 118 816 1280 1450 8 1 6 1 2 8 0 1 4 5 0 Firing temperature, °C Firing temperature, °C Figure 6.2-7. Influence of 5% of different ultrafine spinel on (a) B D , (b) A P , (c) P L C and (d) C C S of tri-modal mixes D2-3 and D4-3 with 5% UF-alumina after firing at different temperatures. 6.2.4 Contribution of Ultrafine Powders to Total Specific Surface Area It has been demonstrated in our experiments that the incorporation of the ultrafine powders is beneficial to sintering. In order to illustrate the contribution of the ultrafine powders to the sintering driving force of the mixtures, the specific surface area ratio of UF-spinel-2 to the experimental mixtures (Dl -1 , D1-2 and Dl-3) is presented in Figure 6.2-8. It is seen that the specific surface Chapter 6 Sintering of Binding Systems for Refractory Castables 119 area of the mixtures increases sharply with UF-spinel-2 (up to 2%). Addition of 5% UF-spinel-2 accounts for -96% of the total specific surface area of the mixture (Dl-2). Therefore, the addition of small amounts of the ultrafine powders increases the driving force for sintering. As seen from Figure 6.2-8, when the ultrafine spinel powder increases from 5% to 10%, its contributions to the total specific surface area of the mixtures is limited (from -96% to -98%). Therefore, as discussed in Section 6.1-2, addition of UF-spinel-2 up to 5% increased the sintering of the compacts after firing at 1280°C and 1450°C, but higher content up to 10% showed little further promoting effect on sintering (refer to Figure 6.2-4d). uf-spinel-2, % Figure 6.2-8 Contribution of UF-spinel-2 to the total specific surface area of the mixtures. However, the powders with high specific surface areas do not necessarily contribute to improving the sinterability of the mixtures. For example, the hydratable alumina has a Chapter 6 Sintering of Binding Systems for Refractory Castables 120 specific surface area of -200 m 2 /g [92], but its median particle size is 2.5 pm, indicating that this material is agglomerated with a porous structure. Fast sintering may happen in the agglomerates, but that fast sintering does not necessarily contribute to the sintering of the whole compacts [112]. Therefore, incorporation of the hydratable alumina with a high specific surface area did not enhance the sintering of the systems. This is the same case for the in-house spinel MA-9-900 without mill ing (refer to Section 6.1.3). In comparison, the ultrafine spinel powders like UF-spinel-2 (Figure 6.2-9a), UF-spinel-1 (Figure 6.2-9b) and the in-house spinel MA-9-900-mil led (Figure 5.5-5) are composed of loosely agglomerated particles, and they can be easily dispersed and uniformly distributed in the mixed powders. The uniformly distributed ultrafine particles, together with either ultrafine particles or larger particles, can form neck areas where the high vacancy concentration serves as the driving force for the mass transfer and the sintering of the whole compacts. 200 nm Figure 6.2-9 S E M Micrograph of (a) UF-spinel-1 and (b) UF-spinel-2. Chapter 6 Sintering of Binding Systems for Refractory Castables 121 6.3 Sintering Behavior of Two and Tri-modal Compacts 6.3.1 Systems without Ultrafine Powders It is seen from Figure 6.3-la that the tri-modal mixtures had higher green density than the bimodal mixtures. A s describe in Sections 6.2 and 6.3, the two mixtures without any ultrafine powder had little sintering at 816°C. Then the apparent porosity after firing at this temperature should represent the true apparent porosity of the green samples because the binder and chemical water were removed through the firing. The apparent porosity results demonstrate that the packing density in the tri-modal mixtures is higher than that in the bimodal mixtures [76]. When firing temperature was increased from 816°C to 1450°C, as shown in.Figure 6.3-la, the bulk density of the bimodal compacts increased 5.9% from 2.21 ± 0 . 0 l g / c m 3 to 2.34±0.01g/cm ; in contrast, the bulk density of the tri-modal compacts increased only 4.0% from 2.52±0.01g/cm to 2.62±0.02g/cm . A s described above, the bimodal compacts had higher porosity than the tri-modal compacts. Consequently, the particle rearrangement could happen easier in the bimodal compacts than in the tri-modal compacts during sintering, leading to the higher increase in the bulk density and linear shrinkage of the bimodal compacts than the tri-modal ones. It is shown in Figure 6.3-lc that the linear shrinkage of the bimodal mixtures was higher than that of the tri-modal mixtures after firing at temperature ranging from 816°C to 1450°C. As discussed above, the bimodal compacts had lower packing density than the tri-modal compact, and the particle rearrangement could happen easier in the bimodal compacts than in the tri-modal compacts during sintering. Therefore the linear shrinkage of the bimodal compacts was higher than that of the tri-modal ones after firing at the three temperatures. Chapter 6 Sintering of Binding Systems for Refractory Castables 122 110 816 1280 Temperature (°C) 1450 3 2.5 g aT 2 2 •i 1.5 1 0.5 0 ( C ) E3 Bi-modal B Tri-modal ^ , 816 1280 1450 Temperature (°C) 37 C? 35 C O S 33 o a. g 31 re Q. Q. ** 29 27 (b) H Bi-modal • Tri-modal 816 1280 1450 Temperature (°C) 70 S. 60 2 £ 50 D) I 40 in O) E 30 JZ CO B 20 u o 10 O (d) a Bi-modal • Tri-modal t l 816 1280 1450 Temperature (°C) Figure 6.3-1 Comparison between bimodal mix AO and tri-modal mix D l - 1 without ultrafine powder for (a) B D , (b) A P , (c) P L C and (d) C C S . As described above, the porosity of the bimodal compacts was higher than that of the tri-modal compacts after firing at temperatures from 816°C and 1450°C. This means that the compacts with higher porosity could have fewer particle contacts than the compacts with lower porosity. Consequently, the compacts with a larger number of particle contacts would have higher sintering than those with a smaller number of particle contacts, because sintering happens at the particle contacts that impart strength to the whole compacts. Chapter 6 Sintering of Binding Systems for Refractory Castables 123 6.3.2 Systems with Ultrafine Powders The tri-modal compacts with 5% UF-spinel-1 (Figure 6.3-2) and UF-spinel-2 (Figure 6.3-3) exhibited higher green density, bulk density and compressive strength, and lower apparent porosity and shrinkage after firing between 816°C and 1450°C than the bimodal compacts, respectively. Comparing the compressive strength in Figure 6.3-ld and Figure 6.3-2d reveals that the compressive strength of the tri-modal compacts with 5% UF-spinel-1 was 10.2% and 4.3% higher than that of the tri-modal compacts without the ultrafine spinel after firing at 1280°C and 1450°C respectively; in contrast, the compressive strength of bimodal compacts with 5% ultrafine spinel was 3.6% and 1.3% higher than that of the bimodal compacts without the ultrafine spinel after firing at 1280°C and 1450°C respectively. In a similar manner, the compressive strength of the tri-modal compacts with 5% U F -spinel-2 ultrafine spinel (Dl-2) (Figure 6.3-3d) was 37.8% and 12.9% higher than that of the tri-modal compacts without the ultrafine spinel (Dl-1) (Figure 6.3-ld) after firing at 1280°C and 1450°C respectively; in contrast, the compressive strength of bimodal compacts with 5% UF-spinel-2 ultrafine spinel (A2-4) (Figure 6.3-3d) was 13.3% and 2.2% higher than that of the bimodal compacts without the ultrafine spinel (AO) (Figure 6.3-ld) after firing at 1280°C and 1450°C respectively. These results indicate that addition of ultrafine powders into the tri-modal compacts is more effective in increasing strength than in the bimodal compacts. Chapter 6 Sintering of Binding Systems for Refractory Castables 124 2.8 £ 2.6 \ o "B) in g 2.2 I 2 m 1.8 1.6 (a) ES Bi-modal • Tri-modal 110 816 1280 Temperature (°C) 1450 2.5 « 2 H R3 •= 1-5 H (0 I <u 1 0.5 0 (c) H Bi-modal • Tri-modal 816 1280 1450 Temperature (°C) Figure 6.3-2 Comparison between bimodal UF-spinel-1 for (a) B D , (b) A P , (c) P L C and 35 £ 33 £• in o §. 31 c tu «j a. Q- 29 27 (b) I • Bi-modal • Tri-modal 816 1280 1450 Temperature (°C) 70 -i Pa) 60 s 50 O) c 0) k_ 40 0) at c 30 -!E V) 3 k_ 20 u •D O 10 o 0 (d) s Bi-modal • Tri-modal tl 816 1280 1450 Temperature (°C) mix A1-4 and the tri-modal mix D4-1 with 5% (d) C C S . As shown in Figure 6.3-4, the ultrafine particles are partly distributed between the larger particles and partly in the voids formed by the larger particles in the bimodal compacts; in comparison, more ultrafine particles are scattered between the larger particles and fewer ultrafine particles in the voids formed by larger particles. Accordingly, the addition of the same amount of the ultrafine spinel increases the "effective" contacts, referring to those between the large particles, more in the tri-modal compacts than in the bimodal compacts. In other words, the ultrafine spinel particles filled in the voids do not contribute to the Chapter 6 Sintering of Binding Systems for Refractory Castables 125 effective contacts. It is proposed that only the sintering of the effective contacts add to the strength of the whole compacts. Therefore, the same amount of the ultrafine powder increases strength more in tri-modal systems than in the bimodal ones because the same addition of the ultrafine powders would produce more effective contacts in the tri-modal systems than in the bimodal ones. 3.0 - i 2.8 CO™"* E u 2.6 I? 2.4 -'in c cu -TJ _: 2.2 3 CD 2.0 1.8 (a) l Bi-modal • Tri-modal II 1 110 816 1280 Temperature (°C) 1450 3 2.5 «T 2 is •i 1.5 - C CO k_ CO 1 cu 1 c • 0.5 0 ( C ) E S Bi-modal • Tri-modal 816 1280 Temperature (°C) 1450 35 £. 33 _• CO o I 31 c 2 CO a Q. 29 < 27 (b, 1 II 1 Q Bi-modal • Tri-modal 816 1280 1450 Temperature (°C) 70 -I co MP 60 .—. 50 u> c 2 40 CO O) c 30 !E CO 3 20 O TJ 10 O o 0 (d) Q Bi-modal • Tri-modal 816 1280 1450 Temperature (°C) Figure 6.3-3 Comparison between bimodal mix A2-4 and tri-modal mix D l - 2 with 5% U F -spinel-2 for (a) B D , (b) A P , (c) P L C and (d) C C S . Chapter 6 Sintering of Binding Systems for Refractory Castables 126 Figure 6.3-4 Schematic particle packing in (a) bimodal and (b) tri-modal compacts with ultrafine powder It is seen from Figures 6.3-1 to 6.3-3 that the influence of the ultrafine powders on sintering is related to the firing temperatures. For the bimodal compact, the addition of 5% UF-spinel-1 increased the compressive strength by 3.6% and 1.3% after firing at 1280°C and 1450°C respectively (compare A l - 4 and AO), and the addition of 5% UF-spinel-2 increased the compressive strength by 13.3% and 2.2% after firing at 1280°C and 1450°C respectively (compare A2-4 and AO). Similarly, for the tri-modal compacts, the addition of 5% U F -spinel-1 increased the compressive strength 10.2% and 4.3% after firing at 1280°C and 1450°C respectively (compare D4-1 and D l - 1 ) ; and the addition of 5% UF-spinel-2 increased the compressive strength by 37.8% and 12.9% after firing at 1280°C and 1450°C respectively (compare D l - 2 and D l - 1 ) . These results demonstrated that, for both the bimodal systems and tri-modal systems, addition of 5% ultrafine spinel powders increased strength of the compacts to a higher extent at lower firing temperature (1280°C) and to a lower extent at higher temperature (1450°C). Chapter 6 Sintering of Binding Systems for Refractory Castables 127 6.3.3 Discussion on Sintering Evaluation Normally, higher densification or shrinkage of compacts with relatively uniform particle size distribution indicates a higher degree of sintering, and.accordingly the strength increase is proportional to the shrinkage [113]. However, this relationship was not observed for the compacts with wide particle size distributions. The bimodal compacts with 5% UF-spinel-1 (A 1-4) (Figure 6.3-2c) and without the ultrafine spinel (AO) (Figure 6.3-lc) displayed linear shrinkage of 2.44% and 2.55% respectively after firing at 1450°C; in comparison, the tri-modal compacts with 5% the ultrafine spinel (D4-1) (Figure 6.3-2c) and without the ultrafine spinel (Dl-1) (Figure 6.3-lc) exhibited linear shrinkage of 0.43% and 0.46% respectively after the same heat treatment. However, as displayed in Figure 6.3-ld and Figure 6.3-2d, the tri-modal compacts had higher compressive strength than the bimodal compacts whether with 5% ultrafine spinel (Figure 6.3-2d) or without the ultrafine spinel (Figure 6.3-ld). These results demonstrate that the strength is not proportional to the shrinkage when comparing the systems with different particle size compositions. Comparing the bimodal and tri-modal compacts with 5% UF-spinel-2 ultrafine spinel (A2-4 and D l - 2 in Figure 6.3-3c and Figure 6.4.3d) supports the above statement. The linear shrinkage of the bimodal compacts with 5% UF-spinel-2 (2.54%) is higher than that of the tri-modal compacts with 5% UF-spinel-2 (0.50%) after firing at 1450°C. However, the compressive strength of the former compacts (56.8MPa) is lower than that of the latter ones (65.36MPa) after the same heat treatment. Therefore, the shrinkage parameter seems not to completely represent the solid-state sintering degree of the multi-component systems. The linear shrinkage in multi-modal systems is not totally due to mass transport from the grain boundary to the necks between the particles, but possibly also due to particle Chapter 6 Sintering of Binding Systems for Refractory Castables 128 rearrangement because of the presence of the pores. A s discussed before, the bimodal systems had considerably higher porosity than the tri-modal systems, with or without the ultrafine spinels, after firing at 816-1450°C. The higher linear shrinkage of the bimodal systems possibly results mainly from particle rearrangement because the presence of the higher porosity facilitates particle rearrangement. This linear shrinkage would not add up to the strength because only the mass transport to the necks of the above defined "effective contacts" contributes to the strength. As discussed in Sections 6.3.1 and 6.3.2, the tri-modal systems have lower porosity and more effective contacts than the bimodal systems. Therefore, the tri-modal systems undergo less shrinkage resulting from particle rearrangement and higher strength development resulting from mass transfer to the necks of the effective contacts, than the bimodal systems. Therefore, it is proposed here that the change in strength can better represent the solid-state sintering of the, multi-component systems than the linear shrinkage. 6.3.4 Sintering Kinetics The linear shrinkage of the bimodal compacts with constant heating rates of 2°C/min and 4°C/min is shown in Figure 6.3-5. Addition of 5% UF-spinel-1 (A 1-4), UF-spinel-2 (A2-4) and the in-house spinel (A4-4) shifted the onset of shrinkage to lower temperatures and increased the total shrinkage at 1500°C. Similarly, addition of 5% UF-spinel-2 (D2-1), U F -spinel-1 (D4-1) and UF-alumina (D5-1) decreased the incipient shrinkage temperature of tri-modal compacts and raised their total shrinkage at 1500°C (Figure 6.3-6 and Table 6.3-1). These results indicate that addition of the ultrafine powders increased the sintering of the bimodal and tri-modal compacts because the ultrafine particles have higher specific surface area and, as a result, higher sintering driving force. Chapter 6 Sintering of Binding Systems for Refractory Castables 129 0.5 0 T -0-5 • -1 at g -1-5 t " 2 8 -2.5 e •J -3 -3.5 -4 (a) \X AO \ ^ A 1 - 4 ^ A 2 - 4 A4-4 njn 1000 1200 1400 Temperature (°C) 0.5 i 0 o? -0.5 0) U) c ra r -1 o ra s c -1.5 -2 -2.5 (b) N AO ^ \ A1-4 1 N A2-4 < A4-4 1000 1200 1400 Temperature (°C) Figure 6.3-5 Linear shrinkage of bimodal mixtures at heating rates of (a) 2°C/min and (b) 4°C/min; AO (without ultrafine powder), A l - 4 (5% UF-spinel-1), A2-4 (5% UF-spinel-2) and A4-4 (5% in-house spinel MA-9-900-milled). -1 i i ; i i ; ; ! -0.7 4 , , , , , H 900 1100 1300 1500 900 1100 1300 1500 Temperature (°C) Temperature (°C) Figure 6.3-6 Linear shrinkage of tri-modal mixtures at heating rates of (a) 2°C/min and (b) 4°C/min; D l - 1 (without ultrafine powder), D2-1 (5% UF-spinel-2), D4-1 (5% UF-spinel-1) and D5-1 (5% in-house spinel MA-9-900-milled). Chapter 6 Sintering of Binding Systems for Refractory Castables Table 6.3-1 Summary of sintering kinetics 130 Mix Heating Linear Sintering Apparent code rate shrinkage at shrinkage onset activation (°C/min) 1500°C(%) ( ° Q energy (kJ/mol) AO 2 3.387 1061 144 ± 6 4 2.078 1067 A l - 4 2 3.614 3025 143 ± 6 4 2.129 1056 A2-4 2 3.709 1027 142 ± 5 4 2.140 1053 A4-4 2 3.749 1026 141 ± 5 4 2.151 1051 Dl-1 2 0.755 945 141 ± 3 4 0.531 961 D2-1 2 0.906 921 130 ± 3 4 0.602 937 D4-1 2 0.893 924 131 ± 2 4 0.597 943 D5-1 2 0.889 925 133 ± 2 4 0.592 945 0.001 0.0001 I . . . . 5 6 7 8 9 6 7 8 9 104/T(K'1) Figure 6.3-7 Constant-rate-of-heating plots of (a) bimodal mixtures and (b) tri-modal systems at a heating rate of 2°C/min. Figure 6.3-7 shows the plot of In T t / (A^r /z<>)] versus 1/T of the bimodal and tri-modal compacts at a heating rate of 2°C/min. These results were also obtained for the bimodal and Chapter 6 Sintering of Binding Systems for Refractory Castables 131 tri-modal compacts at a heating rate of 4°C/min, but the plots are not shown here because they are very similar to the plots in Figure 6.3-7. Sintering activation energy was calculated from the slopes of the plots in Figure 6.3-7 and is listed in Table 6.3-1. The addition of U F -spinel-1, UF-spinel-2 and the in-house spinel decreases the sintering activation energy of the bimodal systems by 0.7%, 1.4% and 2.1% respectively. In comparison, the addition of the three ultrafine powders decreased the apparent activation energy of the tri-modal systems by 7.1% (UF-spinel-1), 7.8% (UF-spinel-2) and 5.7% (the in-house spinel). These results suggest that the sintering of the bimodal systems was not considerably influenced by the addition of the ultrafine powders because, as discussed before, the incorporation of the ultrafine particles does not significantly increase the number of the "effective" contacts for sintering. On the other hand, the sintering of the tri-modal systems was influenced by the addition of the ultrafine powders because the ultrafine particles add to the number of the "effective" contacts for sintering. It is seen from Table 6.3-1 that incipient shrinkage temperatures in the bimodal systems are about 100°C higher than the tri-modal systems. It is also seen from Table 6.3-1 that the bimodal compacts without ultrafine powder have higher apparent sintering activation energy (AO, 144±6kJ/mol) than the tri-modal compacts without ultrafine powder (Dl -T , 141±3kJ/mol). Similarly, the bimodal compacts with 5% UF-spinel-1 (A 1-4) and UF-spinel-2 (A2-4) have higher sintering energy (143±6kJ/mol and 141±5kJ/mol respectively) than the tri-modal compacts with 5% UF-spinel-1 (D4-1, 131±2kJ/mol) and UF-spinel-2 (D2-1, 130±3kJ/mol) respectively. These results indicate that the tri-modal compacts have lower sintering activation energy than the bimodal compacts. These results are in consistence with that addition of ultrafine powders into the tri-modal compacts is more effective in increasing Chapter 6 Sintering of Binding Systems for Refractory Castables 132 sintering than in the bimodal compacts. As described before, there are more effective particle contacts in the tri-modal compacts than in the bimodal compacts and, accordingly there are more necks to which materials move from the surface or grain boundary. As a result, the initial sintering rate in tri-modal compacts is higher than in the bimodal compacts. It must be noted that the sintering activation energy described above is apparent or effective sintering activation. For the grain-boundary diffusion and volume diffusion, the activation energy values is equal to 2 and 3 times the apparent activation energy respectively [73]. Because it could not be determined i f grain-boundary diffusion or volume diffusion dominated the sintering of the compacts in this study, apparent activation energy is used here. As proposed by Young and Cutler [73], surface diffusion effects can be recognized in the C R H data. When surface diffusion occurs in sintering, a neck between particles is formed through material transport. As a result, the material flow from the grain boundary to the neck is impeded until a temperature is reached at which mass transport due to the grain boundary or volume diffusion predominates. Because of the lengthened diffusion paths and larger neck radius, the rate of shrinkage due to grain-boundary or volume diffusion is retarded until a higher temperature is reached. The net effect is a steepened initial slope in the C H R plot (see Figure 6.3-8). The shape of curves in Figure 6.3-7 is similar to the one with surface diffusion in Figure 6.3-8. That is, material transport by surface diffusion occurred before grain-boundary and/or volume diffusion in all the bimodal and tri-modal compacts. It is seen from Figure 6.3-7 that addition of the ultrafine powders decreased the temperatures where the grain-boundary diffusion or volume diffusion overwhelms the surface diffusion. The addition of the ultrafine powders on sintering mechanisms (to determine i f grain-boundary or volume diffusion is the Chapter 6 Sintering of Binding Systems for Refractory Castables 133 predominant mass transport mechanism) of the powder compacts with a wide size particle distribution is worth further investigation. 1/T Figure 6.3-8. Schematic representation of influence of surface diffusion on constant-rate-of-heating data plot [73] 6.2.5. Summary After heat treatment at 1280°C and 1450°C, the tri-modal compacts showed higher bulk density (2.66, 2.70 g/cm respectively) and compressive strength (28, 65 M P a respectively) than the binary compacts (2.26, 2.38 g/cm 3 respectively, and 22, 57 M P a respectively) because the number of effective particle contacts in the tri-modal compacts is higher than in the binary ones. The addition of ultrafine powders increases the number of effective contacts more in the tri-modal compacts than in the bimodal. As a result, we observed enhanced sintering and strength development at 1280°C and 1450°C. The tri-modal compacts had lower incipient sintering temperatures (937°C) and sintering activation energy (130+3 Chapter 6 Sintering of Binding Systems for Refractory Castables 134 kJ/mol) than the bimodal compacts (1053°C and 142±5 kJ/mol respectively) (These values refer to samples containing 5% UF-spinel-2 and heated at 4°C/min). For both the binary and tri-modal systems, the addition of the ultrafine alumina (0.5 pm median) and spinel (<0.3 pm median) up to 10% increased the green density and the bulk density and decreased the apparent porosity of the compacts after heat treatment from 110°C to 1450°C because the ultrafine spinel particles filled the voids formed by the large particles. However, the addition of the above ultrafine powders up to 10% did not promote sintering of the binary compacts after firing at 1280°C and 1450°C because the ultrafine particles did not increase the effective contacts significantly and the sintering was controlled by the large particles which formed the skeleton of the compacts. The tri-modal compacts have higher bulk density and lower apparent porosity because packing density in the tri-modal mixtures is higher than that in the bimodal mixtures. The higher linear shrinkage of the binary compacts fired at 816°C and 1280°C than the tri-modal compacts is related to the easier particle rearrangement in the bimodal compacts than in the tri-modal ones. As a result, the relationship between the strength and shrinkage for the powder compacts consisting of uniform particle size distribution does not apply to the powder compacts with wide particle size distributions. It is proposed that the solid-state sintering of compacts with multi-modal size distributions is better evaluated using strength data than using shrinkage/densification data. This hypothesis assumes that compressive strength is weakly dependent on flaw size distribution in the test samples. The particle size of the ultrafine powders is a critical parameter in enhancing the sintering of powder compacts with a wide particle size distribution (0.2-90 pm). The order of promoting sintering with the same amounts (up to 10%) of the ultrafine spinel powders at Chapter 6 Sintering of Binding Systems for Refractory Castables 135 1280°C and 1450°C is: the synthesized spinel MA-9-9'00-milled (0.6 pm median, 105 m2/g)> ultrafine spinel (0.2 pm median, 30 m 2/g) > ultrafine spinel (0.2 pm median, 12 m 2/g), because the specific surface area of the powders decreases in the same order. The higher specific surface area and smaller particle size of the ultrafine powders provides the compacts with higher driving force for sintering. Based on the same reason, the addition of 5% the ultrafine spinel (0.2 pm median) increased the sintering of the tri-modal compacts at 816°C, but the equal amount of the ultrafine alumina (0.5 pm) and the ultrafine spinel (0.3 pm), respectively, did not increased sintering at the same temperature. The agglomeration of the ultrafine powders affects both the packing and sintering of the mixture compacts. The addition of the in-house processed spinel powder with large agglomerates reduced the green density of the compacts, accordingly decreased the bulk density and compressive strength and increased the apparent porosity after firing at 1280°C and 1450°C. On the contrary, the addition of the milled in-house prepared spinel powder with smaller aggregates increased the green density, the bulk density, linear shrinkage and compressive strength and decreased the apparent porosity of the mixtures after firing at the two temperatures. A small amount (5%) of the ultrafine spinel powder (200 nm) accounts for -96% of the total surface area of the mixture. The optimum addition of ultrafine powders (<0.5pm median particle size) should be around 5% with regard to the cost effectiveness and sintering promotion. For both the bimodal systems and tri-modal systems, addition of 5% ultrafine spinel powders increased strength of the compacts to a higher extent at lower firing temperature (1280°C) and to a lower extent at higher temperature (1450°C), indicating that addition of ultrafine powders could be more effective in enhancing the strength at relatively low temperatures than at higher temperatures. Chapter 7 Spinel Bonded Basic Castables 136 CHAPTER 7 SPINEL BONDED BASIC CASTABLES This part of the work was primarily to investigate the influence of various compositions of the binding systems studied previously on the strength development of basic castables with hydratable alumina as the binder, and to develop basic castables with high strength after heat treatment at 110-1300°C. The hydration behavior of hydratable alumina in the presence of three forms of magnesia and the relationship between the strength and morphology and thermal decomposition behavior of the hydrates in the castables were also investigated. The bonding mechanism of hydratable alumina-bonded castables with the presence of magnesia after drying and firing was explored. These fundamental studies were aimed at raising the strength of hydratable alumina-bonded castables after drying and firing at intermediate temperatures (800-1000°C) using a combination of the hydratable alumina and magnesia in the binding systems. 7.1 Influence of Ultrafine Alumina on Castable Properties 7.1.1 Castables with Magnesia Aggregate After Firing at 816°C A s presented in Table 4.3-2, the contents of the ultrafine alumina (dso = 0.5 pm) in M l - 0 to M l - 6 with magnesia aggregates were varied from 0% to 6%. The influence of the ultrafine alumina on the bulk density, apparent porosity, linear change and modulus of the three mixes after heat treatment at different temperatures is shown in Figure 7.1-1. The increase in ultrafine alumina up to 6% slightly increased the bulk density (Figure 7.1-la) and decreased the apparent porosity (Figure 7.1-lb) of the samples after drying at 110°C and firing at 816°C, indicating that, as discussed in Chapter 6, the ultrafine particles filled the voids between the larger particles. Chapter 7 Spinel Bonded Basic Castables 137 2 4 6 Ultrafine alumina (%) 2 4 6 Ultrafine alumina (%) 22 20 £ 18 U) o g. 16 c a> o3 14 t D . Q . •* 12 10 (b) - • — 110°C- 816°C- • 1280°C 2 4 6 Ultrafine alumina (%) 2 4 6 Ultrafine alumina (%) Figure 7.1-1 Influence of UF-alumina on B D , (b) A P , (c) P L C and (d) M O R of magnesia-based castables M l - 0 , M l - 3 and M l - 6 after heat treatment at different temperatures. The castables after firing at 816°C have lower bulk density and higher apparent porosity than the castables after drying because the hydrates formed during curing of the castables were decomposed during firing at 816°C. Hydratable alumina reacts first with water to yield pseudoboehmite gels and then transforms to large amounts of bayerite and traces of boehmite [93, 94]. Aluminum hydroxides and magnesium hydroxide, formed from hydratable alumina and magnesia during curing of the castables, are dehydrated as follows: Chapter 7 Spinel Bonded Basic Castables 138 2 A 1 0 0 H 5 0 0 " S 5 0 ° C > A 1 2 0 3 + H 2 0 [60] M g ( O H ) 2 ~ 4 0 0 ° c ) M g O + H 2 0 [32] The above dehydration reactions indicate that the hydraulic bonding of the castables was lost during heat treatment at 816°C. As shown in Figure 7.1-ld, the strength of the castables with or without the ultrafine powder after firing at 816°C was low (~ 2MPa), demonstrating that the ceramic bonding was not produced at this temperature. In other words, no significant sintering occurred at 816°C. This result is in agreement with the conclusion in Chapter 6: no significant sintering happened at this temperature in the binding systems with the presence of the ultrafine alumina (0.5 pm). Figure 7.1-2 Micrograph of the matrix of the castable M l - 0 after heat treatment at (a) 110°C and (b) 816°C. Figure 7.1-2 shows the microstructure of M l - 0 before and after firing at 816°C. The matrix microstructure of the castable after firing at 816°C was almost the same as that after drying at 110°C. The particles in the matrix of the castable before and after firing at 816°C had very similar shapes, indicating that mass transport, i f any happened, was not significant Chapter 7 Spinel Bonded Basic Castables 139 during firing at 816°C. The observation confirms that the matrix of the castable underwent little sintering after firing at the temperature. After Firing at 1280°C When the firing temperature was raised to 1280°C, the bulk density of the castables (Figure 7.1-la) was lower than that after drying at 110°C and firing at 816°C and the apparent porosity (Figure 7.1-lb) was higher than that after drying at 110°C and firing at 816°C. This is attributed to the following reaction occurring in the castables during firing: M g O + A 1 2 0 3 > M g A l 2 0 4 The reactants magnesia and alumina have molar volumes of 11.26 and 25.55 cmVmole respectively and the product of M g - A l spinel a molar volume of 39.52cm 3/mole. A s a result, the formation of M g - A l spinel from magnesia and alumina is accompanied by an increase in structural volume of 2.71cm /mol. This value represents a 7.35% volume increase, equivalent to a 2.45% linear expansion [46]. Magnesia powder (including the magnesia aggregate) and alumina powder in the castables formed spinel during firing at 1280°C, which resulted in volume expansion and, as a result, decreased the bulk density (Figure 7.1-la) and increased the linear shrinkage of the castables (Figure 7.1-lc) after firing at 1280°C. As the castables are not homogenous in phase composition and particle size composition, the local volume expansion accompanying spinel formation could lead to cracks in the castables, resulting in increased porosity of the castables (Figure 7.1-lb) after firing at 1280°C. As shown in Figure 7.1-ld, the castables after firing at 1280°C had low strength. This is attributed to two factors. In the first place, as described above, the local volume expansion accompanying spinel formation could lead to cracks in the castables, which would damage the strength of the castables. Secondly, the difference in thermal expansion coefficients of Chapter 7 Spinel Bonded Basic Castables 140 the phases present in the castables could lead to cracking between the phases. The major phases in the castables are corundum (AI2O3), periclase (MgO) and magnesium aluminate spinel ( M g A l 2 0 4 ) . The mean thermal expansion coefficients ( 1 0 " 6 x ° C \ 0-1000°C) of the three phases are listed below [47]: A 1 2 0 3 : 8.8 M g O : 13.5 M g A l 2 0 4 : 7.6 During cooling, magnesia tends to shrink to a higher degree than alumina and spinel, resulting in intergranular stress. Intergranular cracking is generated when the stress exceeds the local shear or tensile strength between the grains/particles. This was evidenced by the microstructure of M l - 6 after firing at 1280°C (Figure 7.1-3), in which, cracks around the magnesia aggregate and between the aggregate and matrix are clearly visible. if' - * ^ SE'' • ' * Wp;5.5mm 2C'\ OkV x200 200um Figure 7.1-3 Micrograph of M l - 6 after firing at 1280°C; (a) cracks around magnesia aggregate and (b) cracks between magnesia aggregate and matrix. Chapter 7 Spinel Bonded Basic Castables 141 Influence of in-situ Spinel As presented in Table 4.3-1, M l - 6 contained 4% magnesia powder; in contrast, M2-6 did not include magnesia powder. The comparison of the properties of the two castables is shown in Figure 7.1-4. The two castables had similar properties after drying and firing at 816° and 1280°C, except that, after firing at 1280°C, M2-6 had noticeably lower linear expansion than M l - 6 after firing (Figure 7.1-4c). This is ascribed to the fact that the in-situ spinel formed from the magnesia powder and alumina powder in M l - 6 ; in contrast, such in-situ spinel did not form in M2-6 because the magnesia powder was not available. Figure 7.1-4d shows that the two castable M l - 6 and M2-6 had almost the same modulus of rupture after firing at 1280°C, implying a higher amount of the in-situ spinel in matrix of M l - 6 did not obviously deteriorate the strength of the castables. As seen in Figure 7.1-4c, M2-6 showed linear expansion after firing at 1280°C; this expansion is caused by the in-situ spinel formed from the magnesia aggregate and alumina in M2-6 . It is inferred that the in-situ spinel formation around the magnesia aggregates would generate cracks (Figure 7.1-3) and, as a result, decrease the strength of the castable. Additionally, the difference in thermal expansion coefficients between the aggregate and matrix could lead to cracking between the aggregate and matrix. Therefore, the influence of the in-situ spinel formation in the matrix on the strength of the M l - 6 was weakened. The differences in apparent porosity and modulus of rapture between M l - 6 and M2-6 after drying and firing at 816°C wi l l be discussed in Section 7.4 focusing on the effect of the reaction between hydratable alumina and magnesia on the strength of the castables. Chapter 7 Spinel Bonded Basic Castables 142 110 816 1280 Temperature (2C) 1.0 0.8 Si 0.6 0.4 v D) C <o (0 CD c 0.2 0.0 -0.2 (C) • 0% E3 4% 816 1280 Temperature (2C) (0 I 6 Q . a 4 (0 3 % 2 o 2 110 816 1280 Temperature (2C) (d) • 0% 0 4% 110 816 1280 Temperature (SC) Figure 7.1-4 Influence of magnesia powder (0-4%) on (a) BD, (b) A P , (c) P L C and (d) M O R of magnesia-based castables M l - 6 and M2-6 after firing at different temperatures. 7.1.2 Castables with Spinel Aggregate After Firing at 816°C As presented in Table 4.3-2, the content of the ultrafine alumina (d5o = 0.5 pm) in S l - 0 , SI-3 and SI-6 with spinel aggregates was varied from 0% to 6%. The properties of the three mixes after heat treatment at different temperatures with respect to the contents of the Chapter 7 Spinel Bonded Basic Castables 143 ultrafine alumina powder are shown in Figure 7.1-5. As seen from the figure, the addition of ultrafine alumina had no obvious influence on the bulk density, apparent porosity, linear change and flexural strength of the castables after firing at 816°C. A s discussed in Section 6.2.1, the addition of ultrafine alumina (dso = 0.5 pm) up to 6% does not significantly improves mass transport in the binding systems of the castables at 816°C because the temperature is not high enough to activate the mass transport. E O) in c o TJ m 2.8 0.4 — 0.2 v U) c CO u CO £ 0.0 (a) 110°C 816°C - _ — 1280°C -0.2 0 1 I I I I 2 4 6 Ultrafine alumina (%) (c) _ • _ 816°C 1280°C Tfc • 20 2 4 6 Ultrafine alumina (%) 18 &^  it o 16 o Q . •ent 14 Appai 12 -10 (b) -•— 110°C 816°C • 1280°C 2 4 6 Ultrafine alumina (%) 2 4 6 Ultrafine alumina (%) Figure 7.1-5 Influence of UF-alumina on (a) B D , (b) A P , (c) P L C and (d) M O R of spinel-based castables S1 -0, S1 -3 and S1 -6 after firing at different temperatures. Chapter 7 Spinel Bonded Basic Castables 144 As illustrated in Figure 7.1-5, the castables after firing at 816°C have lower bulk density, higher apparent porosity and lower flexural strength than the castables after drying; and the linear change of the castables after firing at 816°C is close to zero. This is, as discussed in the above section, attributed to the structural water release of hydration products and insignificant sintering of the castables at this temperature. After Firing at 1280°C It is seen from Figure 7.1-5 that the bulk density slightly increased, and apparent porosity and linear expansion slightly decreased with the ultrafine alumina contents of the castables after firing at 1280°C. Flexural strength of the castable was markedly higher than that of the castable without the ultrafine alumina and with 3% ultrafine alumina. The increased strength, and decreased linear shrinkage indicate that the addition of ultrafine alumina up to 6% promoted sintering of the castables at this temperature, which is confirmed by the well-sintered matrix and the boundary between the aggregate and the matrix of the castable with 6% ultrafine alumina (Figure 7.1-6). (a) P i * 1 A!IB Figure 7.1-6 S E M micrograph of (a) matrix and (b) the boundary between spinel aggregate and matrix in the castable S l - 6 with 4% M g O powder after firing at 1280°C. Chapter 7 Spinel Bonded Basic Castables 145 As shown in Figure 7.1-5, the castables fired at 1280°C had lower bulk density and higher linear expansion than those fired at 816°C. The variation of the properties is related to the volume expansion accompanying the in-situ spinel formation in the castables during firing at 1280°C because the castables contained 4% magnesia powder. In-situ Spinel Formation As presented in Table 4.3-2, SI-6 contained 4% magnesia powder; on the other hand, S2-6 included no magnesia powder. The comparison of the properties of the two castables is shown in Figure 7.1-7. It is seen that SI-6 has higher bulk density (Figure 7.1-7a), lower apparent porosity (Figure 7.1-7b) than S2-6 after drying and firing at 816°C, probably due to the difference in compositions between the two mixes. The two castables exhibit almost the same slight shrinkage after firing at 816°C (Figure 7.1-7c), indicating that little sintering happened at the temperature. Wi th regard to the higher strength of SI-6 than that of S2-6 after drying and firing at 816°C (Figure 7.1-7d), this wi l l be detailed in Section 7.4. The castables without magnesia powder shows shrinkage after firing at 1280°C, while the castables with magnesia powder displays' expansion after firing at the same temperature (Figure 7.1-7d). Noticeable shrinkage in S2-6 means that sintering happened in the matrix of the castables during firing at 1280°C. Sintering with the same extent should have taken place in S l - 6 with 4% magnesia. Therefore, the expansion of S l - 6 after firing at 1280°C is due to the in-situ spinel formation which was, as discussed before, accompanied by volume expansion. A s shown in Figure 7.1-6, cracks are present in the matrix and at the boundary between the aggregate and matrix in S l - 6 because of the volume expansion. In contrast, such cracks did not exist in the matrix and at the boundary of S2-6 without the in-situ spinel formation (Figure 7.1-8). Chapter 7 Spinel Bonded Basic Castables 146 Temperature (9C) Temperature (9C) Figure 7.1-7 Influence of magnesia powder on (a) BD, (b) A P , (c) P L C and (d) M O R of the spinel-based castables (S l -6 : with 4% magnesia powder; S2-6: without magnesia powder) after firing at different temperatures. Normally the castables with in-situ spinel formation has lower strength than castables without in-situ spinel [114] because the in-situ spinel formation produces volume expansion and cracks. However, as seen in Figure 7.1-7, S l - 6 with the in-situ spinel formation at 1280°C showed considerably higher strength (12.0 MPa) than S2-6 without the in-situ spinel formation (8.0 MPa) . The cause of the above results needs to be investigated. Chapter 7 Spinel Bonded Basic Castables Figure 7.1-8 S E M micrographs of (a) the matrix and (b) the boundary between spinel aggregate and matrix in the castable S2-6 without M g O powder after firing at 1280°C. 7.1.3 Influence of Aggregate Type Here discussed is the influence of the magnesia and spinel aggregates on the properties of the castables after firing at 1280°C. The variation of the castable properties with the aggregate types after drying at 110°C and firing at 816°C wi l l be addressed in Section 7 .4 . As presented in Tables 4.3-2 and 4.3-3, S l - 0 , S I - 3 and S l - 6 had spinel aggregates and M l -0, M l - 3 and M l - 6 had magnesia aggregates, and all the mixes contained 4% magnesia powder. Figure 7.1-9 shows that the castables with spinel aggregates had lower linear expansion and higher flexural strength than the castables with magnesia aggregates after firing at 1280°C. This difference arises from the following two aspects. For the castables with magnesia aggregate, the in-situ spinel forms not only in the matrix, but also on the surface of the magnesia aggregate during firing at 1280°C. In comparison, the in-situ spinel forms only in the matrix of the castables with the spinel aggregate. The in-situ spinel formed on the surface of the aggregate leads to the higher expansion of the castables with magnesia aggregate. A s presented in Tables 4.3-2 and 4.3-3, S2-6 and M2-6 Chapter 7 Spinel Bonded Basic Castables 148 had spinel and magnesia aggregates respectively and the mixes did not contain magnesia powder. From the compositions of the two mixes, it is expected that no in-situ spinel forms in S2-6, and the in-situ spinel forms on the surface of the magnesia aggregates of M2-6 . The castable with spinel aggregates showed shrinkage ( P L C = -0.17%) after firing at 1280°C, while the castable with magnesia aggregates showed expansion ( P L C = 0.12%) after firing at the same temperature. These results confirm that the in-situ spinel formed on the surface of the magnesia aggregates contributed expansion of the castables fired at 1280°C. Because in-situ spinel formation and the thermal expansion mismatch (discussed in detail below) between the aggregate and matrix, the magnesia-base castable M2-6 had lower strength ( M O R = 2.0 MPa) than the spinel-based castable S-6 ( M O R = 7.8 MPa) 1.6 1.2 0) D> C 03 0.8 03 0) C 0.4 0.0 (a) • Spinel aggregate • Magnesite aggregate 0 3 6 Ultrafine alumina (0.5 um median) (%) 16 14 12 S. 10 O (b) E3 Spinel aggregate • Magnesite aggregate 0 3 6 Ultrafine alumina (0.5 um median) (%) Figure 7.1-9 Comparison between castables with spinel aggregate S l - 0 , SI-3 and S l - 6 and magnesia aggregate M l - 0 , M l - 3 and M l - 6 fired at 1280°C for (a) P L C and (b) M O R . Chapter 7 Spinel Bonded Basic Castables 149 Another factor leading to the higher linear expansion of magnesia-based castables than the spinel-based castables is ascribed to the difference in thermal expansion coefficient between the two aggregates. The mean thermal expansion .coefficients of magnesia and magnesium aluminate spinel are 13.5xl0" 6 and 7.6xl0" 6 C C " 1 , 0 -1000°C) respectively [47]. For the spinel castables, the aggregates and matrix are composed of spinel and have the same amount of the shrinkage during cooling. As a result, the cracks were not generated between the aggregates and matrix (Figure 7.1-8b). Compared with the castables with spinel aggregate, the castables with magnesia aggregate would experience higher expansion during heating up because the magnesia aggregate has higher thermal expansion coefficient than the spinel aggregate. However, the same amount of shrinkage happens in the magnesia aggregate during cooling, while the castable would not undergo that amount of the shrinkage because the sintered spinel matrix would prevent the shrinkage of the castables to some extent. Consequently cracks would be generated between the aggregates and the matrix (Figure 7.1-6b). Therefore, the spinel-based castables exhibited lower linear expansion than the magnesia-based castables (see Figure 7.1-9a). Figure 7.1-9b show that the castables with spinel aggregate had higher flexural strength than the castables with magnesia aggregate after firing at 1280°C. A s discussed above, cracks are not generated at the boundary between the matrixes and aggregates in spinel castables (Figure 7.1-8b), while such cracks are produced between the aggregates and the matrix in magnesia castables (Figure 7.1-6b). The cracks between the spinel matrixes and magnesia aggregates would decrease the strength of the castables. Therefore, the castables with the spinel aggregate exhibited obviously higher strength than the castables with the magnesia aggregate. Chapter 7 Spinel Bonded Basic Castables 150 7.2 Influence of Ultrafine Spinel on Castable Properties 7.2.1 Influence of Ultrafine Spinel (dso=0.3 pm) The formulations of the castables containing UF-spinel-1 are presented in Table 4.3-4. Two different aggregates, fused magnesia and fused spinel, were used, and the amount of the ultrafine spinel varied between 0% and 5%. As shown in Figure 7.2-1, the flowability of magnesia-based castables (M3-0, M3-3 and M3-5) increased with the content of the ultrafine spinel powder. For the two spinel-based castables (S3-0 and S3-5), the flowability of S3-0 without the ultrafine spinel was 107% and the flowability of S3-5 with 5% the ultrafine spinel was 148%. The ultrafine spinel particles replace the water in the micro voids, resulting in increased flowability at the same level of water addition or decreased water demand for similar flowability [115]. 135 100 0 2 4 6 Ultrafine spinel (%) Figure 7.2-1 Influence of UF-spinel-1 on flowability of magnesia-based castables M3-0 , M3-3 and M3-5 with water addition of 5.1%. Chapter 7 Spinel Bonded Basic Castables 151 2.95 0.5 -1-0.3 (V D> C <o JZ 0.1 u s c Li -0.1 -0.3 -2 4 Ultrafine spinel (%) 2 4 Ultrafine spinel (%) 2 4 Ultrafine spinel (%) 2 4 Ultrafine spinel (%) Figure 7.2-2 Influence of UF-spinel-1 on (a) B D , (b) A P , (c) P L C and (d) M O R of magnesia-based castables M3-0 , M3-3 and M3-5 . The properties of the magnesia-based castables (M3-0, M3-3 and M3-5) after heat treatment with respect to the content of the ultrafine spinel powder are shown in Figure 7.2-2. After heat treatment at 110°C and 816°C, the addition of 3-5% ultrafine spinel increased the bulk density (Figure 7.2-2a) and apparent porosity (Figure 7.2-2b) because the ultrafine spinel filled the voids between the large particles. However, its addition had no obvious influence on linear change (Figure 7.2-2c) and modulus of rupture (Figure 7.2-2d) at 816°C, Chapter 7 Spinel Bonded Basic Castables 132 indicating that the addition of ultrafine spinel up to 5% did not significantly improve sintering in the binding systems of the castables at the temperature. As discussed in Chapter 6, the temperature of 816°C is not high enough to activate mass transport in the binding systems of the castables. Figure 7.2-2c shows that the linear change of the castables after firing at 816°C was close to zero, confirming that no significant sintering happened at this temperature. The castables after firing at 816°C had lower bulk density, higher apparent porosity and lower modulus of rupture than the castable after drying because, as discussed in Section 7.2.1, the hydrates were dehydrated and the ceramic bonding was not significantly formed at this temperature. After firing at 1280°C, M3-0 (without the ultrafine spinel) had lower bulk density (Figure 7.2-2a) and higher apparent porosity (Figure 7.2-2b) than after firing at 816°C. This is attributed to the volume expansion accompanying the in-situ spinel formed at 1280°C (see Figure 7.2-2c). The cracks were observed in M3-0 and M3-5 after firing at 1280°C (Figure 7.2-3) resulted from the volume expansion. Figure 7.2-3 S E M micrograph of magnesia based castables (a) without ultrafine spinel (M3-0) and (b) with 5% UF-spinel-1 (M3-5) after firing at 1280°C. Chapter 7 Spinel Bonded Basic Castables 153 As show in Figure 7.2-2, the addition the ultrafine spinel (0.3 pm median), especially from 3% to 5%, increased the strength at 1280°C and shifted the dimension change from expansion to shrinkage, indicating that the addition of the ultrafine spinel promoted sintering of the castables at this temperature. 110 816 1280 Temperature (SC) 110 816 1280 Temperature (9C) -0.3 Temperature (eC) 816 1280 110 816 1280 Temperature (2C) Figure 7.2-4 Influence of UF-spinel-1 on (a) B D , (b) A P , (c) P L C and (d) M O R of the spinel-based castables S3-0 and S3-5. Chapter 7 Spinel Bonded Basic Castables 154 For the castables based on spinel aggregate, as shown in Figure 7.2-4, the variations of the properties with the amount of the ultrafine spinel and the heat treatment temperatures present almost the same trend as the castables based on magnesia aggregate ( Figure 7.2-2). It is seen from Figure 7.2-5a that the matrix of the castable without the ultrafine spinel (0.3 pm) contained voids in the matrix after the castable was fired at 1280°C. On the other hand, the matrix of the castable with 5% the ultrafine spinel was relatively dense and well sintered at the temperature (Figure 7.2-5b). These observations illustrate that the addition of 5% the ultrafine spinel filled the voids in the matrix and improved the sintering of the matrix of the castable at the temperature. Figure 7.2-5 S E M micrograph of spinel-based castables (a) without ultrafine spinel (S3-0) and (b) with 5% UF-spinel -1 (S3-5) after firing at 1280°C. The major difference between the fused magnesia-based and fused spinel-based castables after firing at 1280°C was in regard to their linear change and flexural strength (Figure 7.2-6). As discussed before, the in-situ spinel could form on the surface of the magnesia aggregate, but not on the surface of the spinel aggregate. The in-situ spinel formation is Chapter 7 Spinel Bonded Basic Castables 155 accompanied by volume expansion, and could generate local micro-cracks in the castables. As a result, the magnesia-based castables had lower shrinkage (Figure 7.2-6a) and lower strength (Figure 7.2-6H) than the spinel-based castables after firing at 1280°C. Ultrafien spinel (%) Ultrafine spinel (%) Figure 7.2-6 Comparison between magnesia-based (M3-0 and M3-5) and spinel-based (S3-0 and S3-5) castables with different amounts of UF-spinel-1 after firing at 1280°C for (a) P L C and (b) M O R . 7.2.2 Influence of Ultrafine Spinel (dso=0.2 pm) The formulations of the castables containing UF-spinel-2 (0.2 pm median) are presented in Table 4.3-5. The properties of the spinel-based castables (S4-0, S4-2 and S4-5) after heat treatment at different temperatures are shown in Figure 7.2-7. As shown in Figure 7.2-7a and Figure 7.2-2b, the addition of the ultrafine spinel up to 5% increased the bulk density and decreased the apparent porosity of the castables after drying at 110°C and firing at 816°C and 1280°C, indicating that, as discussed before, the ultrafine spinel particles occupied the voids between larger particles. Chapter 7 Spinel Bonded Basic Castables 156 2.95 CO E 2.90 u * '55 2.85 c a TJ m 2.80 < 2.75 0 2 4 6 Ultrafine spine (d50=0.2um) (%); 0 2 4 6 Ultrafine spinel (d50=0.2um) (%) 20 -0.4 0 2 4 6 Ultrafine spinel (d50=0.2um) (%) (d) — • — 1 1 0 ° C - » — 8 1 6 ° C - * - 1 2 8 0 ° C 0 2 4 6 Ultrafine spinel (d50=0.2um) (%) Figure 7.2-7 Influence of UF-spinel-2 on (a) B D , (b) A P , (c) P L C and (d) M O R of spinel-based castables S4-0, S4-2 and S4-5. The strength of the castables after drying at 110°C was independent of the addition of the ultrafine spinel (0.2 pm) because the hydraulic bonding of the hydratable alumina is not influence by the presence of the ultrafine spinel. The shrinkage (Figure 7.2-7c) and flexural strength (Figure 7.2-2d) of the castables increased with the content of UF-spinel-2 after firing at 816°C, indicating that addition of the ultrafine spinel powder promoted sintering of Chapter 7 Spinel Bonded Basic Castables 157 the castables at this temperature. The addition of the ultrafine spinel considerably increased flexural strength (Figure 7.2-2d) of the castables after firing at 1280°C, indicative of the enhanced sintering of the binding systems of the castables at this firing temperature. The improved sintering of the castable through introduction of the ultrafine spinel powder was confirmed by the higher linear shrinkage of the castables (Figure 7.2-7c). It is noted in Figure 7.2-7d that the incorporation of 2% the ultrafine spinel substantially increased the strength by 77% from 8.8 M P a to 15.6 M P a after firing at 1280°C. However, the higher addition of the ultrafine spinel from 2% to 5% increased the strength by 9% from 15.6 M P a to 17.1 M P a . These results indicate that a small amount (2%) of U F -spinel-2 ultrafine spinel powder may be enough for improving the sintering of the matrixes of castables at the relatively low temperature. Figure 7.2-8 Micrographs of spinel based castables (a) without ultrafine spinel (S4-0) and (b) with 5% UF-spinel-2 (S4-5) after firing at 1280°C. The microstructures of the two castables after firing at 1280°C are shown in Figure 7.2-8. There are noticeable voids in the matrix of S4-0 and the boundary between the Chapter 7 Spinel Bonded Basic Castables 158 agglomerate and the matrix is not well sintered (Figure 7.2-8a). In contrast, the matrix of S4-5 is dense and well sintered. The microstructure shows that the addition of the ultrafine spinel UF-spinel-2 improved the sintering of the matrix of the castables after firing at 1280°C and explains why the castables with the ultrafine spinel had higher strength of the castables without the ultrafine spinel. It is especially worth noting that the shrinkage (Figure 7.2-7c) and flexural strength (Figure 7.2-2d) of the castables increased with the content of UF-spinel-2 (0.2 pm median) after firing at 816°C, indicating that addition of the ultrafine spinel powder promoted sintering of the castables at this temperature. However, the addition of UF-spinel-1 (0.3 pm median) did not influence the shrinkage (Figure 7.2-2c) and strength (Figure 7.2-2d) of the castables. These results imply that the sintering of the matrix of the castables at 816°C is related to the particle size of the ultrafine spinels. If the ultrafine spinel is fine enough (such as UF-spinel-2 with a median size of 0.2 pm), the firing temperature of 816°C can activate mass transport which leads to sintering; on the other hand, the mass transport cannot happen at this temperature i f the ultrafine spinel is not fine enough (such as UF-spinel-1 with a median size of 0.3 pm). Because 0.5 pm alumina is not fine enough, addition of this ultrafine powder did not increase the shrinkage (Figure 7.1-lc, and Figure 7.1-3c) and strength (Figure 7.1-ld and Figure 7.1-3d) of the castable after firing at 816°C. It was found in Chapter 6 that addition of UF-spinel -2 up to 10% in the tri-modal systems increased the linear shrinkage (Figure 6.2-4c) and compressive strength (Figure 6.2-4d) after firing at 816°C, indicating that addition of the ultrafine spinel promoted the sintering of the binding systems at the temperature. It was noticed that the mix with 5% U F -Chapter 7 Spinel Bonded Basic Castables 159 spinel-2 (0.2 pm median) had higher linear shrinkage (Figure 6.2-5c) and compressive strength (Figure 6.2-5d) than the mix with 5% UF-spinel-1 (0.3 pm median) after firing at 816°C, confirming that addition of the ultrafine spinel with finer particle size can improve sintering the binding systems at this temperature. 18 -i 16 Q. 14 5 <D 12 3 Q. 10 3 4— o 8 » 3 6 -3 TJ O 4 5 2 0 • 6% ultrafine alumina (0.5 um) • 5% ultrafine spinel (0.3 urn) m 5% ultrafine spinel (0.2 um) 816 1280 Temperature (9C) Figure 7.2-9 Influence of ultrafine powders on the strength of spinel-based castables S2-6, S3-5, S4-5 after heat treatment at 816°C and 1280°C. The influence of UF-alumina (0.5 pm median) and UF-spinel-1 (0.3 pm) and UF-spinel -2 (0.2 pm) on the strength of the spinel-based castables after firing at 816°C and 1280°C is shown in Figure 7.2-9. It is clearly seen that the castable with UF-spinel-2 had the highest strength, the castable with UF-alumina had the lowest strength and the castable with U F -spinel-1 had the strength in between. The results indicate that the finer the powder, the higher the strength of the castable, because the finer powder has higher specific surface area and driving force. It can be inferred that the sintering of the matrix of the castable can Chapter 7 Spinel Bonded Basic Castables 160 happen to a higher extent at 816°C or can happen at a lower temperature (<816°C) i f the particle size of the ultrafine powders is further decreased (<0.2 pm). This conclusion provides a direction for increasing the strength of the castables used at relatively low temperatures (<1300°C). 7.3 Hydration of Hydratable Alumina in the Presence of Magnesia It was observed in this work that castables containing magnesia aggregate and/or magnesia powder had higher strength than the castables without magnesia aggregate and magnesia powder after drying at 110°C and firing at 816°C. It has been reported that the hydration products of hydratable alumina with the presence of reactive magnesia is different from those without the presence of magnesia [39]. It was assumed that the presence of magnesia influenced the strength of the hydratable-alumina bonded castables. As described in Chapter 1, deadburnt and fused magnesia, rather than reactive magnesia, are normally used for production of refractory castables, and hydration of hydratable alumina in castables happens during curing at room temperatures and continues during drying/heating-up. In this work, hydratable alumina mixed with reactive magnesia, deadburnt magnesia and fused magnesia respectively were hydrated under the conditions representing curing and drying/heating-up of the castables in industrial vessels. The objective of this work is to find out the possible difference in the hydration products between the mixtures of hydratable alumina with reactive magnesia and those of hydratable alumina with deadburnt or fused magnesia powder at the temperatures of 20°C and 110°C, and understand why hydratable alumina-bonded castables with magnesia aggregate had higher Chapter 7 Spinel Bonded Basic Castables 161 strength after drying at 110°C and firing at 816°C than those with magnesium aluminate spinel aggregate [95]. 7.3.1. Hydration Hydratable Alumina and Different Forms of Magnesia The sample preparation for investigating the hydration of hydratable alumina with the presence of various forms of magnesia was described in Section 4.3.6. Figure 7.3-la shows that bayerite and boehmite were formed in hydratable alumina hydrated under Hydration-A and Hydration-B conditions. Compared with the sample hydrated under Hydration-A, the peak intensity of boehmite and bayerite was increased under Hydration-B, indicating that crystallization of bayerite and boehmite from the hydratable alumina was improved under hydration at the higher temperatures. 2 Theta (2) Temperature (9C) Figure 7.3-1 (a) X R D patterns and (b) T G curves of hydratable alumina hydrated under (A) Hydration-A and (B) Hydration-B ( O = boehmite, • = bayerite). Chapter 7 Spinel Bonded Basic Castables 162. The T G results (Figure 7.3-lb) show that the mass loss (from 150°C to 800°C) of hydratable alumina after Hydration-A is 14.2wt% (= total mass loss - mass loss up to 110°C), indicating that about 31% of hydratable alumina was hydrated under Hydration-A. D T A results (not shown here) illustrated that the dominant hydration product was bayerite. In Hydration-B, about 70% of hydratable alumina was hydrated and the dominant hydration products were bayerite and pseudoboehmite. The results indicate that hydratable alumina was not completely hydrated under both Hydration-A and Hydration-B. Figure 7.3-2 compiles X R D data for the hydration products of reactive magnesia, fused magnesia and deadburnt magnesia under the A and B conditions. Brucite is the predominant crystalline phase in the reactive magnesia under Hydration-A. In contrast, relatively weak peaks of brucite are visible in fused magnesia samples and no brucite peaks are noticeable in the deadburnt magnesia under the same hydration condition. The T G results for the same samples of magnesia (Figure 7.3-3a) show that the weight losses of the reactive, deadburnt and fused magnesia after Hydration-A were 27.7%, 4.4% and 3.2% respectively, confirming that the reactive magnesia was hydrated to a significantly higher extent than the deadburnt and fused magnesia. Crystallization of brucite from reactive magnesia was promoted under Hydration-B (Figure 7.3-2b), in comparison with the hydration product of reactive magnesia under Hydration-A (Figure 7.3-2a). Under Hydration-B, the brucite peaks are strong in the fused and deadburnt magnesia (Figure 7.3-2b), but periclase peaks are still noticeable in the two samples. The T G results (Figure 7.3-3b) show that the weight losses of reactive magnesia, fused magnesia and deadburnt magnesia hydrated under Hydration-B were approximately 28.6%, 26.9% and 24.8% respectively. Comparing with the theoretical water content of 30.9% in brucite (Mg(OH)2), the T G results indicate that the reactive magnesia was hydrated Chapter 7 Spinel Bonded Basic Castables 163 to a higher degree than the fused and deadburnt magnesia in Hydration-B. The X R D results (Figure 7.3-2) and T G results (Figure 7.3-3) indicate that reactive magnesia is more liable to hydration than the fused and deadburnt magnesia at room temperature and 110°C. 3 c CD (a) A 1 A |<c, • • 1 Ik . .5 . -1^ (A) • ° • • •2->. '5> c CD 10 20 60 70 10 20 30 40 50 2 Theta (9) 60 70 30 40 50 2 Theta (2) Figure 7.3-2 X R D patterns of samples hydrated under (a) Hydration-1 and (b) Hydration-2; (A) reactive magnesia, (B) fused magnesia and (C) deadburnt magnesia ( A = periclase, O = boehmite, • = brucite). 0> B> C A CO CO «J 2 5 0 -5 -10 -15 -20 -25 -30 -35 >~ Fused A—— \ Deadburnt —Beactive (a) 200 400 600. Temperature (°C) 800 CD c (0 CO CO CO 2 -5 -10 -15 -20 -25 -30 -35 \ Fused Deadburnt (b) Reactive 200 400 600 Temperature (°C) 800 Figure 7.3-3 T G curves for the thermal decomposition of various forms of magnesia after (a) Hydration-A and (b) Hydration-B conditions. Chapter 7 Spinel Bonded Basic Castables 164 7.3.2. Hydration Products of Mixtures of Hydratable Alumina and Magnesia Figure 7.3-4a exhibits that, under both hydration conditions, brucite [Mg(OH) 2 ] and hydrotalcite-like compound [(MgioAl 2)(OH)24(C03)(H 20)9] were formed in Mixture I. B y contrast, neither brucite nor hydrotalcite-like compound was formed in Mixture II and Mixture III under Hydration-A. The T G results (Figure 7.3-5a) indicate that the weight losses of Mixture II and Mixture III after Hydration-A were only 6.9%; in contrast, the weight loss of Mixture I was 37%. The higher weight loss of Mixture-I in Figure 7.3-5a resulted from the hydration of reactive magnesia (Figures 7.3-2a and 3a) and hydratable alumina (Figure 7.3-1) and formation of the hydrotalcite-like compound (Figure 7.3-4a). However, little deadburnt and fused magnesia were hydrated after Hydration-A (Figures 7.3-2a and 3a) and crystalline hydrotalcite-like compound was not detected in Mixtures II and III. Accordingly the weight losses of Mixtures II and HI were much lower than Mixture I after Hydration-A. 2 Theta (9) 2 Theta (») Figure 7.3-4 X R D patterns of samples (with A l 2 0 3 : M g O ratio = 16:84) hydrated under (a) Hydration-1 and (b) Hydration-2; (A) Mixture I, (B) Mixture II, (C) Mixture III ( • = (Mg 1 0 Al 2 ) (OH ) 2 4 (CO 3 ) (H 2 O )9 , 0 = ( M g 4 A l 2 ) ( O H ) 1 2 ( C 0 3 ) ( H 2 0 ) 6 , A = periclase, • = brucite). Chapter 7 Spinel Bonded Basic Castables 165 It could be inferred that the presence of sufficient Mg(OH)2 , not the presence of M g O , in the mixtures of hydratable alumina and magnesia during Hydration-A is necessary for formation of hydrotalcite-like compound. In Mixture I, the reactive magnesia was hydrated to a high degree during Hydration-A, and accordingly the hydrotalcite-like compound was formed. On the other hand, the fused and deadburnt magnesia in Mixtures II and III respectively were much less hydrated under the same hydration condition, and consequently the hydrotalcite-like compound was not observed. 5 0 _ -5 Co" ^ -10 CD c CO 15 o -20 CO 8 -25 -30 -35 -40 Mixture III Mixture II Mixture I (a) c IS u in m is 5 0 -5 -10 -15 -20 -25 -30 -35 -40 ^ • ^ ^ Mixture III \ \ ^ ^ Mixture II ^ ^ ^ ^ M i x t u r e I (b) 200 400 600 Temperature (°C) 800 200 400 600 Temperature (°C) 800 Figure 7.3-5 T G curves for the thermal decomposition of the magnesia-alumina mixtures after (a) Hydration-A and (b) Hydration-B conditions. Figure 7.3-4b shows that, after Hydration-B, hydrotalcite-like compound [(MgioAl2)(OH) 2 4(C03)(H20)9] was present in Mixture I, and hydrotalcite-like hydrate [ (Mg 4 Al2 ) (OH)i2(C03)(H 2 0)6] in Mixtures II and III. These results confirm the above conclusion that formation of the hydrotalcite-like compounds depends on the presence of Mg(OH)2 in the mixtures during hydration. It is inferred from Figures 7.3-4b and 5b that the Chapter 7 Spinel Bonded Basic Castables 166 three forms of magnesia in the three mixtures were hydrated under Hydration-B. As a result, the hydrotalcite-like compounds was formed in all the three mixtures during Hydration-B. Periclase peaks were observed in Mixtures II and III after hydration-B (Figure 7.3-4b), indicating that only part of the magnesia was hydrated in Mixtures II and III under the Hydration-B condition. This was confirmed by the T G results (Figure 7.3-5b) of the three mixtures after Hydration-B. The weight loss of Mixture I after Hydration-B was 37%, while the weight losses of the Mixture II and III were 24% and 10% respectively. The results shown in Figure 7.3-4b also indicate that more magnesia in Mixture I was hydrated than in Mixture II and Mixture III under Hydration-B, because brucite was detected in the former hydrated mixture and not detected in the latter two mixtures (Figure 7.3-4b). These results are in agreement with those in Figures 7.2-2b and 3b, where the three types of magnesia were individually hydrated under Hydration-B 7.3.3 Phase Compositions of Hydrotalcite Hydrates Figure 7.3-4b shows that the hydrotalcite hydrates formed in Mixture I had a different composition from that formed in Mixture II and Mixture III. The hydrotalcite hydrate in the former mixture had a molar M g : A l ratio of 5:1; by contrast, the hydrotalcite hydrate in the latter two mixtures had a molar M g : A l ratio of 2:1, though the three starting mixtures had the same MgO:Ai203 weight ratio of 84:16 ( M g : A l molar ratio of over 6:1). A recent investigation [39] has shown that the hydrotalcite-like hydrate formed from hydratable alumina-reactive magnesia mixture with a MgO :Al2C»3 ratio of 65:35 ( M g : A l molar ratio of about 3:1) had a molar M g : A l ratio of 3:1 [ (Mg 6 Al2 ) (OH ) i 6 (C0 3 ) (H 2 0 )4 ] . Based on our experimental results and [39], it is proposed that the varying compositions of hydrotalcite hydrates formed through the hydration-reaction of hydratable alumina-magnesia mixtures Chapter 7 Spinel Bonded Basic Castables 167 are related to the ratio of M g ( O H ) 2 to the hydrated-Al 2 0 3 in the hydrated mixtures. In Mixture I, the reactive magnesia was completely hydrated, and correspondingly the formed hydrotalcite hydrate has a high molar M g : A l ratio (5:1). However, deadburnt or fused magnesia was only partially hydrated under the hydration conditions. Consequently, the hydrotalcite hydrate has a low molar M g : A l ratio (2:1) in Mixture II and Mixture III. 7.4 Effect of Magnesia on the Bonding of Hydratable Alumina-Bonded Castables The hydration products of hydratable alumina with the presence of different forms of magnesia were studied in the above section. This section is focused on examining the bonding formation between hydration products and magnesia and its contribution to the strength of hydratable alumina-bonded castables after drying at 110°C and firing at 816°C. 7.4.1 Strength and Porosity Figure 7.4-1 shows the influence of magnesia powder and magnesia aggregates on the flexural strength of the castables after drying at 110°C and firing at 816°C. The difference between S2-6 and S l - 6 in strength should result from the difference in the composition: S l - 6 with higher strength contained 4% deadburnt magnesia (<75 pm) in the matrix, while S2-6 with lower strength contained no magnesia powder in the matrix. As described in the above section, hydrotalcite-like [ (Mg 4 Al 2 ) (OH) i 2 (C03) (H 2 0)6 ] was formed in the mixtures of hydratable alumina with deadburnt or fused magnesia, when the mixtures were hydrated at room temperature for 48 h at 20°C and then 12 h at 110°C. The surface hydroxyl groups of the neighboring hydrotalcite and M g ( O H ) 2 surrounding the magnesia particles in S l - 6 formed hydrogen bonding [49]. Accordingly, the hydrotalcite formed hydrogen bonding with M g ( O H ) 2 on the surfaces of magnesia particles in S l - 6 ; in contrast, the hydrotalcite-like compound was not formed in S2-6 without magnesia powder Chapter 7 Spinel Bonded Basic Castables 168 and consequently hydrogen bonding between the hydrotalcite hydrate and M g ( O H ) 2 on the surfaces of magnesia powder was not present in S2-6. As a result, S l - 6 had higher strength than S2-6 after drying at 110°C. These results are in agreement with the finding [116, 117] that the addition of 0.25% co-precipitated magnesium aluminate hydrate ( M g O - A l 2 0 3 16H 2 0) into deadburnt magnesia samples cured at 110°C for 24 h improved their strength because M g ( O H ) 2 formed on the surface of the M g O particles and hydrogen bonding <was created between the magnesium aluminate hydrate and the magnesium hydroxide on the surfaces of the M g O particles. cs 6 tL §" 4 o in 3 3 "§ 2 2 D Spinel aggregate, without MgO powder • Spinel aggregate, with 4 % MgO powder • Magnesia aggregate, w ithout MgO pow der El Magnesia aggregate, w ith 4 % MgO pow der 110 816 Temperature (SC) Figure 7.4-1 Comparison of M O R of castables with different aggregates and different amounts of M g O powder after drying at 110°C and firing at 816°C. Figure 7.4-1 shows that, after drying at 110°C, the castable with magnesia aggregate and 4% magnesia powder ( M l - 6 ) had higher flexural strength than the castable with magnesia aggregate, but without magnesia powder (M2-6). The above discussions apply here. The Chapter 7 Spinel Bonded Basic Castables 169 hydrotalcite hydrate formed hydrogen bonding with M g ( O H ) 2 on the surfaces of both magnesia powder and magnesia aggregate in M l - 6 ; in comparison, this hydrate formed hydrogen bonding with Mg(OH )2 on the surfaces of only magnesia aggregate in M2-6 . Figure 7.4-1 exhibits that after drying at 110°C, the castable with magnesia aggregate (M2-6) had higher flexural strength than the castable with spinel aggregate (S2-6), In the former castable, the hydrotalcite hydrate formed hydrogen bonding with M g ( O H ) 2 on the surfaces of magnesia aggregate. B y contrast, the hydrotalcite hydrate was not formed in the latter castable because no magnesia was present and accordingly no hydrogen bonding was formed between the matrix and the aggregate. As shown in Figure 7.4-1, after drying at 110°C, the castable with magnesia aggregate and 4% magnesia powder ( M l - 6 ) had higher flexural strength than the castable with spinel aggregate and 4% magnesia powder (Sl-6). In the former castable, the hydrotalcite hydrate formed hydrogen bonding with M g ( O H ) 2 on the surfaces of both magnesia powder and magnesia aggregate; on the other hand, the hydrotalcite hydrate formed hydrogen bonding with M g ( O H ) 2 on the surfaces of only magnesia powder in the latter castable. Figure 7.4-1 exhibits that the castables possessing higher strength after drying at 110°C had higher strength after firing at 816°C. This is attributed to the reactions between the hydrates existing in the castables. The hydroxyls leading to formation of hydrogen bonding lose water during heating and the dehydroxylation reaction of the hydrotalcite hydrate and M g ( O H ) 2 on the surface of magnesia powder and/or magnesia aggregate results in a polycondensation-type reaction [49, 116, 117]: H O - M g - O H + H O - ( M g , A l ) - O H + H O - M g - O H + Chapter 7 Spinel Bonded Basic Castables 170 > -0-Mg-0-(Mg, A l ) - 0 - M g - 0 - + n H 2 0 The condensation reaction during heating of castable M l - 6 occurred between the hydrotalcite hydrate and M g ( O H ) 2 on the surface of both magnesia aggregate and magnesia powder in the matrix. In comparison, this condensation reaction took place only on the surface of magnesia aggregate in M2-6 and only on the surface of magnesia powder in the matrices in SI-6. It is clear that this condensation reaction did not occur in S2-6. Consequently, M l - 6 had the highest strength, S2-6 had the lowest strength, and the other two castables had strength in between because the condensation reaction contributed to the strength of the castables after firing at 816°C. It is worth noting in Figure 7.4-1 that SI-6 with spinel aggregate and 4% magnesia powder, after thermal treatment at 110°C and 816°C, had lower strength than M2-6 with magnesia aggregate and no magnesia powder, implying that the bonding between hydrotalcite and M g ( O H ) 2 on the surfaces of magnesia aggregate of M2-6 played a larger role in strength contribution than the bonding between hydrotalcite and M g ( O H ) 2 on the surfaces of 4% magnesia powder in the matrix of SI-6. It is normally accepted that the bonding between the binders and aggregates of castables contribute relatively little to the strength of castables after drying and firing at intermediate temperature. The results of this work (Figure 7.4-1) suggest that the bonding between the binder (hydratable alumina) and the magnesia aggregate contribute significantly to the strength after heat treatment at 110°C and 816°C. Accordingly, use of hydratable alumina in magnesia-aggregate castables is beneficial to the strength of the castable after drying and firing at intermediate temperatures. Chapter 7 Spinel Bonded Basic Castables 171 Figure 7.4-2 shows that, after drying at 110°C, M l - 6 had the lowest apparent porosity, castable S2-6 had the highest apparent porosity and the apparent porosity of castables S1-6 and M2-6 was in between. A s discussed before, M g ( O H ) 2 was formed on the surfaces of both the magnesia aggregate and the magnesia powder in castable M l - 6 . On the contrary, M g ( O H ) 2 was not formed in S2-6 because neither magnesia aggregate nor magnesia powder was present in the castable. In comparison, M g ( O H ) 2 was formed on the surfaces of only magnesia aggregate in M2-6 and on the surfaces of only the magnesia powder in S l - 6 . As a result, the hydrotalcite was bonded with the magnesia aggregate and magnesia powder in M l - 6 , with only magnesia aggregate in M2-6 and with only magnesia powder in S l - 6 after drying at 110°C. A n d the hydrotalcite was not formed in S2-6. Therefore, the amount and bonding status of the hydrotalcite hydrate with the aggregate and the powder led to the difference in the apparent porosity in the three castables. 22 20 4 18 4 (0 o | 16 c g. 14 Q. < 12 4 10 • Spinel aggregate, w ithout MgO pow der • Spinel aggregate, w ith 4% MgO pow der • Magnesia aggregate, w ithout MgO pow der H Magnesia aggregate, with 4% MgO powder 110 816 Temperature (9C) Figure 7.4-2 Comparison of A P of castables with different aggregates and different amounts of M g O powder after drying at 110°C and firing at 816°C. Chapter 7 Spinel Bonded Basic Castables 172 When the hydrotalcite hydrate was bonded with both the magnesia aggregate and magnesia powder, the castable had the lowest porosity ( M l - 6 ) . When the hydrotalcite hydrate was not present, the castable had the highest porosity. If the hydrotalcite hydrate was bonded to only the magnesia aggregate (M2-6) or only the magnesia powder (SI-6), the castables had apparent porosity in between. Figure 7.4-2 illustrates that the apparent porosity variation after firing at 816°C was similar to that after drying. As discussed above, hydrotalcite was bonded to magnesia aggregate and magnesia powder in M l - 6 , to only magnesia aggregate in M2-6 and to only magnesia powder in S1-6; and the hydrotalcite hydrate was not present in S2-6. After the decomposition of the hydrates, M l - 6 is expected to have a higher increase in apparent porosity than M2-6 and S1-6, and S2-6 expected to have a lower increase in apparent porosity than M 2 - 6 and S1-6. In fact, comparing with the apparent porosity of the castables after drying, the increase in apparent porosity of M l - 6 , S l - 6 , M2-6 and S2-6 are 3.6, 3.1, 2.8 and 2.7% respectively. During heat treatment, the dehydroxylation of the hydrotalcite hydrate and M g ( O H ) 2 was accompanied by the condensation reaction of the hydroxyls of the hydrotalcite and M g ( O H ) 2 on the magnesia particles, also leaving voids in the materials. 7.4.2 Morphology of Hydrates Bayerite and boehmite were formed from hydratable alumina after hydration at 20°C for 48 h and 110°C for 12 h [40]. It could be inferred that the hydroxyls of bayerite and boehmite can form hydrogen bonding with the hydroxyls of M g ( O H ) 2 on the surfaces of magnesia. However, the hydration products of the hydratable alumina had a granular shape (Figure 7.4-3a) after hydration at 20°C for 48 h and 110°C for 12 h. This morphology was also observed after the hydratable alumina was hydrated at 25°C for 100 h [93]. The granular Chapter 7 Spinel Bonded Basic Castables 173 shape of bayerite and boehmite make the hydrates less accessible to M g ( O H ) 2 on the surfaces of magnesia and limited their contribution to the strength of castables. Figure 7.4-3 S E M micrographs of (a) hydratable alumina and (b) Mixture III after hydration at 20°C for 48 h and 110°C for 12 h. The morphology of hydrotalcite formed from Mixture HI (refer to Section 4.3.6) after hydration at 20°C for 48 h and 110°C for 12 h is presented in Figure 7.4-3b. Comparison of the morphologies of hydration products from pure hydratable alumina and the mixture reveals that hydrotalcite formed in castables closely wrap the magnesia particles and has much more contacts with M g ( O H ) 2 on the surfaces of magnesia than bayerite and boehmite. The morphological features of hydrotalcite promote its bonding with Mg(OH) 2 on the surfaces of magnesia. 7.4.3 Dehydroxylation of the Hydrates The increased strength of hydratable alumina-bonded castables in the presence of magnesia is also possibly related to the thermal decomposition behavior of the hydrates in the castables. For pure hydratable alumina after hydration (Figure 7.4-4a), the formed bayerite and boehmite decomposed around 280°C [118] and 426°C [119] respectively. Chapter 7 Spinel Bonded Basic Castables 174 Figure 7.4-4b exhibits the thermal decomposition behavior of the mixture of hydratable alumina and magnesia powder. The endothermic peaks in the temperature range of 150-220°C resulted from the decomposition of the interlayer water in hydrotalcite and those of 300-400°C were due to loss of the (OH) and ( C 0 3 ) groups in hydrotalcite [39, 120, 121]. 3 o CO 0) X o TJ LU 3 o CO CD X o TJ c U J 0 100 200 300 400 500 600 Q ^ 2 Q Q 3 Q Q 4 Q Q 5 Q Q m Q Temperature PC) Temperature P C ) Figure 7.4-4 T G - D T A curves for the thermal decomposition of (a) hydratable alumina and (b) Mixture III after hydration at 20°C for 48 h and 110°C for 12 h. Comparison of the T G results in Figures 7.4-4a and 7.4-4b reveals that the weight loss in hydratable alumina sample occurred in a narrower temperature range and with fewer steps than in the mixture of hydratable alumina and magnesia. In hydratable alumina-bonded castable after curing and drying, rapid decomposition of bayerite and boehmite possibly created local explosion spalling, because of the local high vapor pressure [37, 38]. It is proposed that the polycondensation of hydrates in the matrices and on the surfaces of aggregates was impaired by the local explosion spalling within and between the disassociated hydrates, resulting in lower strength of the castables during heating. On the Chapter 7 Spinel Bonded Basic Castables 175 other hand, decomposition of interlayer water in hydrotalcite in the lower temperature range (150-220°C) generated a micro-porous microstructure [39], which provided escape channels for the vapor disassociated at the higher temperature range (300-400°C). Consequently, the decrease in the local vapor pressure mitigated the local explosion spalling, and hydrotalcite present in hydratable alumina-bonded castables after curing and drying could prevent the strength decrease of the castables after heat treatment at the intermediate temperature. 7.5 Summary The increase in ultrafine alumina (0.5 pm median particle size) up to 6% increased the bulk density and decreased the apparent porosity of magnesia-based and spinel-based castables after drying at 100°C and firing at 816°C. This indicates that, as also discussed in Chapter 6, the ultrafine particles filled the voids between the larger particles. The increase in ultrafine alumina up to 6% has little influence on the linear change and modulus of rupture of magnesia-based and spinel-based castables after firing at 816°C, demonstrating that this temperature is not high enough to activate the mass transport in the binding systems. Addition of the ultrafine alumina (0.5 pm median particle size) up to 6% decreased the linear expansion of the magnesia-based castables fired at 1280°C, indicating that addition of the ultrafine alumina enhanced the sintering at this firing temperature. However, the modulus of rupture of the castables did not increase with the addition of the ultrafine alumina powder because of the cracks generated by in-situ spinel formation and mismatch in thermal expansion coefficients of the aggregate and matrix. The spinel-based castable with 6% ultrafine alumina (0.5 pm median particle size) showed significantly higher modulus of rupture than the castables without the ultrafine powder after firing at 1280°C. These castables had higher strength than the magnesia-based Chapter 7 Spinel Bonded Basic Castables 176 castables with the aggregate after firing at 1280°C because the cracks between the aggregate and matrix did not occur in the spinel-based castables. The addition of 5% the ultrafine spinel (0.3 pm median particle size) increased the strength of the spinel-based castables after firing at 1280°C, but not the strength of the magnesia-based castables. The in-situ spinel formation and the mismatch in thermal expansion coefficients between the aggregate and matrix generated microcracks between the matrix and aggregate in magnesia-based castables and, as a result, reduced the strength of the castables. The particle size of the ultrafine power played an important role in enhancing strength of the castables after heat treatment at 816°C and 1280°C. Strength of the castable was not obviously increased with 2% of the ultrafine spinel with a median size of 0.3 pm at 1280°C, but significantly increased with 2% the ultrafine spinel with a median size of 0.2 pm. Addition of the ultrafine spinel (0.2 pm median) increased the strength of the castables after firing at 816°C, while addition of the ultrafine alumina (0.5 pm median) and spinel (0.3 pm median) did not increase strength of the castable after firing at the same temperature. These results indicate that sintering of the binding systems of the castable progresses to a higher extent at 816°C or can possibly happen at even lower temperatures (<816°C) i f the particle size of the ultrafine powders is further decreased (<0.2 pm). Spinel-based castables with high strength (MOR>15 MPa) were obtained with the addition of 2% of the ultrafine spinel (0.2 pm median) after firing at 1280°C. Hydratable alumina was not completely hydrated at 20°C for 48 h and then at 110°C for 12 h. A hydrotalcite hydrate formed in the mixture of hydratable alumina and reactive magnesia after hydration at 20°C for 48 h, while such a hydrate was not observed in the Chapter 7 Spinel Bonded Basic Castables 177 mixtures of hydratable alumina and deadburnt or fused magnesia under the same hydration condition. This is because reactive magnesia was significantly hydrated, but deadburnt or fused magnesia was not significantly hydrated under this hydration condition. After hydration at room temperature for 48 h and then at 110°C for 12 h, hydrotalcite compounds formed in the mixtures of hydratable alumina with reactive magnesia, deadburnt magnesia and fused magnesia respectively as the three types of magnesia were all significantly hydrated under these hydration conditions. The compositions of hydrotalcite hydrates formed through the hydration-reaction of hydratable alumina and various forms of magnesia varied with the M g ( O H ) 2 : A l 2 0 3 ratio in the hydrated mixtures, rather than the M g O : A l 2 0 3 ratio in the starting mixtures. The presence of magnesia aggregate and powder in hydratable alumina-bonded castables improved the strength of the castables after drying at 110°C because of the hydrogen bonding between hydrotalcite and M g ( O H ) 2 on the surface of magnesia particles. The polycondensation accompanying dehydroxylation of hydrotalcite and M g ( O H ) 2 at 816°C contributed to the strength of hydratable alumina-bonded castables containing magnesia. The morphological features of hydrotalcite facilitated bonding formation of hydrotalcite with M g ( O H ) 2 on the surfaces of magnesia. The hydration products of the hydratable alumina have a granular shape and, as a result, have limited contacts with M g ( O H ) 2 on the surfaces of magnesia. In contrast, hydrotalcite closely wrapped magnesia particles. Decomposition of hydrotalcite at the lower temperatures (150-220°C) could provide escape channels for the vapor released at higher temperatures (300-400°C) and accordingly, alleviate explosion spalling during heating-up and benefit the strength of castables after firing at intermediate temperatures. Chapter 8 Conclusions 178 CHAPTER 8 CONCLUSIONS In this work, novel approaches have been developed for the preparation of agglomerate-free and ultrafine magnesium aluminate spinel powders. Solid-state sintering of multi-modal binding systems was enhanced through the incorporation of such spinel powders. The resulting new knowledge provided insight into the development of hydratable alumina-bonded castable refractories to achieve high strength after firing at relatively low temperatures (<1300°C). The hydration behavior of hydratable alumina in the presence of various forms of magnesia was also elucidated. This knowledge threw light on the possibility of increasing strength of hydratable alumina-bonded castables after drying and firing at temperatures of 800-1000°C. The goal of this thesis was accomplished as hydratable alumina-bonded castables with high strength (MOR>15 MPa) were obtained after firing at 1280°C. The following principal conclusions are drawn from this work: Synthesis of Spinel Powders 1. The incorporation of the combustible ingredients, such as carbon black and polyvinyl alcohol ( P V A ) , into the spinel precursor suspensions prepared by coprecipitation effectively reduced agglomeration of the precursor particles during drying and calcination, and did not hamper the spinel formation during calcination. 2. Single-phase spinel was obtained from the heterogeneous sol-gel precursors prepared with reactive magnesia (3.09 pm median particle size) after calcination at 1200°C. It was suggested that spinel formation took place in one-step reaction from the precursors, as opposed to the two-step process from the precursor using nano-sized magnesia (10 nm median). Spinel-coated magnesia was obtained using this synthesis after the precursor was calcined at 1200°C. Chapter 8 Conclusions 179 3. The mechanical mill ing transformed the starting magnesium and aluminum hydroxides into an amorphous state, decreased the particles from 6-15 pm to 200-300 nm, mixed aluminum and magnesium-bearing ingredients uniformly and, as a result, enhanced the spinel formation. Completely crystallized spinel was obtained from the mixtures milled for 20 h and calcined at 1000°C. 4. Grinding of the spinel precursor prepared by the heterogeneous sol-gel process decreased the A l ( O H ) 3 dehydroxylation temperature from 190°C to about 130°C. Grinding of the precursor for 1 h lowered the incipient temperature of spinel formation from 900°C to 800°C, and the temperature of complete spinellization from >1280°C to 900°C. The activation energy for spinel formation decreased from 688 kJ/mol for the as-prepared precursor to 468 kJ/mol for the precursor subjected to 5 h of mill ing. 5. The novel Sol-Gel-Precipitation (SGP) process enhanced the precursor's homogeneity through formation of alumina sol and magnesium hydroxide precipitation in two separate steps, thus allowing synthesis of complete crystallized spinel at temperatures as low as 900°C. Ultrafine spinel powder (dso = 600 nm, specific surface area = 105 m 2 /g, crystallite size 13 nm) was obtained after calcination at 900°C. The optimum process temperature for acquiring spinel powder with high sintering reactivity was determined in the range of 900-1000°C. Sintering of the Binding Systems 1. The.relationship between the strength and shrinkage for the powder compacts consisting of uniform particles does not apply to the binding systems with wide particle size distributions. It is therefore proposed in this study that the solid-state sintering of compacts with multi-modal particle size distributions is better evaluated using strength Chapter 8 Conclusions 180 than using shrinkage/densification curves. This is because the particle rearrangement leads to shrinkage, but not to strength. 2. For both the binary and tri-modal systems, the addition of the ultrafine alumina (0.5 pm median) and spinel (<0.3 pm median) up to 10% increased the green bulk density and decreased the apparent porosity of the compacts after heat treatment up to 1450°C. However, the addition of the above ultrafine powders up to 10% did not promote sintering of the binary compacts after firing at 1280°C and 1450°C. 3. After heat treatment at 1280°C and 1450°C, the tri-modal compacts showed higher bulk density (2.66 and 2.70 g/cm 3 respectively) and compressive strength (28 and 65 M P a respectively) than the binary compacts (2.26 and 2.38 g/cm 3 respectively, and 22 and 57 M P a respectively). The tri-modal compacts had lower incipient sintering temperatures (937°C) and sintering activation energy (130±3 kJ/mol) than the bimodal compacts (1053°C and 142±5 kJ/mol respectively) (The values refer to samples containing 5% the ultrafine spinel with a median size of 0.2 pm). 4. In respect to promoting sintering, the synthesized spinel MA-9-900-mil led (0.6 pm median, 105 m /g) showed the highest effect, followed by the ultrafine spinel (0.2 pm median, 30 m2/g) and the ultrafine spinel (0.2 pm median, 12 m2/g). The higher specific surface area and smaller particle size of the ultrafine powders provided higher driving force for sintering. For the same reasons, addition of 5% the ultrafine spinel (0.2 pm median) increased the sintering of the tri-modal compacts at 816°C, and equal amounts of ultrafine alumina (0.5 pm) and ultrafine spinel (0.3 pm) did not show that effect at similar temperature. Chapter 8 Conclusions 181 5. For both binary and tri-modal systems, addition of 5% ultrafine spinel powders increased the strength to a higher extent at relatively lower firing temperatures (1280°C) and to a lower extent at relatively higher temperatures (1450°C). Spinel-Bonded Castables 1. The particle size of the ultrafine powders played an important role in enhancing the strength of the castables after heat treatment at 816°C and 1280°C. The addition of the ultrafine spinel (0.2 pm median) increased the strength of the castables after firing at 816°C, while the addition of the ultrafine alumina (0.5 pm median) and spinel (0.3 pm median) did not show the similar effect. The strength was not obviously increased by 2% addition of the ultrafine spinel (0.3 pm median) after firing at 1280°C, but significantly increased by 2% addition of the ultrafine spinel (0.2 pm median). Spinel-based castables with high strength (MOR>15 MPa) were obtained with 2% addition of the ultrafine spinel (0.2 pm median) after firing at 1280°C. 2. The aggregate type has substantial influence on strength after firing at 1280°C. The magnesia-based castables had lower strength than the spinel-based castables because the in-situ spinel formation and mismatch in thermal expansion coefficients between the aggregate and matrix generated cracks in the magnesia-based castables and such cracks did not occur in the spinel-based castables. 3. The hydration of the mixtures of hydratable alumina and non-reactive magnesia (deadburnt or fused) showed different mechanisms from those of hydratable alumina and reactive magnesia. After hydration at 20°C for 48 h, hydrotalcite formed in the mixture of hydratable alumina and reactive magnesia, while such a hydrate was not observed in the mixtures of hydratable alumina and deadburnt or fused magnesia under the same Chapter 8 Conclusions 182 hydration condition. After hydration at room temperature for 48 h and then at 110°C for 12 h, hydrotalcite formed in all experimental mixtures of hydratable alumina with reactive magnesia, deadburnt magnesia and fused magnesia respectively, suggesting that the hydrotalcite hydrate can form in any hydratable alumina-bonded castables containing magnesia during curing and drying/heating-up in industrial vessels. 4. The presence of hydrotalcite increased the bending strength of hydratable alumina-bonded castables from 3.7 M P a (spinel based castable without magnesia powder) to 6.4 M P a (magnesia-based castable with 4% magnesia powder) after drying at 110°C and from 0.6 M P a to 2.3 M P a after firing at 816°C. Chapter 9 Future Work CHAPTER 9 FUTURE WORK 183 1. It was reported that the main phase in the spinel precursor prepared using a heterogeneous sol-gel method and aluminum isopropoxide and nano-sized M g O (10 nm median) as the starting materials [10] was M g 4 A l 2 ( O H ) i 4 - 3 H 2 0 . However, this compound was not observed in our experiments of preparing the M A - 5 precursor using the same method and aluminum isopropoxide and micro-sized M g O (3.09 nm median) as the starting materials. The critical particle size of M g O for forming M g 4 A l 2 ( O H ) i 4 - 3 H 2 0 in the spinel precursors needs to be investigated. 2. The permeability of hydratable alumina-bonded castables is known to be relatively low because of the hydration products from hydratable alumina. A s a result, the explosive spalling resistance is poor. Our observations indicate that the hydration products of hydratable alumina and their dehydration temperatures were changed when magnesia was present. The hydrotalcite dehydrated at relatively lower temperatures (150-220°C and 300-400°C) than bayerite and boehmite from pure hydratable alumina (280°C and 426°C respectively). We propose that the explosive spalling resistance of hydratable alumina-bonded castable with the presence of magnesia could be improved because the dehydration at 150-220°C may provide escape channels for the vapor disassociated at the higher temperature range (300-400°C). The improvement of the thermal shock resistance of hydratable alumina-bonded castables in the presence of magnesia needs to be confirmed. 3. It was found that the bending strength of the spinel-based castable fired at 816°C was increased from 1.3 M P a to 4.0 M P a when the median particle size of the ultrafine spinel Chapter 9 Future Work 184 powders (5%) decreased from 0.3pm to 0.2 pm. More work needs to be done explain this effect and to further increase the strength at the temperature using spinel powder with a median particle size below 100 nm. References 185 REFERENCES [I] S. Banerjee, Monolithic Refractories - A Comprehensive Handbook, Wor ld Scientific, Singapore, 1998, p.vi i . [2] M . Kobayashi, K . Kataoka, Y . Sakamoto, and I. Kifune, "Use of Alumina and Magnesia Castables in Steel Ladle Sidewalls", Taikabutsu Overseas, 16 [3] 39-44 (1997). [3] R.W. McConnell, R . A Marra, and F.A. Fullington, "Benefits of Sintered Magnesium Aluminate Spinels in High Purity Castable Refractories", Refractories Applications, 7 [2] 23-28 (2002). [4] M . E . Schlesinger, M . Karakus, M . Crites, M . A . Somerville and S. Sun, "Chrome-Free Refractories for Copper Production Furnaces", pp. 1703-1706 in Proc. of UNITECR, 1997, New Orleans, U S A . [5] L . J. Davies and J. M . M c C o l l u m , "Refractory Selection for Non-Ferrous Smelting Application", pp. 195-214 in Advances in Refractories for the Metallurgical Industries, edited by M . Riguad, C I M , Montreal, 1987. [6] F. Golestani-Fard, K . Bagheri, H . R Hassani, and N . Nobari, "Refractory Practices in Iranian Sarchesmeh Copper Complex", pp. 372-374 in Proc. of UNITECR, 1999, Berlin, Germany. [7] R. E . Fisher, L . P. Krietz, L . Leung and J. G . Beets, "Refractory Usage in Non-Ferrous Metallurgical Applications", in Proc. of UNITECR, 1993, Sao Paolo, Brazi l , pp. 372-374. [8] D . J. Bray, "Toxicity of Chromium Compounds Formed in Refractories", Am. Ceram. Soc. Bull., 64 [4] 1012-16 (1985). [9] C . G . Marvin, "Chrome-Bearing Hazardous Waste", Am. Ceram. Soc. Bull., 72 [6] 66-68 (1993). [10] T. Shiono, K . Shiono, K . Miyamoto, and G . Pezzotti, "Synthesis and Characterization of (MgAl 2 04) Spinel Precursor from a Heterogeneous Alkoxide Solution Containing Fine M g O Powder", J, Am. Ceram. Soc, 83 [1] 235-37 (2000). [II] M . Sindel, N . A . Travtzky and N . Claussen, "Influence of Magnesium Aluminate Spinel on the Direct Oxidation of Molten Aluminum Al loys" , J. Am. Ceram. Soc, 73 [9] 2615-18 (1990). References 186 [12] G . Baudin, R. Martinez, and P. Pena, "High Temperature Mechanical Behavior of Stoichiometric Magnesium Spinel", J. Am. Ceram. Soc, 78 [7] 1857-62 (1995). [13] J . -G. L i , T. Ikegami, J . -H. Lee, T. M o r i and Y . Yajima, " A Wet-Chemical Process Yielding Reactive Magnesium Aluminate Spinel (MgA^CU) Powder", Ceramics International, 27, 481-489 (2001). [14] S.-K. Chen, M . - Y . Cheng, S.-J. Lin, and Y . C. Ko, 'Thermal Characteristics of A l 2 0 3 - M g O and Al2C>3-Spinel Castable for Steel Ladles", Ceramics International, 28, 811-817 (2002). [15] Y . C. K o and C. F. Chan, "Effect of Spinel Content on Hot Strength of Alumina-Spinel Castables in the Temperature Range 1000-1500°C", J. Eur. Ceram. Soc, 19, 2633-2639 (1999). [16] P. Korgul, D . R. Wilson, and W . E . Lee, "Microstructural Analysis of Corroded Alumina-Spinel Castable Refractories", J. Eur. Ceram. Soc, 17, 77-84 (1997). [17] J. M o r i , W . Watanabe, M . Yoshimura, Y . Oguchi, and T. Kawakami, "Material Design of Monolithic Refractories for Steel Ladle", Bull. Am. Ceram. Soc, 69 [7] 1172-1176 (1990). [18] B . Nagai, O. Matsumoto, and T. Isobe, "Development of Monolithic Refractory Linings for B O F Ladle in Japan Mainly for the Last Decade", pp. 116-122 in Proc. ofUNITECR, 1991, Aachen, Germany. [19] K . Kurata, T. Matsui, and S. Sakaki, "Castable lining technique to bottom of teeming ladle", Taikabutsu Overseas, 12 [1] 29-39 (1992). [20] B . Nagai, O. Matsumoto, T. Isobe, and Y . Nishiumi, "Wear mechanism of castable for steel ladle by slag", Taikabutsu Overseas, 12 [1] 15-20 (1992). [21] F. N . Cunha-Duncan and R. C. Bradt, "Synthesis of Magnesium Aluminate Spinels from Bauxite and Magnesias", J. Am. Ceram. Soc, 85 [12] 2995-3003 (2002). [22] S. G . Mukherjee and B . N . Samaddar, "Spinel Formation from Coprecipitated Hydroxides of Aluminum and Magnesium", Trans. Ind. Ceram. Soc, 25 [1] 33-34 (1966). References 187 [23] S. Hokazono, K. Manako and A. Kato, "The Sintering Behavior of Spinel Powders Produced by a Homogeneous Precipitation Technique", British Ceram. Trans, and J., 91, 77-79 (1992). [24] D. Lepkova, A . Bartarjav, B. Samuneva, Y . Ivanova, and L. Georgieva, "Preparation and Properties of Ceramics from Magnesium Spinel by Sol-Gel Technology", J. Mater. Sci., 26, 4861-64(1991). [25] J .G.M. Delau, "Preparation of Ceramic Powders from Sulfate Solution by Spray Drying and Roasting", Amer. Ceram. Soc. Bull., 49 [6] 572-74 (1970). [26] Y . Suyama, and A. Kato, "Characterization and Sintering of Magnesium-Aluminium Spinel by Spay-Pyrolysis Technique", Ceram. Int., 8 [1] 17-21 (1982). [27] C. T. Tang, L. S. Lin, and S. J. Yang, "Preparation of M g A l 2 0 4 Spinel Powders via Freeze-Drying of Alkoxide Precursors", / . Amer. Ceram. Soc, 75 [8] 2240-43 (1992). [28] Z. Nakagawa, K. Hamano, M . Sakaguchi, and S. Kanzaki, "Characterization and Sinterability of Mg-Al Spinel Powders Prepared with a Thermal Decomposition of a Freeze-Dried Sulfate", Yogyo Kyokaishi, 90 [6] 313-19 (1982). [29] R. Baranwal, M . P. Villar, R. Garcia, and Richard M . Laine, "Flame Spray Pyrolysis of Precursors as a Route to Nano-Mullite Powder: Powder Characterization and Sintering Behavior", 7. Amer. Ceram. Soc, 84 [5] 951-61 (2001). [30] C. R. Bickmore, K. F. Waldner, D. R. Treadwell, and R. M . Laine, "Ultrafine Spinel Powders by Flame Spray Pyrolysis of a Magnesium Aluminium Double Alkoxide", J. Amer. Ceram. Soc, 79 [5] 1419-23 (1996). [31] L. Lu, M . O. Lai and G. Ye, " A Method for Producing Oxide Compounds", WO0132583 (2001). [32] W. Kim, F. Saito, "Effect of Grinding on Synthesis of M g A l 2 0 4 Spinel from a Powder Mixture of Mg(OH) 2 and Al(OH) 3 ", Powder Technology, 113, 109-113 (2000). [33] K. J. D. Mackenzie, J. Temuujin, T. Jadambaa, M . E. Smith, and P. Angerer, "Mechanochemical Synthesis and Sintering Behavior of Magnesium Aluminate Spinel", J. of Materials Science, 35, 5529-5535 (2000). References 188 [34] D. W. Johnson, Jr., "Innovations in Ceramic Powder Preparation", pp. 3-19 in Advances in Ceramics, Vol . 21, edited by G. L . Messing, E. R. Fuller, Jr., and H. Hausner, The American Ceramic Society, 1988. [35] F. Azizian, "Improving the Performances of Cement-Free Castables", Ceramic Industry, 147 [2] 42-48 (1997). [36] Y . Hongo, "p-Alumina Bonded Castable Refractories", Taikabutsu Overseas, 9 [1] 35-38 (1988). [37] M . D. M . Innocentini, A . R. F. Pardo, and V. C. Pandolfelli, "Permeability of High-Alumina Refractory Castables Based on Various Hydraulic Binders", J. Am. Ceram. Soc, 85 [6] 1517-21 (2002). [38] F. A . Cardoso, M . D . M . Innocentini, M . F.S. Miranda, F. A.O. Valenzuela, and V. C. Pandolfelli, "Drying Behavior of Hydratable Alumina-Bonded Refractory Castables", Journal of the European Ceramic Society, 24, 797-802 (2004). [39] K. Ghanbari Ahari, J.H. Sharp, and W.E. Lee, "Hydration of Refractory Oxides in Castable Bond Systems—I: alumina, magnesia, and alumina-magnesia mixtures", J. of the European Ceramic Society, 22, 495-503 (2002). [40] G. Ye and T. Troczynski, "Hydration of Hydratable Alumina in the Presence of Various Forms of MgO", Ceramics International (accepted). [41] W. D. Kingery, "Introduction to Ceramics", 2nd Ed, John Wiley & Sons, 1976, pp.474-476. [42] M . W. Barsoum, "Fundamentals of Ceramics", Institute of Physics Publishing, 2003, pp.311-315. [43] B. G. Ravi and R. Chaim, "Sintering of Bimodal Alumina Powder Mixtures with a Nanocrystalline Component", NanoStructured Materials, 11 [7] 853-859 (1999). [44] T.-S. Yeh and M.D. Sacks, "Effect of Particle Size Distribution on the Sintering of Alumina", J. Am. Ceram. Soc, 71 [12] C484-C487 (1988). References 189 [45] M . W . Barsoum, "Fundamentals of Ceramics", Institute of Physics Publishing, 2003, pp.66-67. [46] F. N . Cunha-Duncan and R. C . Bradt, "Synthesis of Magnesium Aluminate Spinels from Bauxites and Magnesias," J. Am.Ceram. Soc, 82 [12] 3279-92 (1999). [47] W . D. Kihgery, Introduction to Ceramics, 2nd Ed. ; John Wiley and Sons, New York and London, 1976, p. 595. [48] A . C. Dodd and P. G . McCormick, "Synthesis of Nanoparticulate Zirconia by Mechanochemical Processing", Scripta Mater. 44, 1725-1729 (2001). [49] A . C. Dodd, K . Raviprasad and P. G . McCormick, "Synthesis of Ultrafine Zirconia powders by Mechanochemical Processing", Scripta Materials, 44, 689-694 (2001). [50] L . A . Paulick, Y . F . Y u , and T. L . Mah , "Ceramic Powders from Alkoxide Precursors," pp. 99-107, in Advances in Ceramics, V o l . 21, edited by G . L . Messing, E . R. Fuller, Jr., and H . Hausner, the American Ceramic Society, 1988. [51] O. Yamaguchi, H . Omaki, and K . Shimizu, "Formation of Spinel (MgA^CU) from Hydroxides Prepared by Alkoxy-Method," J. Jpn. Soc. Powder Metall, 22 [6] 202-204 (1975). [52] C. J. Brinker and G . W . Scherer, "Sol-Gel Science", Academic Press, 1990, pp. 456-459. [53] V . Montouillout, D . Massiot, A . Douy and J. P. Coutures, "Characterization of M g A l 2 0 4 Precursor Powders Prepared by Aqueous Route", J. Am. Ceram. Soc, 82 [12] 3299-304 (1999). [54] F.J . Schnettler, F .R. Monforte, and W . W . Rhodes, pp.79-90 in Science of Ceramics, V o l . 4. Edited by G . H . Steward, The British Ceramic Society, 1968. [55] E . Avvakumov, M . Senna and N . Kosova, "Soft Mechanochemical Synthesis: A Basis for New Chemical Technologies", Kluwer Academic Publishers, 2001. [56] W . D . Kingery, "Introduction to Ceramics", 2nd Ed , John Wiley & Sons, 1976, pp.227-228. References 190 [57] N . N . Thadhani, "Shock-Induced and Shock-Assisted Solid-State Chemical Reactions in Powder Mixtures", 7. Appl. Phys., 76, 2129-38 (1994). [58] T. G. Shen, C. C. Koch, T. L. McCormick, R. I. Nemanich,.J. Y . Huang and J. G. Huang, "The Structure and Properties of Amorphous/Nanocrystallite Silicon Produced by Ball Milling", 7. Mater. Res., 10, 139-48 (1995). [59] T. Tsuchida and K. Horigome, "The Effect of Grinding on the Thermal Decomposition of Alumina Monohydrates, a- and P - A I 2 O 3 H 2 O " , Thermochimica Acta, 254, 359-370 (1995). [60] C. C. Yong and J. Wang, "Mechanical-Activation-Triggered Gibbsite-to-Boehmite Transition and Activation-Derived Alumina Powders", 7. Am. Ceram. Soc, 84 [6] 1225-30 (2001). [61] J.Temmujin, K. J. D. MacKenzie, M . Schmucker, H. Schneider, J. McManus, and S. Wimperis, "Phase Evolution in Mechanically Treated Mixtures of Kaolinite and Alumina Hydrates (Gibbsite and Boehmite)", 7. of the European Ceramic Society, 20, 413-23 (2000). [62] J. Wang, J. Xue and D. Wan, "How Different is Mechanical Activation from Thermal Activation? A Case Study with PZN and PZN-based Relaxors", Solid State Ionics, 127, 169-175 (2000). [63] J. Xue, D. Wan and J. Wang, "Mechanochemical Synthesis of Nanosized Lead Titanate Powders from Mixed Oxides", Materials Letters, 39, 364-369 (1999). [65] J. Temuujin, K. Okada and K. J. D. MacKenzie, "Formation of Mullite from Mechanochemically Activated Oxides and Hydroxdes", 7. of Euro. Ceram. Soc, 18, 831-835 (1998). [66] R. M . German, "Sintering Theory and Practice", John Wiley and Sons, Inc., 1996, pp.12-13. [67] M . W. Barsoum, "Fundamentals of Ceramics", Institute of Physics Publishing, 2003, pp.304-305. References 191 [68] R. M . German, "Sintering Theory and Practice", John Wiley and Sons, Inc., 1996, pp.4-5. [69] R. M . German, "Sintering Theory and Practice", John Wiley and Sons, Inc., 1996, pp.75-76. [70] R. M . German, "Sintering Theory and Practice", John Wiley and Sons, Inc., 1996, pp.100-103. [71] J. P. Smith and G. L. Messing, "Sintering of Bimodally Distributed Alumina Powders", J. Am. Ceram Soc, 6 7 , 238-42 (1984). [72] J. Han, A . M . R. Senos and P. Q. Mantas, "Nonisothermal Sintering of Mn Doped ZnO", Journal of the European Ceramic Society, 19 (1999) 1003-1006. [73] W. S. Young and I. B. Cutler, "Initial Sintering with Constant Rates of Heating", J. Am. Ceram. Soc, 5 3 , [12] 659-663 (1970). [74] H. Su and D. L. Johnson, "Master Sintering Curve: A Practical Approach to Sintering", J. Am. Ceram. Soc,19 [12] 3211-17 (1996). [75] M . J. O'Hara and I. V . Cutler, "Sintering Kinetics of Binary Mixtures of Alumina Powders", Proceedings of the British Ceramic Society, No. 12, 145-154 (1969). [76] R. M . German and M . Bulger, "The Effects of Bimodal Particle Size Distribution on Sintering of Powder Injection Molded Compacts", pp.55-62 in Proceedings of the Fifth International Symposium on the Science and Technology of Sintering, Vancouver, Canada, 1991. [77] R. L . Coble, "Effects of Particle Size Distribution in the Initial-Stage Sintering", J. Am. Ceram. Soc, 5 6 [9] 461-466 (1973). [78] G. L. Messing and G. Y . Onoda, Jr., "Sintering of Inhomogeneous Binary Powder Mixtures", J. Am. Ceram. Soc, 6 4 [8] 468-472 (1981). [79] R. K. McGeary, "Mechanical Packing of Spherical Particles", J. Am. Ceram. Soc, 4 4 [10] 513-522 (1961). References 192 [80] A . G. Evans, "Consideration of Inhomogeneity Effects in Sintering", J. Am. Ceram. Soc, 65 [ 10 ] 4 9 7 - 5 0 1 ( 1 9 8 2 ) . [81] R. M . German, "Sintering Theory and Practice", John Wiley and Sons, Inc., 1996, pp.188-191. [82] S. Tamura, T. Ochiai, S. Takanaga, T. Kanai, and Hi.Nakamura, "The Development of the Nano Structural Matrix", pp. 517-520 in Proceedings of UNITECR'03, Osaka, Japan, 2003. [83] S. Tamura, T. Ochiai, S. Takanaga, T. Kanai, and Hi.Nakamura, "The Application of the Nano Structural Matrix to MgO-C bricks", pp. 521-524 in Proceedings of UNITECR '03, Osaka, Japan, 2003. [84] H. Zhao, H . Wang, W. Zhang and H. Gu, "The Influence of Nano-Alumina and Nano-Silica on Sintering and Mechanical Performance of Corundum Refractory", pp. 283-288 in Proceedings of the 4th International Symposium on Refractories, Dalian, China, March 2003. [85] H. Zhao, S. Hu, and H. Wang, "The Influence of Nano-Fe 20 3 on Sintering and Mechanical Performance of Magnesite-Chrome Refractory", pp. 284-287 in Proceedings of UNITECR '03, Osaka, Japan, 2003. [86] S. M . Bushnell-Watson and J. H. Sharp, "On the Cause of the Anomalous Setting Behavior with Respect to Temperature of Calcium Aluminate Cements", Cement and Concrete Research, 20 [5] 677-686 (1990). [87] D. D. Poddar and S. Mulkhopadhyay, "Spinel-Bonded Basic Castables in Relation to Spinel Formation Agents", Inteceram, 51 [4] 282-288 (2002). [88] B. Clavaud, J. P. Kiehl and J. P. Radal, "New Development in Monolithic Refractories", pp. 274-283, in Advances in Ceramics, Vol . 13, edited by R.E. Fisher, 1985. [89] Y . Wei, N . L i and H. Wu, "Influence of Mierosilica on the Formation of Aquo-Compounds in the MgO-Si02-H 2 0 System", Naihuo Cailiao (Refractories), 35 [5] 255-257 (2001). References 193 [90] T. A . Bier, C. Parr, C. Revails, and H. Fryda, "Chemical Interactions in Calcium Aluminate Cement Based Castables Containing Magnesia", pp. 15-21, in Proceedings of Unitecr'97, 1997. [91] B. Myhre and A. M . Hundere, "Substitution of Reactive Alumina with Microsilica in Low Cement and Ultra Low Cement Castables. Part.l: Properties Related to Installation and Demoulding", pp.43-52, in Proceedings of Unitecr'97, 1997. [92] W. Ma and P. W. Brown, "Mechanisms of Reaction of Hydratable Aluminas", J. Am. Ceram. Soc, 82, 453-456 (1999). [93] W. Mista, and J. Wrzyszcz, "Rehydration of Transition Aluminas Obtained by Flash Calcination of Gibbsite", Thermochimica Acta, 331, 67-72 (1999). [94] S. D. Vaidya, and N . V. Thakkar, "Effect of Temperature, pH and Ageing Time on Hydration of Rho-Alumina by Studying Phase Composition and Surface Properties of Transition Alumina Obtained after Thermal Dehydration", Materials Letters, 51, 295-300 (2001). [95] G. Ye and T. Troczynski, "Effect of Magnesia on Strength of Hydratable Alumina-Bonded Castable Refractories", Journal of Materials Science (accepted). [96] J. Parmentier, M . Richard-Plouet, and S. Vilminot, "Influence of the Sol-Gel Synthesis on the Formation of Spinel M g A l 2 0 4 " , Materials Research Bulletin, 33 [11] 1717-1724 (1998). [97] H. E. Kissinger, "Variation of peak temperature with heating rate in differential thermal analysis", J. Res. Nat. Bur. Stds. 57 [4] 217-221 (1956). [98] R. M . German, Particle Packing Characteristics, p. 185, Metal Powder Industries Federation, Princeton, NT (1989). [99] W. Schulle, J. Ulbricht and A. Altun, "Investigations in the Development of Self-Flowing Castables", Ceramic Forum International, 78 [5] E39-E42 (2001). [100] Y . F. Chen, M.-C. Wang, and M . H. Hon, "Secondary Mullite Formation in Kaolin-A1 2 0 3 Ceramics", J. Mater. Res., 19 [3] 806-814 (2004). [101] Y . F. Chen, M.-C. Wang, and M . H. Hon, "Kinetics of Secondary Mullite Formation in K a o l i n - A l 2 0 3 Ceramics", Script Materialia, 51, 231-235 (2004). References 194 [102] M . A . Sainz, F.J. Serrano, J .M. Amigo, J. Bastida, and A. Caballero, " X R D Microstructural Analysis of Mullites Obtained from Kaolinite-Alumina Mixtures", J. Euro. Ceram. Soc, 20 [4] 403-142 (2000). [103] Y . F. Chen, M.-C. Wang, and M . H. Hon, "Transformation Kinetics for Mullite in Kaol in-Al 2 0 3 Ceramics", J. Mater. Res., 18 [6] 1355-1362 (2003). [104] C.Y. Chen, G.S. Lan, and W.H. Tuan, "Preparation of Mullite by the Reaction Sintering of Kaolinite and Alumina", J. Euro. Ceram. Soc, 20, 2519-2525 (2000) [105] R. J. Bratton, "Coprecipitates Yielding MgAl 2 04 Spinel Powders", Amer. Ceram. Soc. Bulletin, 48 [8] 759-762 (1968). [106] D. Domanski, G. Urretavizcaya, F. J. Castro and F. C. Gennari, "Mechanochemical Synthesis of Magnesium Aluminate Spinel Powder at Room Temperature", J. Am. ceram. Soc, 87 [11] 2020-2024 (2004) [107] P. J. Sanchez-Soto, M . C. J. Haro, L. A. Perez-Maqueda, I. Varona, and J. L. Perez-Rodriguez, "Effects of Dry Grinding on the Structural Changes of Kaolinite Powders", J. Am. Ceram. Soc, 83 [7] 1649-57 (2000). [108] M . Sathiyakumar and F. D. Gnanam, "Synthesis of Sol-Gel Derived Alumina Powder: Effect of Milling and Calcination Temperatures on Sintering Behavior", British Ceramic Transactions, 98 [2] 87-92 (1999). [109] C. S. Nordahl, and G. L. Messing, "Thermal Analysis of Phase Transformation Kinetics in Alpha Alumina Seeded Boehmite and Gamma Alumina", Thermochimica Acta, 318 [2] 187-200 (1998). [110] J. Katanic-Popovic, N . Miljevic and S. Zee, " Spinenel Formation from Coprecipitated Gel", Ceramics International, 17, 49-52 (1991). [ I l l ] P. Duran, J. Tartaj, J.F. Fernandez, M . Villegas, C. Moure, "Crystallisation And Sintering Behaviour of Nanocrystalline Y-TZP Powders Obtained by Seeding-Assisted Chemical Coprecipitation", Ceramics International, 25, 125-135 (1999). [112] F.F. Lange, "Powder Processing Science and Technology for Increased Reliability", J. Amer. Ceram. Soc, 72 [1] 3-15 (1989). References 195 [113] W. D. Kingery, Introduction to Ceramics, 2nd Ed.; John Wiley and Sons, New York and London, 1976, p. 469. [114] S. Mukhopadhyay and P. K. Das Poddar, "Effect of Preformed and In-Situ Spinels on Microstructure and Properties of a Low Cement Refractory Castable", Ceramics International, 30, 369-380 (2004). [115] Z. L i , L . Zhang, N . Zhou and G. Ye, "Effects of Ultra-Fine Oxides on the Properties of Ultra-Low Cement Corundum-Based Castables", pp. 149-152 in Proc. of International Colloquium on Refractories, 1992, Aachen, Germany. [116] D. K. Mukherjee and B. N . Samaddar, "Magnesium Aluminate Hydrate and a Chemical Bond of Magnesia Refractories - Part 1: Cold Setting Behavior", Transactions of the Indian Ceramic Society, 47 [5] 141-146 (1988). [117] D. K. Mukherjee and B. N . Samaddar, "Magnesium Aluminate Hydrate and a Chemical Bond of Magnesia Refractories - Part 2: Bond Formation", Transactions of the Indian Ceramic Society, 48 [2] 23-27 (1989). [118] N . Koga, T. Fukagawa and H. Tanaka, "Preparation and Thermal Decomposition of Synthetic Bayerite", Journal of Thermal Analysis and Calorimetry, 64, 965-972 (2001). [119] C. Sunil Kumar, U . S. Hareesh, A. D. Damodaran and K. G. K. Warrier, "Monohydroxy Aluminum Oxide (Boehmite, AlOOH) as a Reactive Binder for Extrusion of Alumina Ceramics", Journal of the European Ceramic Society, 17, 1167-1 172 (1997). [120] L. Pesic, S. Salipurovic, V . Markovic, D. Vucelic, W. Kagunya, and W. Jones, "Thermal Characteristics of a Synthetic Hydrotalcite-Like Material", /. Mater. Chem., 2, 1069-1073 (1992). [121] K. J. D. MacKenzie, R. H. Meinhold, B. L. Sherri and Z. Xu , " 2 7 A1 and 2 5 M g Solid-State Magic-Angle Spinning Nuclear Magnetic Resonance Study of Hydrotalcite and Its Thermal Decomposition Sequence", 7. Mater. Chem., 3, 1263-1269 (1993) [122] Yung-Chao Ko, "Influence of the Characteristics of Spinels on the Slag Resistance of A l 2 0 3 - M g O and Al 2 0 3 -Spinel Castables", J. Am. Ceram. Soc, 83 [9] 2333-35 (2000). Thesis-Related Publications 196 THESIS-RELATED PUBLICATIONS Journal Papers 1. G . Ye and T. Troczynski, "Mechanical Activation of a Heterogeneous Sol-Gel Precursor for Synthesis of M g A l 2 0 4 Spinel", J. Am. Ceram. Soc. (Accepted on April 14, 2005). 2. G . Ye, G. Oprea and T. Troczynski, "Synthesis of M g A l 2 0 4 Spinel Powder by Combination of Sol-Gel and Precipitation Processes", J. Am. Ceram. Soc. (Accepted on March 22, 2005). 3. G . Ye and T. Troczynski, "Effect of Magnesia on Strength Of Hydratable Alumina-Bonded Castable Refractories", Journal of Materials Science (in print, accepted on February 28, 2005). 4. G . Ye and T. Troczynski, "Hydration of Hydratable Alumina in the Presence of Various Forms of MgO", Ceramics International (in print, accepted on January 19, 2005). 5. Y. He, G . Ye, T. Troczynski and G. Oprea, "Hydration Behavior of Magnesia in Binder Systems for Basic Castables", Canadian Metallurgical Quarterly, 43 [2] 173-176 (2004). Conference Papers 1. G . Ye, T. Troczynski, and G. Oprea, "Sintering Studies on Binding Systems for Spinel Bonded Castables", Unified International Technical Conference on Refractories, Orlando, USA, November 8-11, 2005 (accepted). 2. G . Ye, T. Troczynski, G. Oprea, M . Brothers, D. Harris and J. Rigby, "Spinel-Bonded Magnesia versus Spinel-Bonded Spinel Castables", Unified International Technical Conference on Refractories, Orlando, USA, November 8-11, 2005 (accepted). 3. G . Ye, T. Troczynski, G. Oprea, J. Rigby and D. Harris, "Sintering Reactivity of Nano-Powders Used in the Binding Systems for Refractory Castables", pp. 799-808 in Proc. Thesis-Related Publications 197 of 4th International Symposium on Advances in Refractories for Metallurgical Industries, Hamilton, Canada, August 22-25, 2004. 4. S. Murugesan, G . Ye, T. Troczynski, G. Oprea, D. Harris and J. Rigby, "Spinel Bonded Basic Castables", pp. 346-361 in Proc. of 4th International Symposium on Advances in Refractories for Metallurgical Industries, Hamilton, Canada, August 22-25, 2004. 5. G . Ye, T. Troczynski and G. Oprea, "Development of Reactive MgAl204 Spinel Powders", pp. 168-171 in Proc. of Unified International Technical Conference on Refractories, Osaka, Japan, 2003. 6. G . Ye, T. Troczynski and G. Oprea, "Synthesis of M g A l 2 0 4 Spinel by Modified Coprecipitation and Sol-Gel Processing", pp. 78-82 in Proc. of International Symposium on Refractories, Dalian, China, 2003. Appendix 198 APPENDIX Appendix A. Sample Descriptions AO: Bi-modal mixture without ultrafine powder (see Table 4.2-2) A l - 1 , A l - 2 , A l - 3 , A l - 4 , A l - 5 : Bi-modal mixtures with ultrafine spinel (0.3 pm) A2-1, A2-2, A2-3, A2-4; A2-5: Bi-modal mixtures with ultrafine spinel (0.2 pm) A3-1, A3-2, A3-3, A3-4, A3-5: Bi-modal mixtures with M A - 9 fired at 900°C A4-1, A4-2, A4-3, A4-4, A4-5: Bi-modal mixtures with MA-9 fired at 900°C and followed by milling D l -1 : Tri-modal mixture without ultrafine powder (see Table 4.2-3) Dl -2 , Dl -3 : Tri-modal mixture with 5% ultrafine spinel (0.2 pm) D2-1, D2-2, D2-3: Tri-modal mixtures with 5% ultrafine spinel (0.2 pm) and ultrafine alumina (0.5 pm) D4-1, D4-2, D4-3: Tri-modal mixture with 5% ultrafine spinel (0.3 pm) and ultrafine alumina (0.5 pm) D5-1: Tri-modal mixture with 5% MA-9 fired at 900°C and followed by milling M A - 1 : Precursor prepared using coprecipitation • MA-2 : Precursor prepared using coprecipitation with addition of carbon black MA-3 : Precursor prepared using coprecipitation with addition of polyvinyl alcohol MA-5 : Precursor prepared by heterogeneous sol-gel process using magnesia (d5o=3.09 pm) MA-5A: MA-5 with 3 h of further refluxing after the addition of light MgO MA-6: Precursor prepared using heterogeneous sol-gel process using magnesia (<45 pm) MA-8: Precursor prepared by heterogeneous sol-gel process using magnesia (d5o = 1.82 pm) MA-9: Precursor prepared by sol-gel precipitation process Appendix 199 MA-21: Mixture of Mg(OH) 2 and AlOOH MA-22: Mixture of Mg(OH) 2 and Al(OH) 3 Mixture I: 16% hydratable alumina (Alphabond 300) and 84% reactive magnesia Mixture II: 16% hydratable alumina (Alphabond 300) and 84% fused magnesia Mixture III: 16% hydratable alumina (Alphabond 300) and 84% deadburnt magnesia M l - 0 , M l - 3 , M l - 6 , M2-6, M3-0, M3-3, M3-5: Castables with fused magnesia aggregate (Table 4.3-2 and Table 4.3-4) Sl-0, Sl-3, Sl-6, S2-6, S3-0, S3-5, S4-0, S4-2, S4-5: Castables with fused spinel aggregate (Table 4.3-3, Table 4.3-4, Table 4.3-5) Appendix 200 Appendix B. Physical Properties of the Bi-modal Mixes Mix code Green Density*® (g/cm3) Heat treatment Linear change© (%) Bulk density® (g/cm3) Apparent porosity® (%) Compressive strength® (MPa) AO 2.35 816°Cx5h -0.18 2.21 35.47 2.4 1280°Cx5h -0.70 2.23 35.10 19.6 1450°Cx5h -2.44 2.34 31.45 55.6 Al -1 2.36 816°Cx5h -0.19 2.21 35.40 2.3 1280°Cx5h -0.70 2.23 34.96 19.5 1450°Cx5h -2.47 2.34 31.36 55.8 A l - 2 2.37 816°Cx5h -0.21 2.21 35.10 2.5 1280°Cx5h -0.70 2.24 34.67 19.6 1450°Cx5h -2.48 2.35 31.28 55.5 A l - 3 2.38 816°Cx5h -0.23 2.22 34.82 2.4 1280°Cx5h -0.71 2.25 34.15 19.7 1450°Cx5h -2.50 2.36 31.10 55.7 A l - 4 2.39 816°Cx5h -0.24 2.23 34.13 2.5 1280°Cx5h -0.71 2.26 33.42 20.3 1450°Cx5h -2.55 2.37 30.97 56.3 A l - 5 2.41 816°Cx5h -0.50 2.28 33.68 3.7 1280°Cx5h -0.72 2.30 32.96 24.1 1450°Cx5h -2.64 2.38 30.09 57.2 A2-1 2.36 816°Cx5h -0.18 2.21 35.31 2.4 1280°Cx5h -0.70 2.23 34.92 20.2 1450°Cx5h -2.45 2.34 31.31 55.9 A2-2 2.37 816°Cx5h -0.19 2.22 35.06 2.5 1280°Cx5h -0.71 2.24 34.63 21.2 1450°Cx5h -2.48 2.35 30.96 56.1 A2-3 2.38 816°Cx5h -0.19 2.23 34.93 2.6 1280°Cx5h -0.71 2.25 34.51 21.7 1450°Cx5h -2.51 2.36 30.84 56.0 Appendix Appendix B. Physical properties of the bi-modal mixes (continued) 201 Mix code Green Density*® (g/cm3) Heat treatment Linear change® (%) Bulk density® (g/cm3) Apparent porosity® (%) Compressive strength® (MPa) A2-4 2.40 816°Cx5h -0.20 2.24 34.42 2.7 1280°Cx5h -0.72 2.26 33.96 22.2 1450°Cx5h -2.54 2.38 30.11 56.8 A2-5 2.43 816°Cx5h -0.21 2.26 33.69 2.8 1280°Cx5h -0.73 2.28 33.15 23.5 1450°Cx5h -2.63 2.41 29.02 57.7 A3-1 2.34 1280°Cx5h -0.71 2.22 35.53 17.4 1450°Cx5h -2.44 2.33 31.93 53.2 A3-2 2.33 1280°Cx5h -0.71 2.22 35.71 16.8 1450°Cx5h -2.45 2.33 32.29 51.5 A3-3 2.35 1280°Cx5h -0.72 2.20 36.18 14.6 1450°Cx5h -2.46 2.31 32.82 47.6 A3-4 2.30 1280°Cx5h -0.72 2.19 36.77 13.3 1450°Cx5h -2.46 2.29 33.36 44.2 A3-5 2.26 1280°Cx5h -0.73 2.16 37.79 11.1 1450°Cx5h -2.47 2.25 34.27 39.3 A4-1 2.36 1280°Cx5h -0.70 2.24 35.03 20.1 1450°Cx5h -2.47 2.35 31.39 56.0 A4-2 2.37 1280°Cx5h -0.71 2.25 34.68 21.6 1450°Cx5h -2.49 2.36 30.90 56.3 A4-3 2.38 1280°Cx5h -0.71 2.26 34.37 22.9 1450°Cx5h -2.52 2.37 30.66 56.5 A4-4 2.40 1280°Cx5h -0.72 2.28 33.61 23.7 1450°Cx5h -2.56 2.39 30.02 57.1 A4-5 2.42 1280°Cx5h -0.74 2.30 32.76 26.2 1450°Cx5h -2.65 2.41 29.17 58.3 *Calcu ated from weight, diameter and height of the samples. Appendix 202 ® Standard deviation: ± (0.01-0.02). © Standard deviation: 816°C: ±(0.01-0.03) 1280°C: ±(0.05-0.07) 1450°C:± (0.08-0.12) @ Standard deviation: ± (0.01-0.02). ® Standard deviation: ± (0.26-0.82). © Standard deviation: 816°C:± (0.03-0.06) 1280°C:± (0.28-0.71) 1450°C:± (0.99-1.43) Note: Standard deviation for linear change and compressive strength falls in three different ranges after firing at three different temperatures. Appendix Appendix C Physical properties of the tri-modal mixes Mix code Green Density*® (g/cm3) Heat treatment Linear change© (%) Bulk density® (g/cm3) Apparent porosity® (%) Compressive strength® (MPa) Dl-1 2.60 816°Cx5h -0.03 2.52 31.25 2.87 1280°Cx5h -0.12 2.54 30.78 20.76 1450°Cx5h -0.43 2.62 29.69 57.87 Dl-2 2.68 816°Cx5h -0.08 2.64 29.15 7.29 1280°Cx5h -0.15 2.66 28.13 28.61 1450°Cx5h -0.50 2.70 27.64 65.36 Dl-3 2.69 816°Cx5h -0.16 2.66 28.79 11.80 1280°Cx5h -0.28 2.67 28.01 30.67 1450°Cx5h -0.64 2.71 27.57 69.02 D2-1 2.67 816°Cx5h -0.08 2.64 29.15 7.29 1280°Cx5h -0.15 2.66 28.13 28.61 1450°Cx5h -0.50 2.70 27.64 65.36 D2-2 2.69 816°Cx5h -0.09 2.67 28.22 8.99 1280°Cx5h -0.18 2.69 27.69 30.69 1450°Cx5h -0.56 2.71 27.53 67.58 D2-3 2.70 816°Cx5h -0.12 2.68 27.91 10.68 1280°Cx5h -0.27 2.71 27.63 32.46 1450°Cx5h -0.63 2.72 27.46 68.39 D4-1 2.63 816°Cx5h -0.03 2.60 31.36 6.61 1280°Cx5h -0.13 2.60 30.94 22.87 1450°Cx5h -0.46 2.61 29.25 60.36 D4-2 2.65 816°Cx5h -0.05 2.62 30.49 7.85 1280°Cx5h -0.16 2.63 30.32 29.33 1450°Cx5h -0.53 2.64 28.72 66.73 D4-3 2.70 816°Cx5h -0.09 2.68 28.06 8.43 1280°Cx5h -0.39 2.69 27.98 30.85 1450°Cx5h -0.77 2.70 27.02 67.16 Appendix 204 *Calculated from weight, diameter and height of the samples. ® Standard deviation: ±(0.01 -0.02). © Standard deviation: 816°C: ± (0.01-0.02) 1280°C:± (0.02-0.05) 1450°C: ± (0.03-0.07) © Standard deviation: ± (0.01-0.02). © Standard deviation: ± (0.26-0.63). © Standard deviation: 816°C:± (0.23-0.37) 1280°C:± (0.46-0.67) 1450°C:± (0.95-1.27) Note: Standard deviation for linear change and compressive strength falls in three different ranges after firing at three different temperatures. Appendix 205 Appendix D. Glossary Related to Refractories Agglomeration: A process by which particles of smaller materials are brought together to form larger particles by interparticle bonding. Alumina gel: Gel developed from boehmite sol through drying or adjusting pH value. Binding systems: Referring to the particulate fractions with a size below 90 pm in castable refractories. Calcination: A heating process by which hydrates, carbonate, or other compounds are decomposed and volatile material is released. Castables: Refractory materials usually supplied as dry mixtures of granular ceramics; need to be mixed with water or other liquid at the applications site. Castables are installed by pouring, usually followed by vibration, to fill complex shapes. Curing: Exposing cement or hydratable alumina to moisture to ensure hydration of the cement or hydratable alumina and proper hardening of the castable. Deadburnt magnesia: Magnesia from magnesia (MgCC>3) after firing at high temperatures (above 1700°C). Dehydration: Decomposition reaction with a release of water. Explosion spalling: Fracture of refractory lining due to stresses caused by temperature gradients or pressure of water vapor inside the linings. Flowability: The ability of castable to flow under vibration or without vibration. Detailed description of flowability is provided in Section 4.3.4. Silica fume (Microsilica): Amorphous silicon oxide (SiC>2) consisting of spherical particles with an average diameter of ~0.1 pm. Hot strength: Strength measured at high temperatures (e.g. above 800°C). Hydratable alumina: P-AI2O3 which hydrates to form bayerite Al(OH) 3 and boehmite AlOOH when exposed to water. Appendix 206 Hydraulic bonding: The bonding generated by the hydration of cements or hydratable alumina. Hydrotalcite: Aluminium magnesium carbonate hydroxide hydrate (Mg4Al 2)(OH)i 2(C0 3)(H 20)6, (Mg 6 Al 2 )(OH)i6(C0 3 )(H 2 0)4. It is also referred to as hydrotalcite hydrate and hydrotalcite-like compound. Modulus of Rupture (MOR): Three-point bending strength. Reactive alumina: Fine alpha alumina (<3 pm) produced by decomposition of aluminum hydroxide at temperatures below 1300°C. Monolithics: Refractory materials supplied as dry or wet mixtures of particulate materials and shaped in installation of industrial vessels, and fired during service. Monolithics are also referred to as unshaped refractories. Transition alumina: Alumina has several polymorphs. The transition alumina referred to in the thesis is p - A l 2 0 3 . Ultrafine powder: Normally referring to powders with a particle size below 3-5 pm in refractories. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078777/manifest

Comment

Related Items